{
  "best_metric": 4.349167346954346,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/re-irr-sv-agr/lstm/0/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": 10.8198,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5444,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0493,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9839,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9381,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.879,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.6972,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994131474623009e-05,
      "loss": 6.5963,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.4992,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.4332,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.375,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3236,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.2567,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.1862,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1315,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0673,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0263,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.9832,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9401,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9022,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8732,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.831,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8059,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.7615,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7477,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7157,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.6952,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6681,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6356,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6184,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.5935,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.5844,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5577,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.5458,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714926921053515e-05,
      "loss": 5.5353,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5128,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5024,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4586,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4545,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9673013562304644e-05,
      "loss": 5.4341,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4197,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4159,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4078,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.3784,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.3794,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.3738,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3611,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3548,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959757279231743e-05,
      "loss": 5.3395,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3168,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3194,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3012,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.2726,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.2845,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2577,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2741,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2461,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2331,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.2281,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2223,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948858823228813e-05,
      "loss": 5.2089,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.2097,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.1824,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.1777,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.1896,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1812,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1656,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1467,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942153340981143e-05,
      "loss": 5.1466,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941316384110464e-05,
      "loss": 5.1461,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.1398,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93963919460836e-05,
      "loss": 5.1359,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938800599857309e-05,
      "loss": 5.1295,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1314,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1229,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.0963,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.0932,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934609263982421e-05,
      "loss": 5.0903,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.0773,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.0854,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 5.0695,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 5.0863,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 5.0594,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295809712368554e-05,
      "loss": 5.0441,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0498,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0495,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0454,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0322,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.028,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0201,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0178,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0082,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92203525635776e-05,
      "loss": 5.0043,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921196661606708e-05,
      "loss": 5.0017,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920358066855656e-05,
      "loss": 5.0073,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919519472104604e-05,
      "loss": 4.9857,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9812,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.974,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9696,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 4.9745,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915329774110091e-05,
      "loss": 4.9612,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.9549,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9466,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.9508,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9226,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911136800354831e-05,
      "loss": 4.9364,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910298205603779e-05,
      "loss": 4.9256,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909459610852727e-05,
      "loss": 4.9319,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.9191,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907784059230996e-05,
      "loss": 4.926,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906945464479944e-05,
      "loss": 4.9108,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9064,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9152,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.8907,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.9082,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.8988,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.8949,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.8823,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.8822,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994013874812214e-05,
