{
  "best_metric": 4.0336785316467285,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl2/transformer/0/checkpoints/checkpoint-152640",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 152640,
  "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": 10.9798,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8315,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.2085,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.0029,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.8364,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.7323,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.6185,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5452,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4657,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.4052,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3625,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.3164,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 5.2673,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 5.2177,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988259673485272e-05,
      "loss": 5.1919,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98742107873422e-05,
      "loss": 5.1458,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1187,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.0946,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.062,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0318,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.0174,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 4.9839,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 4.9684,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.9416,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9354,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 4.9031,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.8876,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8579,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.851,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 4.8285,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 4.8212,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 4.811,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 4.7942,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 4.7832,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 4.7702,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7545,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969820416244367e-05,
      "loss": 4.7525,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.7217,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.7116,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.6904,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.693,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.6795,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.6804,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6593,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 4.6529,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6492,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6431,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6303,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 4.6123,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 4.604,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958085003370758e-05,
      "loss": 4.6218,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957246408619706e-05,
      "loss": 4.6102,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956407813868654e-05,
      "loss": 4.5907,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.5766,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 4.5847,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 4.5632,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 4.5721,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 4.5424,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.5625,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5485,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5277,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5207,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.5342,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.4944,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.5079,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 4.5149,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 4.4988,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.4924,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4754,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421582546222625e-05,
      "loss": 4.4751,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.4699,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 4.4867,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 4.461,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 4.4739,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 4.4674,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 4.4658,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 4.4485,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 4.4464,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346125397431674e-05,
      "loss": 4.443,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 4.4395,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 4.4323,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 4.4245,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 4.4269,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930421203868281e-05,
      "loss": 4.4248,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 4.405,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.928744014366177e-05,
      "loss": 4.4122,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927905419615125e-05,
      "loss": 4.4112,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927066824864073e-05,
      "loss": 4.4133,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92622823011302e-05,
      "loss": 4.3986,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925391273242342e-05,
      "loss": 4.3952,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92455267849129e-05,
      "loss": 4.3888,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923714083740238e-05,
      "loss": 4.3822,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922875488989185e-05,
      "loss": 4.3827,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 4.3874,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 4.3792,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920359704736029e-05,
      "loss": 4.3861,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919521109984978e-05,
      "loss": 4.3766,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 4.3595,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 4.3597,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917006963612195e-05,
      "loss": 4.3693,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.3653,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.3627,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.3532,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136542224883596e-05,
      "loss": 4.3426,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.3571,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.3341,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91114171399595e-05,
      "loss": 4.3392,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910303119244898e-05,
      "loss": 4.3341,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909464524493846e-05,
      "loss": 4.3342,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908625929742794e-05,
      "loss": 4.324,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907787334991742e-05,
      "loss": 4.3249,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90694874024069e-05,
      "loss": 4.3196,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906110145489638e-05,
      "loss": 4.325,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905273188618959e-05,
      "loss": 4.3209,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904434593867907e-05,
      "loss": 4.3114,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903595999116855e-05,
      "loss": 4.3225,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902757404365803e-05,
      "loss": 4.328,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901920447495124e-05,
      "loss": 4.316,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901081852744072e-05,
