{
  "best_metric": 4.342019557952881,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-case/lstm/4/checkpoints/checkpoint-228960",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 228960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8203,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5554,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0617,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9928,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9463,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8938,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7206,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994131474623009e-05,
      "loss": 6.6102,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.5164,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.447,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.3857,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.328,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.2684,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.2061,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1519,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0943,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0496,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.0038,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9633,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9267,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8922,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8559,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8247,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.796,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7777,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7387,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7171,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6922,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.97652262273129e-05,
      "loss": 5.6726,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.975684027980238e-05,
      "loss": 5.6463,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974845433229186e-05,
      "loss": 5.6222,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974006838478134e-05,
      "loss": 5.6155,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973168243727082e-05,
      "loss": 5.5884,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97232964897603e-05,
      "loss": 5.5783,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714926921053515e-05,
      "loss": 5.5648,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5395,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5286,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4887,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4899,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4549,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4551,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4362,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.435,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.4091,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.4015,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.3932,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3755,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3797,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959757279231743e-05,
      "loss": 5.3497,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3351,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3448,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3373,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3124,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.301,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.2988,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.283,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2966,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2574,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.2572,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.2491,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 5.2377,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 5.2392,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 5.2173,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.2118,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.2034,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.2166,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1889,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1878,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1617,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942154978861516e-05,
      "loss": 5.1689,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.1778,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404794272397856e-05,
      "loss": 5.1681,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1465,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1595,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1548,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1311,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.1289,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.1113,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934609263982421e-05,
      "loss": 5.0985,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933770669231369e-05,
      "loss": 5.0977,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.1003,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0981,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.1038,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 5.0892,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295809712368554e-05,
      "loss": 5.0694,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0711,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0705,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0586,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0501,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.0399,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0393,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.0403,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0274,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 5.0294,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 5.0181,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920359704736029e-05,
      "loss": 5.0141,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919521109984978e-05,
      "loss": 4.9998,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 4.9976,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 4.9913,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917006963612195e-05,
      "loss": 4.9944,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916168368861143e-05,
      "loss": 4.9953,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915329774110091e-05,
      "loss": 4.9728,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.978,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9667,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128156277373076e-05,
      "loss": 4.9644,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119770329862556e-05,
      "loss": 4.9573,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9506,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9497,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9094612487330996e-05,
      "loss": 4.9416,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.939,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907784059230996e-05,
      "loss": 4.9309,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906945464479944e-05,
      "loss": 4.9213,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906106869728892e-05,
      "loss": 4.9316,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9052682749778396e-05,
      "loss": 4.9214,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9044296802267876e-05,
