{
  "best_metric": 4.337387561798096,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/lstm/2/checkpoints/checkpoint-228960",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 228960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8206,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5517,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0492,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9882,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.943,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8668,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7139,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994131474623009e-05,
      "loss": 6.615,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.5194,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992455923001279e-05,
      "loss": 6.4393,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991617328250227e-05,
      "loss": 6.3772,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990778733499175e-05,
      "loss": 6.3056,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989940138748123e-05,
      "loss": 6.2402,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989101543997071e-05,
      "loss": 6.1777,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988262949246018e-05,
      "loss": 6.1283,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987424354494966e-05,
      "loss": 6.0656,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986585759743914e-05,
      "loss": 6.0332,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985747164992862e-05,
      "loss": 5.9864,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98490857024181e-05,
      "loss": 5.9467,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984069975490758e-05,
      "loss": 5.9056,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983231380739706e-05,
      "loss": 5.8737,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 5.8322,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815558291179756e-05,
      "loss": 5.8041,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807172343669236e-05,
      "loss": 5.7614,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798786396158716e-05,
      "loss": 5.7496,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 5.7106,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 5.6924,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.6523,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6354,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6148,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.5896,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.58,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5538,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5377,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5294,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.506,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.4977,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4519,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4578,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9673013562304644e-05,
      "loss": 5.4188,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9664627614794124e-05,
      "loss": 5.4256,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4043,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4025,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.3707,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963110020355577e-05,
      "loss": 5.3789,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962271425604525e-05,
      "loss": 5.3669,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3595,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3521,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959757279231743e-05,
      "loss": 5.3231,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958918684480691e-05,
      "loss": 5.3145,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958080089729639e-05,
      "loss": 5.3136,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957241494978587e-05,
      "loss": 5.3003,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956402900227535e-05,
      "loss": 5.2938,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955564305476483e-05,
      "loss": 5.271,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.2684,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2476,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2694,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2219,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2387,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.228,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949699055860238e-05,
      "loss": 5.2077,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948860461109186e-05,
      "loss": 5.2068,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.1939,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.1821,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.1751,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.189,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1628,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1535,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942993573612568e-05,
      "loss": 5.144,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942154978861516e-05,
      "loss": 5.1415,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.1472,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.1426,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93963919460836e-05,
      "loss": 5.1174,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1341,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1269,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.0984,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.0971,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.0869,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.0785,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.0734,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.0779,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0732,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 5.0751,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 5.0513,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 5.0322,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.928744014366177e-05,
      "loss": 5.0385,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927905419615125e-05,
      "loss": 5.0402,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927066824864073e-05,
      "loss": 5.039,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92622823011302e-05,
      "loss": 5.0211,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0188,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0187,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0121,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922875488989185e-05,
      "loss": 5.0,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 4.9985,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 4.9846,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 4.9976,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919524385745724e-05,
      "loss": 4.9831,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918685790994672e-05,
      "loss": 4.9624,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91784719624362e-05,
      "loss": 4.9672,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 4.9667,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.9722,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.9544,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.9512,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136542224883596e-05,
      "loss": 4.937,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.9429,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.9218,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111400761155765e-05,
      "loss": 4.9317,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9103014813645245e-05,
