{
  "best_metric": 4.478839874267578,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/det-adj-noun/lstm/1/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": 10.8186,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.551,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0602,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9931,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9434,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9023,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7401,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6279,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.5318,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.4591,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.3961,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3289,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.2531,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.1894,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1285,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0728,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0287,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.9906,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9456,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9063,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8717,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8261,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8038,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7662,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7483,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7099,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.6836,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.659,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6425,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6187,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.5964,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.5864,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5527,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.5444,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5271,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5114,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.496,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4582,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96814158886189e-05,
      "loss": 5.4639,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 5.4313,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 5.4355,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.414,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 5.4168,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.3879,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.3739,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.38,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3681,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3564,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3393,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3183,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3285,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3102,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.31,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.2787,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.2827,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2614,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 5.2826,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 5.2458,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.248,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.2348,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949699055860238e-05,
      "loss": 5.2221,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948860461109186e-05,
      "loss": 5.2231,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.2118,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.1866,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94634467685603e-05,
      "loss": 5.1906,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.2029,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1799,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1696,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1508,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942153340981143e-05,
      "loss": 5.1535,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.1713,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.154,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1396,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1438,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1401,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1216,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.1074,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354494966138465e-05,
      "loss": 5.1118,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.0924,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.0938,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.0992,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.086,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.0879,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0759,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0589,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287407386054305e-05,
      "loss": 5.0594,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279021438543785e-05,
      "loss": 5.0552,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0613,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0436,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0443,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0414,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0348,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0318,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92203525635776e-05,
      "loss": 5.0193,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921196661606708e-05,
      "loss": 5.0059,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920358066855656e-05,
      "loss": 5.0134,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919519472104604e-05,
      "loss": 5.0012,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9967,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9895,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9892,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 4.9931,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915328136229718e-05,
      "loss": 4.9817,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.9703,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9686,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.9592,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119770329862556e-05,
      "loss": 4.9457,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9412,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9463,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9094612487330996e-05,
      "loss": 4.9429,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.9298,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9282,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9277,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9327,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9282,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904432955987534e-05,
      "loss": 4.9188,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903594361236482e-05,
      "loss": 4.9176,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.9164,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.903,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.8971,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.9027,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994013874812214e-05,
      "loss": 4.9014,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89856279273017e-05,
      "loss": 4.8797,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897724197979118e-05,
      "loss": 4.8812,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8798,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896048646357387e-05,
      "loss": 4.8701,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895210051606335e-05,
      "loss": 4.8725,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.894371456855283e-05,
      "loss": 4.8759,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8598,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8609,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8498,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.8669,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.8584,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889343164109717e-05,
      "loss": 4.8482,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888504569358665e-05,
      "loss": 4.8454,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8876659746076134e-05,
      "loss": 4.8462,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8868273798565614e-05,
      "loss": 4.8441,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859887851055094e-05,
      "loss": 4.8347,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851501903544574e-05,
      "loss": 4.8328,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843115956034054e-05,
      "loss": 4.8329,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834730008523534e-05,
      "loss": 4.829,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8293,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.8072,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88095885447957e-05,
      "loss": 4.8192,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880120259728518e-05,
      "loss": 4.8053,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.795,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.813,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.8024,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.802,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.8046,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7832,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.759700298309326,
      "eval_runtime": 311.2482,
      "eval_samples_per_second": 1226.002,
      "eval_steps_per_second": 38.313,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7856,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.7831,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.7993,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7766,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709006311080657e-05,
      "loss": 4.7869,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700620363570137e-05,
      "loss": 4.766,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.7711,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683864847352826e-05,
      "loss": 4.7609,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867549527864604e-05,
      "loss": 4.766,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.7659,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7689,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865033743611448e-05,
      "loss": 4.7688,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864196786740769e-05,
      "loss": 4.7479,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863358191989717e-05,
      "loss": 4.7479,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862519597238665e-05,
      "loss": 4.7431,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861681002487613e-05,
      "loss": 4.7356,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860842407736561e-05,
      "loss": 4.7496,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860003812985509e-05,
      "loss": 4.7346,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859165218234457e-05,
      "loss": 4.7333,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858326623483405e-05,
      "loss": 4.7523,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857488028732352e-05,
      "loss": 4.731,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856649433981301e-05,
      "loss": 4.7313,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855810839230249e-05,