      "loss": 4.8844,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89856279273017e-05,
      "loss": 4.877,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897724197979118e-05,
      "loss": 4.8756,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896885603228066e-05,
      "loss": 4.866,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896047008477014e-05,
      "loss": 4.8683,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895208413725962e-05,
      "loss": 4.8476,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89436981897491e-05,
      "loss": 4.8526,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893531224223858e-05,
      "loss": 4.834,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892694267353179e-05,
      "loss": 4.8415,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891855672602127e-05,
      "loss": 4.8277,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891017077851075e-05,
      "loss": 4.8489,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890178483100023e-05,
      "loss": 4.8541,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889341526229344e-05,
      "loss": 4.8379,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888502931478292e-05,
      "loss": 4.817,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766433672724e-05,
      "loss": 4.8285,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886825741976188e-05,
      "loss": 4.8264,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859871472251365e-05,
      "loss": 4.8203,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851485524740845e-05,
      "loss": 4.8159,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843099577230325e-05,
      "loss": 4.8175,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834713629719805e-05,
      "loss": 4.8132,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882632768220928e-05,
      "loss": 4.8018,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881794173469876e-05,
      "loss": 4.8021,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8809572165991974e-05,
      "loss": 4.803,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8801186218481454e-05,
      "loss": 4.7911,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879280027097093e-05,
      "loss": 4.7917,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878441432346041e-05,
      "loss": 4.7991,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877602837594989e-05,
      "loss": 4.7873,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876764242843937e-05,
      "loss": 4.7853,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875927285973258e-05,
      "loss": 4.7972,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875088691222206e-05,
      "loss": 4.774,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.762792587280273,
      "eval_runtime": 291.1775,
      "eval_samples_per_second": 1310.51,
      "eval_steps_per_second": 40.954,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874250096471154e-05,
      "loss": 4.7628,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873411501720102e-05,
      "loss": 4.769,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872574544849423e-05,
      "loss": 4.7819,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871735950098371e-05,
      "loss": 4.7705,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870897355347319e-05,
      "loss": 4.7759,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870058760596267e-05,
      "loss": 4.7498,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869220165845215e-05,
      "loss": 4.7676,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868381571094163e-05,
      "loss": 4.7419,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867544614223484e-05,
      "loss": 4.7543,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667076573528056e-05,
      "loss": 4.7434,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658690626017536e-05,
      "loss": 4.7575,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8650304678507016e-05,
      "loss": 4.7524,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8641918730996496e-05,
      "loss": 4.7396,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633532783485976e-05,
      "loss": 4.7338,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625146835975456e-05,
      "loss": 4.7208,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616760888464936e-05,
      "loss": 4.7233,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608374940954416e-05,
      "loss": 4.7325,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600021751051354e-05,
      "loss": 4.7283,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591635803540834e-05,
      "loss": 4.7215,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583249856030314e-05,
      "loss": 4.7366,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574863908519794e-05,
      "loss": 4.7296,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566477961009274e-05,
      "loss": 4.7214,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558092013498754e-05,
      "loss": 4.7104,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854970606598824e-05,
      "loss": 4.7308,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854132011847772e-05,
      "loss": 4.7053,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85329341709672e-05,
      "loss": 4.7145,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852454822345668e-05,
      "loss": 4.7124,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851616227594616e-05,
      "loss": 4.7057,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850779270723937e-05,
      "loss": 4.6945,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849940675972885e-05,
      "loss": 4.6944,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849103719102206e-05,
      "loss": 4.697,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848265124351154e-05,
      "loss": 4.6944,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847426529600102e-05,
      "loss": 4.6931,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84658793484905e-05,
      "loss": 4.6999,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845750977978371e-05,
      "loss": 4.6895,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449123832273194e-05,
      "loss": 4.6887,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440737884762674e-05,
      "loss": 4.6683,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432351937252154e-05,
      "loss": 4.6736,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423965989741634e-05,
      "loss": 4.6663,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415596421034843e-05,
      "loss": 4.6614,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840721047352432e-05,
      "loss": 4.6725,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83988245260138e-05,
      "loss": 4.6746,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839043857850328e-05,
      "loss": 4.657,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838205263099276e-05,
      "loss": 4.6639,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373666683482236e-05,
      "loss": 4.6618,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365280735971716e-05,
      "loss": 4.6646,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8356894788461196e-05,
      "loss": 4.6657,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834852521975441e-05,
      "loss": 4.6585,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834013927224389e-05,
      "loss": 4.6461,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833175332473337e-05,
      "loss": 4.6532,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832336737722285e-05,
      "loss": 4.6421,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831498142971233e-05,
      "loss": 4.651,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830661186100554e-05,