      "loss": 4.9222,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.9173,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.9176,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.9086,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.8933,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.901,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994013874812214e-05,
      "loss": 4.9055,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89856279273017e-05,
      "loss": 4.8914,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897724197979118e-05,
      "loss": 4.8845,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896885603228066e-05,
      "loss": 4.882,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896047008477014e-05,
      "loss": 4.8806,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895210051606335e-05,
      "loss": 4.8705,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.894371456855283e-05,
      "loss": 4.8736,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893532862104231e-05,
      "loss": 4.8628,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8634,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8545,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.8728,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.855,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889343164109717e-05,
      "loss": 4.846,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888504569358665e-05,
      "loss": 4.8419,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8876659746076134e-05,
      "loss": 4.8434,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8868273798565614e-05,
      "loss": 4.8446,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859887851055094e-05,
      "loss": 4.8441,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851501903544574e-05,
      "loss": 4.8329,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843115956034054e-05,
      "loss": 4.8338,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834730008523534e-05,
      "loss": 4.8293,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8826344061013014e-05,
      "loss": 4.8271,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.8143,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88095885447957e-05,
      "loss": 4.8265,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880120259728518e-05,
      "loss": 4.8126,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879281664977466e-05,
      "loss": 4.7996,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8145,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.8026,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.8028,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.8086,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7898,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.756908416748047,
      "eval_runtime": 298.6182,
      "eval_samples_per_second": 1277.856,
      "eval_steps_per_second": 39.934,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7827,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.7867,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.788,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717408637394906e-05,
      "loss": 4.7828,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709022689884386e-05,
      "loss": 4.7875,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700636742373866e-05,
      "loss": 4.7709,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692250794863346e-05,
      "loss": 4.7771,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683881226156555e-05,
      "loss": 4.7643,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867549527864604e-05,
      "loss": 4.7734,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.7597,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7591,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865033743611448e-05,
      "loss": 4.7594,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864195148860396e-05,
      "loss": 4.7523,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863356554109344e-05,
      "loss": 4.7478,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862517959358292e-05,
      "loss": 4.7445,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86167936460724e-05,
      "loss": 4.7309,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608407698561874e-05,
      "loss": 4.7425,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600021751051354e-05,
      "loss": 4.7357,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591635803540834e-05,
      "loss": 4.728,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583249856030314e-05,
      "loss": 4.7474,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574863908519794e-05,
      "loss": 4.7299,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566477961009274e-05,
      "loss": 4.7319,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855810839230249e-05,
      "loss": 4.7253,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854972244479197e-05,
      "loss": 4.7435,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854133649728145e-05,
      "loss": 4.7118,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853295054977093e-05,
      "loss": 4.7135,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852456460226041e-05,
      "loss": 4.7168,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851617865474989e-05,
      "loss": 4.7186,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850779270723937e-05,
      "loss": 4.703,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849942313853258e-05,
      "loss": 4.7032,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849105356982579e-05,
      "loss": 4.7127,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848266762231527e-05,
      "loss": 4.7096,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847428167480475e-05,
      "loss": 4.7108,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.7054,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8457526158587444e-05,
      "loss": 4.6966,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449140211076924e-05,
      "loss": 4.7052,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440754263566404e-05,
      "loss": 4.6758,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432368316055884e-05,
      "loss": 4.6963,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423982368545363e-05,
      "loss": 4.6715,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415596421034843e-05,
      "loss": 4.6844,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840721047352432e-05,
      "loss": 4.6757,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83988245260138e-05,
      "loss": 4.6873,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839043857850328e-05,
      "loss": 4.6754,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838205263099276e-05,
      "loss": 4.6738,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373666683482236e-05,
      "loss": 4.676,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365280735971716e-05,
      "loss": 4.6671,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835691116726493e-05,
      "loss": 4.682,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834852521975441e-05,
      "loss": 4.6564,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834013927224389e-05,
      "loss": 4.6505,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833175332473337e-05,
      "loss": 4.67,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832336737722285e-05,
      "loss": 4.6691,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831498142971233e-05,
      "loss": 4.6536,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830659548220181e-05,
      "loss": 4.6484,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829822591349502e-05,