      "loss": 4.7282,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85497388235957e-05,
      "loss": 4.7342,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854135287608518e-05,
      "loss": 4.7226,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853296692857466e-05,
      "loss": 4.7158,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524613738671604e-05,
      "loss": 4.7108,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8516227791161084e-05,
      "loss": 4.7203,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8507841843650564e-05,
      "loss": 4.702,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8499455896140044e-05,
      "loss": 4.7094,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8491069948629524e-05,
      "loss": 4.7088,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8482684001119e-05,
      "loss": 4.7017,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847429805360848e-05,
      "loss": 4.7078,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.6976,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845754253739117e-05,
      "loss": 4.6985,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844915658988065e-05,
      "loss": 4.6997,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844078702117387e-05,
      "loss": 4.6748,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843240107366335e-05,
      "loss": 4.6872,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842401512615282e-05,
      "loss": 4.6704,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84156291786423e-05,
      "loss": 4.6817,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840724323113178e-05,
      "loss": 4.682,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839885728362126e-05,
      "loss": 4.687,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839047133611074e-05,
      "loss": 4.6763,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838208538860022e-05,
      "loss": 4.6731,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837371581989343e-05,
      "loss": 4.6788,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836532987238292e-05,
      "loss": 4.6745,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83569439248724e-05,
      "loss": 4.6708,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834855797736188e-05,
      "loss": 4.6632,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.6516,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8331802461144566e-05,
      "loss": 4.6678,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8323416513634046e-05,
      "loss": 4.6608,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8315030566123526e-05,
      "loss": 4.6604,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8306660997416735e-05,
      "loss": 4.6481,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298275049906215e-05,
      "loss": 4.6555,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289889102395695e-05,
      "loss": 4.6405,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281503154885175e-05,
      "loss": 4.6616,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273117207374655e-05,
      "loss": 4.6405,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264731259864135e-05,
      "loss": 4.6505,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256345312353615e-05,
      "loss": 4.6412,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247959364843095e-05,
      "loss": 4.6331,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823958979613631e-05,
      "loss": 4.6359,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823122022742952e-05,
      "loss": 4.6398,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222834279919e-05,
      "loss": 4.6152,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821444833240848e-05,
      "loss": 4.6241,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820606238489796e-05,
      "loss": 4.6364,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819769281619117e-05,
      "loss": 4.6236,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818930686868065e-05,
      "loss": 4.6181,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818092092117013e-05,
      "loss": 4.6138,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817253497365961e-05,
      "loss": 4.6122,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816414902614909e-05,
      "loss": 4.6334,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155779457442305e-05,
      "loss": 4.6266,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147393509931785e-05,
      "loss": 4.6162,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8139007562421265e-05,
      "loss": 4.622,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8130621614910745e-05,
      "loss": 4.6244,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8122252046203954e-05,
      "loss": 4.615,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113866098693434e-05,
      "loss": 4.6003,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105480151182914e-05,
      "loss": 4.6137,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8097094203672393e-05,
      "loss": 4.6035,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088708256161873e-05,
      "loss": 4.599,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808032230865135e-05,
      "loss": 4.6127,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8071936361140827e-05,
      "loss": 4.6019,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063550413630307e-05,
      "loss": 4.6097,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805518084492352e-05,
      "loss": 4.6053,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046794897413e-05,
      "loss": 4.5833,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803840894990248e-05,
      "loss": 4.5968,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803002300239196e-05,
      "loss": 4.5905,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802165343368518e-05,
      "loss": 4.603,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801326748617466e-05,
      "loss": 4.5916,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800488153866413e-05,
      "loss": 4.5949,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799649559115361e-05,
      "loss": 4.5936,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798810964364309e-05,
      "loss": 4.59,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79797400749363e-05,
      "loss": 4.5925,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797135412742578e-05,
      "loss": 4.5854,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7962984558718996e-05,
      "loss": 4.5728,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7954598611208476e-05,
      "loss": 4.5863,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7946212663697956e-05,
      "loss": 4.5836,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7937826716187436e-05,
      "loss": 4.574,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7929440768676916e-05,
      "loss": 4.5795,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7921054821166396e-05,
      "loss": 4.5776,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912668873655876e-05,
      "loss": 4.5893,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7904299304949085e-05,
      "loss": 4.5732,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895913357438565e-05,
      "loss": 4.5645,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887527409928045e-05,
      "loss": 4.5686,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879141462417525e-05,
      "loss": 4.5651,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870755514907005e-05,
      "loss": 4.5477,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862369567396485e-05,
      "loss": 4.5603,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853983619885965e-05,
      "loss": 4.5593,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845597672375445e-05,
      "loss": 4.561,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837211724864925e-05,
      "loss": 4.5529,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782884215615814e-05,
      "loss": 4.5452,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782045620864762e-05,
      "loss": 4.5532,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78120702611371e-05,
      "loss": 4.5634,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780368431362658e-05,
      "loss": 4.558,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779531474491979e-05,
      "loss": 4.5567,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778692879740927e-05,
      "loss": 4.5524,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777855922870248e-05,
      "loss": 4.5632,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777017328119196e-05,
      "loss": 4.5459,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776178733368144e-05,
      "loss": 4.5429,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775340138617092e-05,
      "loss": 4.5508,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77450154386604e-05,
      "loss": 4.5517,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773662949114988e-05,
      "loss": 4.5316,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772824354363936e-05,
      "loss": 4.5344,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719857596128845e-05,
      "loss": 4.5404,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771147164861832e-05,
      "loss": 4.5322,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77030857011078e-05,
      "loss": 4.538,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694716132401014e-05,
      "loss": 4.5442,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686330184890494e-05,
      "loss": 4.5237,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767794423737997e-05,
      "loss": 4.5332,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766955828986945e-05,
      "loss": 4.5213,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766118872116266e-05,
      "loss": 4.5464,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652802773652136e-05,
      "loss": 4.5379,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644416826141616e-05,
      "loss": 4.5261,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636030878631096e-05,
      "loss": 4.5273,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762764493112058e-05,
      "loss": 4.5275,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761927536241379e-05,
      "loss": 4.5339,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761088941490327e-05,
      "loss": 4.5279,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760250346739275e-05,
      "loss": 4.5229,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759411751988223e-05,
      "loss": 4.5236,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758573157237171e-05,
      "loss": 4.5206,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757736200366492e-05,
      "loss": 4.524,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75689760561544e-05,
      "loss": 4.511,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756059010864388e-05,
      "loss": 4.5212,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755220416113336e-05,
      "loss": 4.5101,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754381821362284e-05,
      "loss": 4.5004,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753544864491605e-05,
      "loss": 4.5219,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527062697405536e-05,
      "loss": 4.5108,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518676749895016e-05,
      "loss": 4.5133,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510290802384496e-05,
      "loss": 4.5144,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7501904854873976e-05,
      "loss": 4.4992,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.478839874267578,
      "eval_runtime": 333.4901,
      "eval_samples_per_second": 1144.235,
      "eval_steps_per_second": 35.758,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 6.272835075160013e+16,
  "trial_name": null,
  "trial_params": null
}