      "loss": 4.6366,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829822591349502e-05,
      "loss": 4.6486,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82898399659845e-05,
      "loss": 4.6268,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828145401847398e-05,
      "loss": 4.6499,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827308444976719e-05,
      "loss": 4.633,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264714881060406e-05,
      "loss": 4.6272,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256328933549886e-05,
      "loss": 4.6359,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247942986039366e-05,
      "loss": 4.6271,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239557038528846e-05,
      "loss": 4.6203,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231171091018326e-05,
      "loss": 4.6338,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222785143507806e-05,
      "loss": 4.6041,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214415574801015e-05,
      "loss": 4.6093,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820604600609423e-05,
      "loss": 4.6234,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819766005858371e-05,
      "loss": 4.6222,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189274111073184e-05,
      "loss": 4.6112,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180888163562664e-05,
      "loss": 4.6033,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172502216052144e-05,
      "loss": 4.6043,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164116268541624e-05,
      "loss": 4.6073,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155730321031104e-05,
      "loss": 4.6143,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147344373520584e-05,
      "loss": 4.6068,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138958426010064e-05,
      "loss": 4.6083,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813057247849955e-05,
      "loss": 4.6134,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812218653098903e-05,
      "loss": 4.6102,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811380058347851e-05,
      "loss": 4.5932,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810541463596799e-05,
      "loss": 4.5924,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80970450672612e-05,
      "loss": 4.5992,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808867549855441e-05,
      "loss": 4.5843,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808028955104389e-05,
      "loss": 4.5975,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807190360353337e-05,
      "loss": 4.5821,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806351765602285e-05,
      "loss": 4.6073,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805513170851233e-05,
      "loss": 4.588,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804674576100181e-05,
      "loss": 4.5731,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803837619229502e-05,
      "loss": 4.5832,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8029990244784504e-05,
      "loss": 4.5852,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021604297273984e-05,
      "loss": 4.587,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013218349763464e-05,
      "loss": 4.5773,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8004832402252944e-05,
      "loss": 4.5764,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996446454742424e-05,
      "loss": 4.5804,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79880605072319e-05,
      "loss": 4.5702,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797967455972138e-05,
      "loss": 4.571,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797130499101459e-05,
      "loss": 4.5689,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79629354223078e-05,
      "loss": 4.5758,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795454947479728e-05,
      "loss": 4.5743,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794616352728676e-05,
      "loss": 4.5692,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793777757977624e-05,
      "loss": 4.5575,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792939163226572e-05,
      "loss": 4.5567,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792102206355894e-05,
      "loss": 4.5633,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791263611604842e-05,
      "loss": 4.561,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79042501685379e-05,
      "loss": 4.558,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789586422102737e-05,
      "loss": 4.5533,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788747827351685e-05,
      "loss": 4.5469,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787909232600633e-05,
      "loss": 4.5575,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787070637849581e-05,
      "loss": 4.5267,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786232043098529e-05,
      "loss": 4.5515,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78539508622785e-05,
      "loss": 4.542,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784556491476798e-05,
      "loss": 4.5491,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783717896725746e-05,
      "loss": 4.5381,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828809398550675e-05,
      "loss": 4.5489,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820423451040155e-05,
      "loss": 4.5371,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812037503529635e-05,
      "loss": 4.5349,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803651556019115e-05,
      "loss": 4.5449,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795265608508595e-05,
      "loss": 4.5284,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786879660998075e-05,
      "loss": 4.5438,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778493713487555e-05,
      "loss": 4.5417,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7770107765977035e-05,
      "loss": 4.5358,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7761721818466515e-05,
      "loss": 4.53,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7753352249759724e-05,
      "loss": 4.5313,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7744966302249204e-05,
      "loss": 4.536,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7736580354738684e-05,
      "loss": 4.5209,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7728194407228164e-05,
      "loss": 4.5374,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771982483852137e-05,
      "loss": 4.5209,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771145526981459e-05,
      "loss": 4.5267,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770306932230407e-05,
      "loss": 4.5194,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769468337479355e-05,
      "loss": 4.5195,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768629742728303e-05,
      "loss": 4.5013,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767791147977251e-05,
      "loss": 4.5151,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766954191106572e-05,
      "loss": 4.5031,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76611559635552e-05,
      "loss": 4.5245,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765277001604468e-05,
      "loss": 4.536,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764438406853416e-05,
      "loss": 4.5184,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763599812102364e-05,
      "loss": 4.5018,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762761217351312e-05,
      "loss": 4.5142,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76192262260026e-05,
      "loss": 4.509,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761084027849208e-05,
      "loss": 4.5127,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602470709785293e-05,