      "loss": 4.6545,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82898399659845e-05,
      "loss": 4.6417,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828145401847398e-05,
      "loss": 4.6647,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827306807096346e-05,
      "loss": 4.6385,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826468212345294e-05,
      "loss": 4.646,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825629617594242e-05,
      "loss": 4.6395,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82479102284319e-05,
      "loss": 4.6333,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823952428092138e-05,
      "loss": 4.6439,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823113833341086e-05,
      "loss": 4.6299,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822275238590034e-05,
      "loss": 4.6259,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214382817193557e-05,
      "loss": 4.6282,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8205996869683037e-05,
      "loss": 4.6424,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197627300976246e-05,
      "loss": 4.6241,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189241353465726e-05,
      "loss": 4.6216,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180855405955206e-05,
      "loss": 4.6148,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172469458444685e-05,
      "loss": 4.6161,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164083510934165e-05,
      "loss": 4.6353,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155697563423645e-05,
      "loss": 4.632,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147327994716854e-05,
      "loss": 4.6147,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138958426010064e-05,
      "loss": 4.627,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813057247849955e-05,
      "loss": 4.6293,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812218653098903e-05,
      "loss": 4.6164,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811380058347851e-05,
      "loss": 4.617,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810541463596799e-05,
      "loss": 4.6064,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809702868845747e-05,
      "loss": 4.5952,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808864274094695e-05,
      "loss": 4.5979,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808025679343642e-05,
      "loss": 4.6045,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807188722472964e-05,
      "loss": 4.6056,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806350127721912e-05,
      "loss": 4.6232,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80551153297086e-05,
      "loss": 4.6089,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804672938219807e-05,
      "loss": 4.5892,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803834343468755e-05,
      "loss": 4.5962,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802995748717703e-05,
      "loss": 4.6017,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802157153966652e-05,
      "loss": 4.5964,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013185592156e-05,
      "loss": 4.5884,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800479964464548e-05,
      "loss": 4.5873,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799643007593869e-05,
      "loss": 4.5864,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798804412842817e-05,
      "loss": 4.5927,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797965818091765e-05,
      "loss": 4.5817,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797127223340713e-05,
      "loss": 4.5869,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796288628589661e-05,
      "loss": 4.5824,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795450033838609e-05,
      "loss": 4.5824,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794611439087557e-05,
      "loss": 4.5724,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793772844336505e-05,
      "loss": 4.5773,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792934249585453e-05,
      "loss": 4.5723,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792095654834401e-05,
      "loss": 4.5749,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791257060083349e-05,
      "loss": 4.5789,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790418465332297e-05,
      "loss": 4.5645,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789581508461618e-05,
      "loss": 4.5674,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788744551590939e-05,
      "loss": 4.5644,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787905956839887e-05,
      "loss": 4.5626,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787067362088835e-05,
      "loss": 4.5564,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786228767337783e-05,
      "loss": 4.5581,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785390172586731e-05,
      "loss": 4.5614,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784553215716052e-05,
      "loss": 4.5513,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783714620965e-05,
      "loss": 4.5545,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782876026213948e-05,
      "loss": 4.5494,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782037431462896e-05,
      "loss": 4.5409,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7811988367118434e-05,
      "loss": 4.5633,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780361879841165e-05,
      "loss": 4.5465,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779523285090114e-05,
      "loss": 4.5533,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786863282194346e-05,
      "loss": 4.5497,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778477334683826e-05,
      "loss": 4.5589,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7770091387173306e-05,
      "loss": 4.5495,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7761705439662786e-05,
      "loss": 4.5353,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775331949215226e-05,
      "loss": 4.5462,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774493354464174e-05,
      "loss": 4.5517,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773654759713122e-05,
      "loss": 4.5364,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77281616496207e-05,
      "loss": 4.5401,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771979208091391e-05,
      "loss": 4.5343,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771140613340339e-05,
      "loss": 4.5426,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7703020185892875e-05,
      "loss": 4.5301,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694634238382355e-05,
      "loss": 4.5373,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686264669675564e-05,
      "loss": 4.5279,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7677878722165044e-05,
      "loss": 4.5341,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669492774654524e-05,
      "loss": 4.5227,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661106827144004e-05,
      "loss": 4.5453,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652720879633484e-05,
      "loss": 4.5349,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644334932122964e-05,
      "loss": 4.52,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7635948984612444e-05,
      "loss": 4.5239,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762757941590565e-05,
      "loss": 4.5253,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761919346839513e-05,
      "loss": 4.5294,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761080752088461e-05,
      "loss": 4.5374,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760242157337409e-05,