      "loss": 4.3028,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90024325799302e-05,
      "loss": 4.3008,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994046632419686e-05,
      "loss": 4.2988,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985660684909166e-05,
      "loss": 4.3051,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977274737398646e-05,
      "loss": 4.3013,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968888789888125e-05,
      "loss": 4.289,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960519221181335e-05,
      "loss": 4.299,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952133273670815e-05,
      "loss": 4.2885,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.2886,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.2837,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926975431139254e-05,
      "loss": 4.2848,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918589483628734e-05,
      "loss": 4.2769,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910203536118214e-05,
      "loss": 4.286,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890181758860769e-05,
      "loss": 4.2858,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.2845,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.2602,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.2759,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.2696,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.2747,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885153466115204e-05,
      "loss": 4.2633,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884314871364151e-05,
      "loss": 4.2722,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883476276613099e-05,
      "loss": 4.2612,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.2783,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881800724991369e-05,
      "loss": 4.2473,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880962130240316e-05,
      "loss": 4.2663,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.2505,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879286578618586e-05,
      "loss": 4.2497,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878447983867534e-05,
      "loss": 4.2518,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877609389116482e-05,
      "loss": 4.2365,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.2413,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.2486,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8750952427436986e-05,
      "loss": 4.2377,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.202905178070068,
      "eval_runtime": 309.1099,
      "eval_samples_per_second": 1234.483,
      "eval_steps_per_second": 38.579,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8742566479926466e-05,
      "loss": 4.2312,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8734180532415946e-05,
      "loss": 4.2274,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8725794584905426e-05,
      "loss": 4.2408,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717425016198635e-05,
      "loss": 4.2327,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709039068688115e-05,
      "loss": 4.2405,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700653121177595e-05,
      "loss": 4.2188,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692267173667075e-05,
      "loss": 4.2217,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683881226156555e-05,
      "loss": 4.2097,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867551165744977e-05,
      "loss": 4.2204,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866712570993925e-05,
      "loss": 4.2159,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865873976242873e-05,
      "loss": 4.2127,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865035381491821e-05,
      "loss": 4.218,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864196786740769e-05,
      "loss": 4.198,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863358191989717e-05,
      "loss": 4.2114,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862521235119038e-05,
      "loss": 4.1947,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861682640367986e-05,
      "loss": 4.1952,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860844045616934e-05,
      "loss": 4.2004,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860005450865882e-05,
      "loss": 4.1946,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.1972,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.2123,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.1869,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566527097420475e-05,
      "loss": 4.1916,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558157528713684e-05,
      "loss": 4.1892,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.1947,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541385633692644e-05,
      "loss": 4.1891,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532999686182124e-05,
      "loss": 4.1874,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852463011747533e-05,
      "loss": 4.1726,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851624416996481e-05,
      "loss": 4.1793,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850785822245429e-05,
      "loss": 4.1705,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849947227494377e-05,
      "loss": 4.1797,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849108632743325e-05,
      "loss": 4.1734,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848271675872646e-05,
      "loss": 4.1754,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847433081121595e-05,
      "loss": 4.1806,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.1686,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845755891619491e-05,
      "loss": 4.1738,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844918934748812e-05,
      "loss": 4.1721,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84408033999776e-05,
      "loss": 4.1595,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.1544,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842403150495656e-05,
      "loss": 4.1453,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841566193624977e-05,
      "loss": 4.1524,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840727598873925e-05,
      "loss": 4.1536,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839889004122873e-05,
      "loss": 4.1615,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839050409371821e-05,
      "loss": 4.1482,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382134525011416e-05,
      "loss": 4.1471,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83737485775009e-05,
      "loss": 4.1524,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836536262999038e-05,
      "loss": 4.1457,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835697668247986e-05,
      "loss": 4.1489,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834860711377307e-05,
      "loss": 4.1337,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834022116626255e-05,
      "loss": 4.132,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833183521875203e-05,
      "loss": 4.1555,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832344927124151e-05,
      "loss": 4.1515,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831507970253472e-05,
      "loss": 4.1349,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83066937550242e-05,
      "loss": 4.1362,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829832418631741e-05,
      "loss": 4.1414,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828993823880689e-05,
      "loss": 4.1242,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828155229129637e-05,
      "loss": 4.1421,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273166343785856e-05,
      "loss": 4.1159,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264780396275336e-05,
      "loss": 4.1424,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256394448764816e-05,
      "loss": 4.137,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248008501254296e-05,
      "loss": 4.1192,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239638932547505e-05,
      "loss": 4.1172,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231252985036985e-05,
      "loss": 4.1309,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222867037526465e-05,