      "loss": 4.9198,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909462886613473e-05,
      "loss": 4.9228,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908624291862421e-05,
      "loss": 4.911,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9157,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9017,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906110145489638e-05,
      "loss": 4.9064,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905271550738586e-05,
      "loss": 4.8975,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904432955987534e-05,
      "loss": 4.8909,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903594361236482e-05,
      "loss": 4.9032,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.8873,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.8853,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.874,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.8793,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.8742,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8674,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8735,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8595,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896048646357387e-05,
      "loss": 4.8598,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895210051606335e-05,
      "loss": 4.8395,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.894371456855283e-05,
      "loss": 4.8452,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8457,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.838,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8288,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.8508,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.8343,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889341526229344e-05,
      "loss": 4.8247,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888502931478292e-05,
      "loss": 4.8251,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8876659746076134e-05,
      "loss": 4.8201,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8868273798565614e-05,
      "loss": 4.822,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859887851055094e-05,
      "loss": 4.8172,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851501903544574e-05,
      "loss": 4.8133,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843115956034054e-05,
      "loss": 4.8159,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834730008523534e-05,
      "loss": 4.8135,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8826344061013014e-05,
      "loss": 4.8021,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8817958113502494e-05,
      "loss": 4.7927,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.8037,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.7904,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879286578618586e-05,
      "loss": 4.793,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878447983867534e-05,
      "loss": 4.7851,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877609389116482e-05,
      "loss": 4.7825,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.787,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.7761,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7721,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.7402849197387695,
      "eval_runtime": 293.2378,
      "eval_samples_per_second": 1301.302,
      "eval_steps_per_second": 40.667,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7586,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.759,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.7823,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7667,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709006311080657e-05,
      "loss": 4.7598,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700620363570137e-05,
      "loss": 4.756,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.7551,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683848468549096e-05,
      "loss": 4.7379,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675462521038576e-05,
      "loss": 4.7479,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667076573528056e-05,
      "loss": 4.7477,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865870700482127e-05,
      "loss": 4.7495,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865032105731075e-05,
      "loss": 4.7478,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864195148860396e-05,
      "loss": 4.7303,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863356554109344e-05,
      "loss": 4.7299,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862517959358292e-05,
      "loss": 4.7187,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86167936460724e-05,
      "loss": 4.7209,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608407698561874e-05,
      "loss": 4.7267,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600021751051354e-05,
      "loss": 4.7171,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591635803540834e-05,
      "loss": 4.7229,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858326623483405e-05,
      "loss": 4.7366,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857488028732352e-05,
      "loss": 4.7178,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856649433981301e-05,
      "loss": 4.7206,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855810839230249e-05,
      "loss": 4.704,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85497388235957e-05,
      "loss": 4.724,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541369254888915e-05,
      "loss": 4.7004,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532983307378395e-05,
      "loss": 4.7044,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524597359867875e-05,
      "loss": 4.6906,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851621141235735e-05,
      "loss": 4.6998,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850782546484683e-05,
      "loss": 4.6908,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8499455896140044e-05,
      "loss": 4.6874,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8491069948629524e-05,
      "loss": 4.6928,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8482684001119e-05,
      "loss": 4.6957,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847429805360848e-05,
      "loss": 4.6893,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.6908,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845754253739117e-05,
      "loss": 4.68,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844915658988065e-05,
      "loss": 4.6897,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844077064237013e-05,
      "loss": 4.6626,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843238469485961e-05,
      "loss": 4.6769,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842399874734909e-05,
      "loss": 4.6532,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841561279983857e-05,
      "loss": 4.6755,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840724323113178e-05,
      "loss": 4.6572,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839885728362126e-05,
      "loss": 4.6693,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839047133611074e-05,
      "loss": 4.6497,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838210176740395e-05,
      "loss": 4.6681,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837371581989343e-05,
      "loss": 4.6645,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836532987238292e-05,
      "loss": 4.6615,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83569439248724e-05,
      "loss": 4.6675,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834855797736188e-05,
      "loss": 4.6374,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.6415,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833178608234084e-05,
      "loss": 4.655,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832340013483032e-05,