      "loss": 4.5103,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7594084762274773e-05,
      "loss": 4.5077,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758569881476425e-05,
      "loss": 4.5049,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757731286725373e-05,
      "loss": 4.5033,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7568926919743206e-05,
      "loss": 4.5038,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7560540972232686e-05,
      "loss": 4.5069,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75521714035259e-05,
      "loss": 4.4991,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543785456015375e-05,
      "loss": 4.4934,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753541588730859e-05,
      "loss": 4.5072,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752702993979807e-05,
      "loss": 4.4999,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751864399228755e-05,
      "loss": 4.4918,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751025804477703e-05,
      "loss": 4.51,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750187209726651e-05,
      "loss": 4.4898,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.483022689819336,
      "eval_runtime": 291.5237,
      "eval_samples_per_second": 1308.953,
      "eval_steps_per_second": 40.906,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749348614975599e-05,
      "loss": 4.4785,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748510020224547e-05,
      "loss": 4.4841,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747671425473495e-05,
      "loss": 4.5,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746832830722443e-05,
      "loss": 4.4919,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745994235971391e-05,
      "loss": 4.4962,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745155641220339e-05,
      "loss": 4.4742,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744317046469287e-05,
      "loss": 4.4932,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743478451718235e-05,
      "loss": 4.4666,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742639856967183e-05,
      "loss": 4.4898,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741801262216131e-05,
      "loss": 4.4717,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740962667465079e-05,
      "loss": 4.4913,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401240727140264e-05,
      "loss": 4.4864,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.739287115843348e-05,
      "loss": 4.4771,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384485210922966e-05,
      "loss": 4.4721,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376099263412446e-05,
      "loss": 4.4657,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736771331590192e-05,
      "loss": 4.4653,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73593273683914e-05,
      "loss": 4.4715,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735094142088088e-05,
      "loss": 4.4718,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734255547337036e-05,
      "loss": 4.4707,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733416952585984e-05,
      "loss": 4.4819,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732578357834932e-05,
      "loss": 4.4805,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73173976308388e-05,
      "loss": 4.4734,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730901168332828e-05,
      "loss": 4.4651,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730064211462149e-05,
      "loss": 4.4773,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729225616711097e-05,
      "loss": 4.4617,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728387021960045e-05,
      "loss": 4.4696,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7275500650893664e-05,
      "loss": 4.4669,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7267114703383144e-05,
      "loss": 4.4637,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258728755872624e-05,
      "loss": 4.4544,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250342808362104e-05,
      "loss": 4.456,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7241956860851584e-05,
      "loss": 4.4581,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7233570913341064e-05,
      "loss": 4.4597,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7225184965830544e-05,
      "loss": 4.4624,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7216799018320024e-05,
      "loss": 4.458,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720844582841696e-05,
      "loss": 4.4543,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720005988090644e-05,
      "loss": 4.4556,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719167393339592e-05,
      "loss": 4.4377,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71832879858854e-05,
      "loss": 4.4479,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717490203837488e-05,
      "loss": 4.4401,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716651609086437e-05,
      "loss": 4.4382,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715814652215758e-05,
      "loss": 4.4441,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714976057464706e-05,
      "loss": 4.4516,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714137462713654e-05,
      "loss": 4.4379,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713298867962602e-05,
      "loss": 4.4387,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71246027321155e-05,
      "loss": 4.4473,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711621678460498e-05,
      "loss": 4.4389,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710783083709446e-05,
      "loss": 4.453,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709944488958393e-05,
      "loss": 4.4379,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7091075320877147e-05,
      "loss": 4.4267,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082689373366627e-05,
      "loss": 4.4413,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074303425856106e-05,
      "loss": 4.4297,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7065917478345586e-05,
      "loss": 4.4388,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7057531530835066e-05,
      "loss": 4.4254,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7049145583324546e-05,
      "loss": 4.4333,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7040776014617755e-05,
      "loss": 4.4218,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7032390067107235e-05,
      "loss": 4.4354,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7024004119596715e-05,
      "loss": 4.4246,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7015634550889924e-05,
      "loss": 4.4238,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007248603379404e-05,
      "loss": 4.4303,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6998862655868884e-05,
      "loss": 4.4198,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990476708358364e-05,
      "loss": 4.4145,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982090760847844e-05,
      "loss": 4.436,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973704813337324e-05,
      "loss": 4.4026,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965318865826804e-05,
      "loss": 4.4075,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695693291831629e-05,
      "loss": 4.417,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694854697080577e-05,
      "loss": 4.4282,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694017740209898e-05,
      "loss": 4.4141,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693179145458846e-05,
      "loss": 4.4055,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.692340550707794e-05,