      "loss": 4.0984,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214481090015945e-05,
      "loss": 4.1139,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206111521309154e-05,
      "loss": 4.1268,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197725573798634e-05,
      "loss": 4.1186,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189339626288114e-05,
      "loss": 4.112,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180953678777594e-05,
      "loss": 4.1011,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817260048887454e-05,
      "loss": 4.1062,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816421454136402e-05,
      "loss": 4.1071,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81558285938535e-05,
      "loss": 4.115,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814744264634298e-05,
      "loss": 4.1046,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813905669883246e-05,
      "loss": 4.1143,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813067075132194e-05,
      "loss": 4.1168,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812228480381141e-05,
      "loss": 4.113,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811389885630089e-05,
      "loss": 4.0988,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810552928759411e-05,
      "loss": 4.1054,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809714334008359e-05,
      "loss": 4.1047,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808875739257306e-05,
      "loss": 4.1025,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808037144506255e-05,
      "loss": 4.1008,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807200187635576e-05,
      "loss": 4.0943,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063615928845243e-05,
      "loss": 4.1026,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055229981334717e-05,
      "loss": 4.0997,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046844033824197e-05,
      "loss": 4.0836,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803847446511741e-05,
      "loss": 4.0898,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030088517606886e-05,
      "loss": 4.0925,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021702570096366e-05,
      "loss": 4.0992,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013316622585846e-05,
      "loss": 4.0898,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800494705387906e-05,
      "loss": 4.0844,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996561106368535e-05,
      "loss": 4.0847,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7988175158858015e-05,
      "loss": 4.0772,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7979789211347494e-05,
      "loss": 4.0782,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797140326383698e-05,
      "loss": 4.0899,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796301731632646e-05,
      "loss": 4.0839,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795463136881594e-05,
      "loss": 4.0936,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794624542130542e-05,
      "loss": 4.0834,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793787585259863e-05,
      "loss": 4.0736,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792948990508811e-05,
      "loss": 4.0739,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792110395757759e-05,
      "loss": 4.0809,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791271801006707e-05,
      "loss": 4.0849,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790434844136028e-05,
      "loss": 4.0827,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789596249384976e-05,
      "loss": 4.0757,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788757654633924e-05,
      "loss": 4.0656,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787920697763245e-05,
      "loss": 4.0827,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870821030121935e-05,
      "loss": 4.0614,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862451461415144e-05,
      "loss": 4.0701,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7854065513904624e-05,
      "loss": 4.0689,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845679566394104e-05,
      "loss": 4.0688,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837293618883584e-05,
      "loss": 4.057,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828907671373064e-05,
      "loss": 4.0678,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820521723862544e-05,
      "loss": 4.0582,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812135776352024e-05,
      "loss": 4.0658,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803749828841504e-05,
      "loss": 4.064,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779538026013471e-05,
      "loss": 4.0543,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778699431262419e-05,
      "loss": 4.0753,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777860836511367e-05,
      "loss": 4.0753,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777023879640689e-05,
      "loss": 4.066,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776185284889637e-05,
      "loss": 4.0556,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775346690138585e-05,
      "loss": 4.0512,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774508095387533e-05,
      "loss": 4.0562,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773669500636481e-05,
      "loss": 4.0545,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772830905885429e-05,
      "loss": 4.0651,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771992311134377e-05,
      "loss": 4.0455,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771155354263698e-05,
      "loss": 4.0551,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770316759512646e-05,
      "loss": 4.0575,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769478164761594e-05,
      "loss": 4.0561,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768639570010542e-05,
      "loss": 4.0434,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76780097525949e-05,
      "loss": 4.0517,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766962380508437e-05,
      "loss": 4.0484,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766123785757386e-05,
      "loss": 4.0522,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765285191006334e-05,
      "loss": 4.0566,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764448234135655e-05,
      "loss": 4.0562,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636096393846026e-05,
      "loss": 4.0365,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7627710446335506e-05,
      "loss": 4.0505,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7619324498824986e-05,
      "loss": 4.0494,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610954930118195e-05,
      "loss": 4.049,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602568982607675e-05,
      "loss": 4.0444,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7594183035097155e-05,
      "loss": 4.052,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7585797087586635e-05,
      "loss": 4.044,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757744389768358e-05,
      "loss": 4.0618,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756905795017306e-05,
      "loss": 4.0364,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756067200266254e-05,
      "loss": 4.0487,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755228605515202e-05,
      "loss": 4.039,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75439001076415e-05,
      "loss": 4.0348,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753551416013098e-05,
      "loss": 4.0387,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752712821262046e-05,
      "loss": 4.0297,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751875864391367e-05,
      "loss": 4.0317,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751037269640315e-05,
      "loss": 4.0364,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750198674889263e-05,
      "loss": 4.0318,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.0336785316467285,
      "eval_runtime": 301.7288,
      "eval_samples_per_second": 1264.682,
      "eval_steps_per_second": 39.522,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.0423436420559053e+17,
  "trial_name": null,
  "trial_params": null
}