{
  "best_metric": 4.768176555633545,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full2/lstm/0/checkpoints/checkpoint-76320",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 76320,
  "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": 10.8214,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5602,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.058,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9951,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9498,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8998,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7169,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994131474623009e-05,
      "loss": 6.6079,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.5124,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.4465,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.3968,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3509,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.2991,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.2314,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1847,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.1285,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0873,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.0405,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9979,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.95,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.9171,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8771,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8495,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.8067,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979875363855125e-05,
      "loss": 5.7943,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979036769104073e-05,
      "loss": 5.762,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 5.7333,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.6998,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6794,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6642,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6328,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 5.6229,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.5966,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5783,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5755,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.558,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.533,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 5.5055,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 5.4963,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 5.472,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 5.4702,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.4463,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 5.4402,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.4154,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.4164,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.4033,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.4023,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3899,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3606,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3523,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 5.3614,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.342,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3285,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.3105,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.3129,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2933,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.3049,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2688,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.2716,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.2677,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949699055860238e-05,
      "loss": 5.2519,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948860461109186e-05,
      "loss": 5.2589,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.2281,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.2234,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.2122,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.2322,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1984,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.2,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1829,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942153340981143e-05,
      "loss": 5.176,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.1845,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.1766,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93963919460836e-05,
      "loss": 5.1683,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938800599857309e-05,
      "loss": 5.1671,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937962005106256e-05,
      "loss": 5.17,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937123410355204e-05,
      "loss": 5.1369,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.1444,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354494966138465e-05,
      "loss": 5.1229,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.1172,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.1179,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.1183,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.1074,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.1141,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0963,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0776,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.073,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0864,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0663,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0641,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0606,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0481,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0503,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.044,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 5.0421,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 5.0395,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 5.0307,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919522747865351e-05,
      "loss": 5.0134,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 5.0157,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 5.0042,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 5.0036,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 5.0109,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.9903,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.994,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.9877,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.977,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.9688,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111400761155765e-05,
      "loss": 4.9678,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9103014813645245e-05,
      "loss": 4.9522,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909462886613473e-05,
      "loss": 4.9542,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908625929742794e-05,
      "loss": 4.9508,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907787334991742e-05,
      "loss": 4.9503,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90694874024069e-05,
      "loss": 4.942,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906110145489638e-05,
      "loss": 4.9391,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905271550738586e-05,
      "loss": 4.9304,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904432955987534e-05,
      "loss": 4.9339,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903594361236482e-05,
      "loss": 4.9347,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902757404365803e-05,
      "loss": 4.9285,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901918809614751e-05,
      "loss": 4.9281,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901080214863699e-05,
      "loss": 4.9148,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900241620112647e-05,
      "loss": 4.9169,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.9137,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8992,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8968,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.895,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896048646357387e-05,
      "loss": 4.8944,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895210051606335e-05,
      "loss": 4.8814,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8867,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8797,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8735,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8651,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910203536118214e-05,
      "loss": 4.8823,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890181758860769e-05,
      "loss": 4.873,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889343164109717e-05,
      "loss": 4.865,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888504569358665e-05,
      "loss": 4.8633,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8876659746076134e-05,
      "loss": 4.8613,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8868273798565614e-05,
      "loss": 4.8595,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859887851055094e-05,
      "loss": 4.8591,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851501903544574e-05,
      "loss": 4.8498,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843115956034054e-05,
      "loss": 4.8446,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834730008523534e-05,
      "loss": 4.8421,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8299,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881799087110995e-05,
      "loss": 4.826,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.8261,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.8302,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879283302857839e-05,
      "loss": 4.8226,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878444708106787e-05,
      "loss": 4.8218,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877606113355735e-05,
      "loss": 4.8172,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876767518604683e-05,
      "loss": 4.8179,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875928923853632e-05,
      "loss": 4.8237,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87509032910258e-05,
      "loss": 4.7998,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.768176555633545,
      "eval_runtime": 323.1793,
      "eval_samples_per_second": 1180.741,
      "eval_steps_per_second": 36.899,
      "step": 76320
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 3.1531030825875588e+16,
  "trial_name": null,
  "trial_params": null
}