      "loss": 4.5186,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759403562586357e-05,
      "loss": 4.5237,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758564967835306e-05,
      "loss": 4.5207,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757728010964627e-05,
      "loss": 4.5229,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756889416213575e-05,
      "loss": 4.5139,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756050821462523e-05,
      "loss": 4.5228,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755212226711471e-05,
      "loss": 4.513,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543769077211646e-05,
      "loss": 4.5074,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535383129701126e-05,
      "loss": 4.5209,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7526997182190606e-05,
      "loss": 4.5128,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518611234680086e-05,
      "loss": 4.5113,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510225287169566e-05,
      "loss": 4.5144,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7501839339659046e-05,
      "loss": 4.5057,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.476215839385986,
      "eval_runtime": 296.0227,
      "eval_samples_per_second": 1289.06,
      "eval_steps_per_second": 40.284,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7493453392148526e-05,
      "loss": 4.4964,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748506744463801e-05,
      "loss": 4.4979,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747668149712749e-05,
      "loss": 4.51,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746829554961697e-05,
      "loss": 4.4999,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7459909602106446e-05,
      "loss": 4.51,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7451523654595926e-05,
      "loss": 4.4954,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7443137707085406e-05,
      "loss": 4.4992,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7434751759574886e-05,
      "loss": 4.4932,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426365812064366e-05,
      "loss": 4.5048,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7417996243357575e-05,
      "loss": 4.4918,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7409610295847055e-05,
      "loss": 4.4915,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401224348336535e-05,
      "loss": 4.4914,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392838400826015e-05,
      "loss": 4.4894,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384452453315495e-05,
      "loss": 4.4841,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376066505804975e-05,
      "loss": 4.4854,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736768055829446e-05,
      "loss": 4.4751,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735929461078394e-05,
      "loss": 4.4805,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735090866327342e-05,
      "loss": 4.4761,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73425227157629e-05,
      "loss": 4.4729,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733413676825238e-05,
      "loss": 4.4989,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732576719954559e-05,
      "loss": 4.479,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.731738125203507e-05,
      "loss": 4.4748,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730899530452455e-05,
      "loss": 4.4789,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730060935701403e-05,
      "loss": 4.4933,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729223978830724e-05,
      "loss": 4.4671,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728385384079672e-05,
      "loss": 4.4709,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.72754678932862e-05,
      "loss": 4.4707,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726708194577568e-05,
      "loss": 4.4775,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725869599826516e-05,
      "loss": 4.4592,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725031005075464e-05,
      "loss": 4.4656,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724192410324412e-05,
      "loss": 4.4678,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.72335381557336e-05,
      "loss": 4.4745,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722515220822308e-05,
      "loss": 4.4796,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721678263951629e-05,
      "loss": 4.4692,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720839669200577e-05,
      "loss": 4.4626,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720001074449525e-05,
      "loss": 4.4718,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719162479698473e-05,
      "loss": 4.4509,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718323884947421e-05,
      "loss": 4.4637,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717485290196369e-05,
      "loss": 4.4459,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71664833332569e-05,
      "loss": 4.4526,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7158097385746384e-05,
      "loss": 4.4548,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7149711438235864e-05,
      "loss": 4.4638,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7141325490725343e-05,
      "loss": 4.4538,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7132939543214823e-05,
      "loss": 4.4517,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7124553595704303e-05,
      "loss": 4.4531,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711616764819378e-05,
      "loss": 4.4486,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710778170068326e-05,
      "loss": 4.4614,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70994285107802e-05,
      "loss": 4.4412,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709104256326968e-05,
      "loss": 4.431,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708265661575916e-05,
      "loss": 4.4559,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.707427066824864e-05,
      "loss": 4.4506,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706588472073812e-05,
      "loss": 4.4405,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70574987732276e-05,
      "loss": 4.44,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704912920452082e-05,
      "loss": 4.4428,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70407432570103e-05,
      "loss": 4.4318,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703235730949978e-05,
      "loss": 4.4518,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702397136198926e-05,
      "loss": 4.432,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701558541447874e-05,
      "loss": 4.4392,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700719946696822e-05,
      "loss": 4.4324,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69988135194577e-05,
      "loss": 4.4331,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990443950750906e-05,
      "loss": 4.4335,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982058003240386e-05,
      "loss": 4.4316,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973672055729866e-05,
      "loss": 4.4238,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965302487023075e-05,
      "loss": 4.4269,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956916539512555e-05,
      "loss": 4.4403,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948530592002035e-05,
      "loss": 4.4267,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940144644491515e-05,
      "loss": 4.4232,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931758696980995e-05,
      "loss": 4.4191,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923372749470475e-05,