      "loss": 4.6475,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83150141873198e-05,
      "loss": 4.645,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830662823980928e-05,
      "loss": 4.6372,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298258671102486e-05,
      "loss": 4.6328,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289872723591966e-05,
      "loss": 4.6271,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281486776081446e-05,
      "loss": 4.6424,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273100828570926e-05,
      "loss": 4.6192,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264714881060406e-05,
      "loss": 4.6311,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256328933549886e-05,
      "loss": 4.635,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247942986039366e-05,
      "loss": 4.6181,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239557038528846e-05,
      "loss": 4.6194,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823120384862579e-05,
      "loss": 4.6182,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822281790111527e-05,
      "loss": 4.6053,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821443195360475e-05,
      "loss": 4.6096,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820604600609423e-05,
      "loss": 4.6276,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819766005858371e-05,
      "loss": 4.609,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189274111073184e-05,
      "loss": 4.6016,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180888163562664e-05,
      "loss": 4.6028,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817251859485588e-05,
      "loss": 4.6033,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816413264734535e-05,
      "loss": 4.6169,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815574669983483e-05,
      "loss": 4.6107,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814736075232432e-05,
      "loss": 4.5989,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81389748048138e-05,
      "loss": 4.6144,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813060523610701e-05,
      "loss": 4.6132,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812221928859649e-05,
      "loss": 4.5955,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811383334108597e-05,
      "loss": 4.5952,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810544739357545e-05,
      "loss": 4.5939,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809706144606493e-05,
      "loss": 4.5882,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808869187735814e-05,
      "loss": 4.5865,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808030592984762e-05,
      "loss": 4.5895,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80719199823371e-05,
      "loss": 4.596,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806353403482658e-05,
      "loss": 4.5971,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805514808731606e-05,
      "loss": 4.5855,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804676213980554e-05,
      "loss": 4.5624,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803837619229502e-05,
      "loss": 4.5784,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8029990244784504e-05,
      "loss": 4.583,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802162067607771e-05,
      "loss": 4.5864,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801323472856719e-05,
      "loss": 4.5765,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80048651598604e-05,
      "loss": 4.5728,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799647921234988e-05,
      "loss": 4.5787,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798809326483936e-05,
      "loss": 4.5711,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797970731732884e-05,
      "loss": 4.5665,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797132136981832e-05,
      "loss": 4.5731,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79629354223078e-05,
      "loss": 4.5579,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795454947479728e-05,
      "loss": 4.576,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794617990609049e-05,
      "loss": 4.5685,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793779395857997e-05,
      "loss": 4.5468,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792940801106946e-05,
      "loss": 4.5608,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792102206355894e-05,
      "loss": 4.5584,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791263611604842e-05,
      "loss": 4.5696,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790426654734163e-05,
      "loss": 4.5567,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895880599831107e-05,
      "loss": 4.5517,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887494652320587e-05,
      "loss": 4.5448,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879108704810067e-05,
      "loss": 4.5547,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870722757299546e-05,
      "loss": 4.5367,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862353188592756e-05,
      "loss": 4.5446,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853967241082236e-05,
      "loss": 4.5442,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845581293571715e-05,
      "loss": 4.5443,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837195346061195e-05,
      "loss": 4.5342,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828809398550675e-05,
      "loss": 4.5442,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820423451040155e-05,
      "loss": 4.5361,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812037503529635e-05,
      "loss": 4.5378,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803651556019115e-05,
      "loss": 4.5388,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795281987312324e-05,
      "loss": 4.5295,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786896039801804e-05,
      "loss": 4.5529,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778510092291284e-05,
      "loss": 4.5383,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777014052358449e-05,
      "loss": 4.5356,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776175457607397e-05,
      "loss": 4.5252,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775336862856345e-05,
      "loss": 4.5332,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774498268105293e-05,
      "loss": 4.5307,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773659673354241e-05,
      "loss": 4.5222,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772821078603189e-05,
      "loss": 4.5387,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771984121732511e-05,
      "loss": 4.5258,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771145526981459e-05,
      "loss": 4.5203,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770306932230407e-05,
      "loss": 4.5165,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769468337479355e-05,
      "loss": 4.5162,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768629742728303e-05,
      "loss": 4.5163,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767791147977251e-05,
      "loss": 4.5189,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766952553226199e-05,
      "loss": 4.5055,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766113958475147e-05,
      "loss": 4.5329,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765277001604468e-05,
      "loss": 4.5228,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764440044733789e-05,
      "loss": 4.5119,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763601449982737e-05,
      "loss": 4.511,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762762855231685e-05,
      "loss": 4.5062,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761924260480633e-05,