      "loss": 4.4084,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691501955956742e-05,
      "loss": 4.4106,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69066336120569e-05,
      "loss": 4.42,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689826404335011e-05,
      "loss": 4.415,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688987809583959e-05,
      "loss": 4.4146,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688149214832907e-05,
      "loss": 4.4188,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687310620081855e-05,
      "loss": 4.4213,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686473663211176e-05,
      "loss": 4.4043,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856367063404974e-05,
      "loss": 4.4028,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6847981115894454e-05,
      "loss": 4.414,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6839595168383934e-05,
      "loss": 4.3953,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6831209220873414e-05,
      "loss": 4.4116,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6822823273362894e-05,
      "loss": 4.3976,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6814437325852373e-05,
      "loss": 4.4179,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6806051378341853e-05,
      "loss": 4.4061,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6797665430831333e-05,
      "loss": 4.3915,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678927948332081e-05,
      "loss": 4.3956,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6780893535810287e-05,
      "loss": 4.4028,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.67725239671035e-05,
      "loss": 4.4117,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676415439839671e-05,
      "loss": 4.3959,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675576845088619e-05,
      "loss": 4.401,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674738250337568e-05,
      "loss": 4.3922,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673899655586516e-05,
      "loss": 4.4006,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673061060835464e-05,
      "loss": 4.3939,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6722257418451576e-05,
      "loss": 4.3903,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6713871470941056e-05,
      "loss": 4.4008,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705485523430536e-05,
      "loss": 4.3993,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6697099575920016e-05,
      "loss": 4.3968,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6688713628409496e-05,
      "loss": 4.3844,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6680327680898976e-05,
      "loss": 4.3869,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6671941733388456e-05,
      "loss": 4.3883,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6663555785877936e-05,
      "loss": 4.3898,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655169838367416e-05,
      "loss": 4.3899,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6646783890856896e-05,
      "loss": 4.3803,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663841432215011e-05,
      "loss": 4.3786,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663002837463959e-05,
      "loss": 4.3902,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66216588059328e-05,
      "loss": 4.3616,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661327285842228e-05,
      "loss": 4.3804,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660488691091176e-05,
      "loss": 4.3718,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6596500963401234e-05,
      "loss": 4.3863,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6588115015890714e-05,
      "loss": 4.3761,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6579729068380194e-05,
      "loss": 4.3826,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6571343120869674e-05,
      "loss": 4.3733,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6562957173359154e-05,
      "loss": 4.3725,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6554571225848634e-05,
      "loss": 4.3827,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6546185278338114e-05,
      "loss": 4.3691,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65377993308276e-05,
      "loss": 4.3825,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652941338331708e-05,
      "loss": 4.3798,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652104381461029e-05,
      "loss": 4.3776,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.651265786709977e-05,
      "loss": 4.371,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650427191958925e-05,
      "loss": 4.3661,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649588597207873e-05,
      "loss": 4.3809,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648750002456821e-05,
      "loss": 4.3633,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647913045586142e-05,
      "loss": 4.3791,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64707445083509e-05,
      "loss": 4.3642,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646235856084038e-05,
      "loss": 4.3697,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645397261332986e-05,
      "loss": 4.3666,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644558666581934e-05,
      "loss": 4.3648,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6437217097112554e-05,
      "loss": 4.3476,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6428831149602034e-05,
      "loss": 4.3639,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6420445202091514e-05,
      "loss": 4.3515,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6412059254580994e-05,
      "loss": 4.3664,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6403673307070474e-05,
      "loss": 4.3797,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639528735955995e-05,
      "loss": 4.3749,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638690141204943e-05,
      "loss": 4.3428,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637853184334264e-05,
      "loss": 4.3678,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637014589583212e-05,
      "loss": 4.3633,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6361759948321596e-05,
      "loss": 4.3635,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6353374000811076e-05,
      "loss": 4.3541,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6344988053300556e-05,
      "loss": 4.3627,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633661848459377e-05,
      "loss": 4.3573,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632824891588699e-05,
      "loss": 4.3558,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631986296837647e-05,
      "loss": 4.3582,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631147702086595e-05,
      "loss": 4.3609,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630309107335542e-05,
      "loss": 4.3564,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.62947051258449e-05,
      "loss": 4.3531,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628631917833438e-05,
      "loss": 4.3596,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627793323082386e-05,
      "loss": 4.3577,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626954728331334e-05,
      "loss": 4.3486,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626117771460655e-05,
      "loss": 4.3637,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625279176709603e-05,
      "loss": 4.3488,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.349167346954346,
      "eval_runtime": 283.9946,
      "eval_samples_per_second": 1343.656,
      "eval_steps_per_second": 41.99,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.167864789759523e+16,
  "trial_name": null,
  "trial_params": null
}