      "loss": 4.4201,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6914986801959955e-05,
      "loss": 4.4415,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690661723325317e-05,
      "loss": 4.4354,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898231285742644e-05,
      "loss": 4.4231,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6889845338232124e-05,
      "loss": 4.4328,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881459390721604e-05,
      "loss": 4.4375,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873073443211084e-05,
      "loss": 4.4212,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864687495700564e-05,
      "loss": 4.4259,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856301548190044e-05,
      "loss": 4.4203,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684793197948325e-05,
      "loss": 4.4105,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683954603197274e-05,
      "loss": 4.4167,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683116008446222e-05,
      "loss": 4.4128,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68227741369517e-05,
      "loss": 4.42,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681438818944118e-05,
      "loss": 4.4327,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680600224193066e-05,
      "loss": 4.429,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679761629442014e-05,
      "loss": 4.4023,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678923034690962e-05,
      "loss": 4.4198,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678086077820283e-05,
      "loss": 4.416,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677247483069231e-05,
      "loss": 4.4188,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676408888318179e-05,
      "loss": 4.4118,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755719314475e-05,
      "loss": 4.4044,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674733336696448e-05,
      "loss": 4.4119,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673894741945396e-05,
      "loss": 4.4094,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673056147194344e-05,
      "loss": 4.4049,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672219190323665e-05,
      "loss": 4.4104,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671380595572613e-05,
      "loss": 4.4042,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670542000821561e-05,
      "loss": 4.4095,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669703406070509e-05,
      "loss": 4.3983,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668864811319457e-05,
      "loss": 4.4009,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668026216568405e-05,
      "loss": 4.3992,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667187621817353e-05,
      "loss": 4.4075,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666350664946674e-05,
      "loss": 4.4067,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665512070195622e-05,
      "loss": 4.3931,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66467347544457e-05,
      "loss": 4.3971,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6638348806935175e-05,
      "loss": 4.3958,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662996285942466e-05,
      "loss": 4.3972,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662157691191414e-05,
      "loss": 4.3863,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661320734320736e-05,
      "loss": 4.3885,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660482139569683e-05,
      "loss": 4.3991,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659643544818631e-05,
      "loss": 4.391,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658804950067579e-05,
      "loss": 4.3873,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657966355316527e-05,
      "loss": 4.3856,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657127760565475e-05,
      "loss": 4.3794,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656289165814423e-05,
      "loss": 4.3972,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655452208943744e-05,
      "loss": 4.3897,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654615252073065e-05,
      "loss": 4.3851,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653776657322013e-05,
      "loss": 4.3887,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529380625709615e-05,
      "loss": 4.4018,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6520994678199095e-05,
      "loss": 4.3901,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512608730688575e-05,
      "loss": 4.3772,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504222783178055e-05,
      "loss": 4.3875,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495836835667535e-05,
      "loss": 4.3912,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487450888157015e-05,
      "loss": 4.3814,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479081319450224e-05,
      "loss": 4.3844,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6470695371939704e-05,
      "loss": 4.3795,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646232580323291e-05,
      "loss": 4.3883,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645393985572239e-05,
      "loss": 4.3748,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644555390821187e-05,
      "loss": 4.3865,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643716796070135e-05,
      "loss": 4.3691,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642878201319083e-05,
      "loss": 4.3849,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642039606568031e-05,
      "loss": 4.368,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64120101181698e-05,
      "loss": 4.3928,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640362417065928e-05,
      "loss": 4.3853,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639523822314876e-05,
      "loss": 4.3749,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638686865444197e-05,
      "loss": 4.3695,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637848270693145e-05,
      "loss": 4.3771,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637009675942093e-05,
      "loss": 4.3769,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636171081191041e-05,
      "loss": 4.3875,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635332486439989e-05,
      "loss": 4.3679,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634493891688936e-05,
      "loss": 4.3778,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633655296937884e-05,
      "loss": 4.3738,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632816702186832e-05,
      "loss": 4.3809,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63197810743578e-05,
      "loss": 4.3641,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631141150565102e-05,
      "loss": 4.377,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630304193694423e-05,
      "loss": 4.3699,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629465598943371e-05,
      "loss": 4.3633,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628627004192319e-05,
      "loss": 4.3757,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277884094412666e-05,
      "loss": 4.3724,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269498146902146e-05,
      "loss": 4.3624,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261112199391626e-05,
      "loss": 4.374,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6252726251881106e-05,
      "loss": 4.3637,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.342019557952881,
      "eval_runtime": 293.1681,
      "eval_samples_per_second": 1301.612,
      "eval_steps_per_second": 40.676,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.39388847862785e+16,
  "trial_name": null,
  "trial_params": null
}