      "loss": 4.5181,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610856657295813e-05,
      "loss": 4.5148,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602470709785293e-05,
      "loss": 4.5101,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75941011410785e-05,
      "loss": 4.5149,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758571519356798e-05,
      "loss": 4.5104,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757734562486119e-05,
      "loss": 4.5089,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756895967735067e-05,
      "loss": 4.4995,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756057372984015e-05,
      "loss": 4.5121,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755218778232963e-05,
      "loss": 4.501,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754380183481911e-05,
      "loss": 4.504,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753541588730859e-05,
      "loss": 4.499,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752702993979807e-05,
      "loss": 4.5013,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751864399228755e-05,
      "loss": 4.5006,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751025804477703e-05,
      "loss": 4.4889,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750188847607025e-05,
      "loss": 4.4938,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.468594074249268,
      "eval_runtime": 293.1863,
      "eval_samples_per_second": 1301.531,
      "eval_steps_per_second": 40.674,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749350252855973e-05,
      "loss": 4.4813,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748511658104921e-05,
      "loss": 4.4803,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747673063353868e-05,
      "loss": 4.5069,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746834468602816e-05,
      "loss": 4.4979,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745995873851764e-05,
      "loss": 4.4866,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745157279100712e-05,
      "loss": 4.486,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.74431868434966e-05,
      "loss": 4.488,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743480089598608e-05,
      "loss": 4.4733,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742641494847556e-05,
      "loss": 4.4907,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741802900096504e-05,
      "loss": 4.4829,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740964305345452e-05,
      "loss": 4.4869,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401257105944e-05,
      "loss": 4.4882,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392887537237216e-05,
      "loss": 4.4709,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384501589726696e-05,
      "loss": 4.4747,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376115642216176e-05,
      "loss": 4.4685,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7367729694705656e-05,
      "loss": 4.4692,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7359343747195135e-05,
      "loss": 4.473,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7350957799684615e-05,
      "loss": 4.4655,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7342571852174095e-05,
      "loss": 4.4732,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733418590466357e-05,
      "loss": 4.4934,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732579995715305e-05,
      "loss": 4.4702,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7317430388446264e-05,
      "loss": 4.4742,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730904444093574e-05,
      "loss": 4.468,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730065849342522e-05,
      "loss": 4.4812,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292272545914704e-05,
      "loss": 4.4606,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7283886598404184e-05,
      "loss": 4.4703,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7275500650893664e-05,
      "loss": 4.4543,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7267114703383144e-05,
      "loss": 4.4588,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258728755872624e-05,
      "loss": 4.4567,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725035918716583e-05,
      "loss": 4.4534,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724197323965531e-05,
      "loss": 4.4594,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723358729214479e-05,
      "loss": 4.4657,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722520134463427e-05,
      "loss": 4.4593,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721683177592748e-05,
      "loss": 4.4649,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720846220722069e-05,
      "loss": 4.4488,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720007625971017e-05,
      "loss": 4.4673,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719169031219965e-05,
      "loss": 4.4346,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718330436468914e-05,
      "loss": 4.4572,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717491841717862e-05,
      "loss": 4.4322,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71665324696681e-05,
      "loss": 4.4539,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715814652215758e-05,
      "loss": 4.4405,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714976057464706e-05,
      "loss": 4.4541,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714139100594027e-05,
      "loss": 4.4352,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7133021437233476e-05,
      "loss": 4.4473,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7124635489722956e-05,
      "loss": 4.4495,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7116249542212436e-05,
      "loss": 4.4501,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7107863594701916e-05,
      "loss": 4.4512,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7099477647191396e-05,
      "loss": 4.4295,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7091091699680876e-05,
      "loss": 4.4266,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082705752170356e-05,
      "loss": 4.4488,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074319804659836e-05,
      "loss": 4.4405,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706595023595305e-05,
      "loss": 4.4344,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705756428844253e-05,
      "loss": 4.4331,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704917834093201e-05,
      "loss": 4.429,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704079239342149e-05,
      "loss": 4.4185,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70324228247147e-05,
      "loss": 4.4373,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702405325600791e-05,
      "loss": 4.4161,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701566730849739e-05,
      "loss": 4.4264,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700728136098687e-05,
      "loss": 4.4377,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699889541347635e-05,
      "loss": 4.4191,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699050946596583e-05,
      "loss": 4.4202,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698212351845531e-05,
      "loss": 4.4228,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.697373757094479e-05,
      "loss": 4.4064,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965351623434276e-05,
      "loss": 4.41,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956982054727485e-05,
      "loss": 4.4309,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948596107216965e-05,
      "loss": 4.4157,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940210159706445e-05,
      "loss": 4.4091,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931824212195925e-05,
      "loss": 4.4105,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69234382646854e-05,
      "loss": 4.4147,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691505231717488e-05,
      "loss": 4.4253,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690666636966436e-05,
      "loss": 4.4222,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689828042215384e-05,
      "loss": 4.4104,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688989447464332e-05,
      "loss": 4.4255,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68815085271328e-05,
      "loss": 4.4213,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687312257962228e-05,
      "loss": 4.4096,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686473663211176e-05,
      "loss": 4.4071,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856367063404974e-05,
      "loss": 4.4138,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6847981115894454e-05,
      "loss": 4.406,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6839595168383934e-05,
      "loss": 4.4043,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6831209220873414e-05,
      "loss": 4.406,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682283965216662e-05,
      "loss": 4.4132,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68144537046561e-05,
      "loss": 4.4174,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680606775714558e-05,
      "loss": 4.4032,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679768180963506e-05,
      "loss": 4.389,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678929586212454e-05,
      "loss": 4.3971,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678090991461402e-05,
      "loss": 4.4031,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677254034590723e-05,
      "loss": 4.4097,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676415439839671e-05,
      "loss": 4.3995,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675576845088619e-05,
      "loss": 4.3979,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674738250337568e-05,
      "loss": 4.4082,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673899655586516e-05,
      "loss": 4.3933,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673062698715837e-05,
      "loss": 4.3946,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672224103964785e-05,
      "loss": 4.3985,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671385509213733e-05,
      "loss": 4.3899,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670546914462681e-05,
      "loss": 4.4069,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669708319711629e-05,
      "loss": 4.3936,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668869724960576e-05,
      "loss": 4.3809,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668031130209524e-05,
      "loss": 4.3924,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6671941733388456e-05,
      "loss": 4.3895,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6663555785877936e-05,
      "loss": 4.4059,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655186217171145e-05,
      "loss": 4.3932,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664680026966063e-05,
      "loss": 4.3851,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663841432215011e-05,
      "loss": 4.3777,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663002837463959e-05,
      "loss": 4.3908,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6621642427129065e-05,
      "loss": 4.3738,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6613256479618545e-05,
      "loss": 4.3838,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6604870532108025e-05,
      "loss": 4.3826,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6596484584597505e-05,
      "loss": 4.3815,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6588098637086985e-05,
      "loss": 4.3703,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6579712689576465e-05,
      "loss": 4.3896,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6571326742065945e-05,
      "loss": 4.3745,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6562940794555425e-05,
      "loss": 4.3746,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6554554847044905e-05,
      "loss": 4.3826,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6546185278338114e-05,
      "loss": 4.3708,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653781570963133e-05,
      "loss": 4.3925,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652942976212081e-05,
      "loss": 4.3814,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652104381461029e-05,
      "loss": 4.3816,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.651265786709977e-05,
      "loss": 4.3709,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650427191958925e-05,
      "loss": 4.3764,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649590235088246e-05,
      "loss": 4.3765,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648751640337194e-05,
      "loss": 4.3663,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647913045586142e-05,
      "loss": 4.3814,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64707445083509e-05,
      "loss": 4.3756,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646235856084038e-05,
      "loss": 4.3677,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645397261332986e-05,
      "loss": 4.3699,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644558666581934e-05,
      "loss": 4.3683,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643720071830882e-05,
      "loss": 4.3599,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6428831149602034e-05,
      "loss": 4.3733,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642046158089524e-05,
      "loss": 4.3581,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641207563338472e-05,
      "loss": 4.3795,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64036896858742e-05,
      "loss": 4.3767,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639530373836368e-05,
      "loss": 4.3633,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638693416965689e-05,
      "loss": 4.3616,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637854822214637e-05,
      "loss": 4.362,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637016227463585e-05,
      "loss": 4.3713,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636177632712533e-05,
      "loss": 4.3675,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635339037961481e-05,
      "loss": 4.3657,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634502081090802e-05,
      "loss": 4.3682,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633663486339751e-05,
      "loss": 4.3681,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632824891588699e-05,
      "loss": 4.3675,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631986296837647e-05,
      "loss": 4.3531,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631147702086595e-05,
      "loss": 4.3731,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630310745215916e-05,
      "loss": 4.3616,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6294737883452366e-05,
      "loss": 4.3583,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6286351935941846e-05,
      "loss": 4.3602,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277965988431326e-05,
      "loss": 4.3614,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269580040920806e-05,
      "loss": 4.3575,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261194093410286e-05,
      "loss": 4.3473,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6252808145899766e-05,
      "loss": 4.3497,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.337387561798096,
      "eval_runtime": 294.5321,
      "eval_samples_per_second": 1295.584,
      "eval_steps_per_second": 40.488,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.460672397103816e+16,
  "trial_name": null,
  "trial_params": null
}