{
  "best_metric": 4.342620849609375,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/det-adj-noun/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": 10.8193,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5516,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0608,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9965,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9474,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9067,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7482,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6243,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5246,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4571,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.4095,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3562,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2988,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.2415,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1819,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.122,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.074,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.0331,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9837,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9423,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.907,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.86,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.838,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7998,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979875363855125e-05,
      "loss": 5.7797,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7431,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7159,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6918,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.97652262273129e-05,
      "loss": 5.674,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.975684027980238e-05,
      "loss": 5.6498,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974845433229186e-05,
      "loss": 5.6279,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.617,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.58,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.573,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714926921053515e-05,
      "loss": 5.5536,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5374,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5226,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4817,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4874,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4534,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4561,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4345,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4366,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.4068,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.3918,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.3965,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3847,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3722,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3542,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.332,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3432,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3239,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3225,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.2913,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954727348605804e-05,
      "loss": 5.2958,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953888753854752e-05,
      "loss": 5.273,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530501591037e-05,
      "loss": 5.2939,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952211564352648e-05,
      "loss": 5.258,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2597,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.2452,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2317,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948858823228813e-05,
      "loss": 5.2339,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948020228477761e-05,
      "loss": 5.2216,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9471816337267086e-05,
      "loss": 5.1945,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9463430389756566e-05,
      "loss": 5.2007,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9455044442246046e-05,
      "loss": 5.2104,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9446658494735526e-05,
      "loss": 5.1875,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9438272547225006e-05,
      "loss": 5.1765,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9429902978518215e-05,
      "loss": 5.159,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421517031007695e-05,
      "loss": 5.1617,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413131083497175e-05,
      "loss": 5.1777,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404745135986655e-05,
      "loss": 5.1614,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.939637556727987e-05,
      "loss": 5.1471,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938798961976935e-05,
      "loss": 5.15,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937960367225883e-05,
      "loss": 5.1461,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937121772474831e-05,
      "loss": 5.1278,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936283177723779e-05,
      "loss": 5.1132,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935444582972727e-05,
      "loss": 5.1184,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934605988221675e-05,
      "loss": 5.0984,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933769031350996e-05,
      "loss": 5.0998,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932930436599944e-05,
      "loss": 5.1044,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932091841848892e-05,
      "loss": 5.0902,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93125324709784e-05,
      "loss": 5.0931,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930416290227161e-05,
      "loss": 5.0813,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0611,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0642,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0609,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0672,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0484,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.0472,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0452,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.0372,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922872213228439e-05,
      "loss": 5.0347,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922033618477387e-05,
      "loss": 5.0237,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921195023726335e-05,
      "loss": 5.0093,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920356428975283e-05,
      "loss": 5.0178,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919517834224231e-05,
      "loss": 5.0058,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918679239473179e-05,
      "loss": 5.0017,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917840644722127e-05,
      "loss": 4.9931,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917003687851449e-05,
      "loss": 4.9923,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916165093100397e-05,
      "loss": 4.9966,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915326498349345e-05,
      "loss": 4.9846,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914487903598292e-05,
      "loss": 4.9739,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91364930884724e-05,
      "loss": 4.9696,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912810714096188e-05,
      "loss": 4.9615,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911972119345136e-05,
      "loss": 4.9481,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911135162474457e-05,
      "loss": 4.9444,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910296567723405e-05,
      "loss": 4.9483,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909457972972353e-05,
      "loss": 4.9445,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908619378221301e-05,
      "loss": 4.9323,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907780783470249e-05,
      "loss": 4.9296,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906943826599571e-05,
      "loss": 4.9301,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906105231848519e-05,
      "loss": 4.9332,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905266637097467e-05,
      "loss": 4.9281,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9044280423464147e-05,
      "loss": 4.9207,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9035910854757356e-05,
      "loss": 4.9198,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.9169,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019155338540045e-05,
      "loss": 4.9064,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010769391029525e-05,
      "loss": 4.8998,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002383443519005e-05,
      "loss": 4.9036,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8993997496008485e-05,
      "loss": 4.9031,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985611548497965e-05,
      "loss": 4.8806,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977225600987445e-05,
      "loss": 4.8813,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968839653476924e-05,
      "loss": 4.8789,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896047008477014e-05,
      "loss": 4.871,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895208413725962e-05,
      "loss": 4.8733,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89436981897491e-05,
      "loss": 4.8763,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893531224223858e-05,
      "loss": 4.8616,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892692629472806e-05,
      "loss": 4.8619,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891854034721754e-05,
      "loss": 4.8506,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891015439970702e-05,
      "loss": 4.8671,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89017684521965e-05,
      "loss": 4.8578,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889338250468598e-05,
      "loss": 4.8493,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888501293597919e-05,
      "loss": 4.8461,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887662698846867e-05,
      "loss": 4.8456,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886824104095815e-05,
      "loss": 4.8457,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885985509344763e-05,
      "loss": 4.8345,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851485524740845e-05,
      "loss": 4.8332,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843099577230325e-05,
      "loss": 4.8331,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834713629719805e-05,
      "loss": 4.8297,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882632768220928e-05,
      "loss": 4.8287,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881794173469876e-05,
      "loss": 4.809,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880955578718824e-05,
      "loss": 4.8195,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880116983967772e-05,
      "loss": 4.8045,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879281664977466e-05,
      "loss": 4.7945,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878443070226414e-05,
      "loss": 4.8131,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877604475475362e-05,
      "loss": 4.8018,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87676588072431e-05,
      "loss": 4.8021,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875927285973258e-05,
      "loss": 4.8048,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875088691222206e-05,
      "loss": 4.7824,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.7586212158203125,
      "eval_runtime": 284.5791,
      "eval_samples_per_second": 1340.896,
      "eval_steps_per_second": 41.904,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874250096471154e-05,
      "loss": 4.7858,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873411501720102e-05,
      "loss": 4.7817,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872574544849423e-05,
      "loss": 4.7995,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871735950098371e-05,
      "loss": 4.7787,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870897355347319e-05,
      "loss": 4.7864,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870058760596267e-05,
      "loss": 4.7659,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869220165845215e-05,
      "loss": 4.7714,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868381571094163e-05,
      "loss": 4.7606,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867544614223484e-05,
      "loss": 4.7651,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866706019472432e-05,
      "loss": 4.7658,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86586742472138e-05,
      "loss": 4.7685,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865028829970329e-05,
      "loss": 4.768,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8641918730996496e-05,
      "loss": 4.7485,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633532783485976e-05,
      "loss": 4.7474,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625146835975456e-05,
      "loss": 4.7442,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616760888464936e-05,
      "loss": 4.7355,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608374940954416e-05,
      "loss": 4.7497,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8599988993443896e-05,
      "loss": 4.7337,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591603045933376e-05,
      "loss": 4.7324,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583217098422856e-05,
      "loss": 4.7522,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574847529716065e-05,
      "loss": 4.7305,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566461582205545e-05,
      "loss": 4.7309,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558075634695025e-05,
      "loss": 4.7267,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549689687184505e-05,
      "loss": 4.733,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541303739673985e-05,
      "loss": 4.7209,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532917792163465e-05,
      "loss": 4.7152,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524531844652945e-05,
      "loss": 4.7121,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8516145897142425e-05,
      "loss": 4.7203,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850779270723937e-05,
      "loss": 4.7009,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849940675972885e-05,
      "loss": 4.7072,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849103719102206e-05,
      "loss": 4.7081,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848265124351154e-05,
      "loss": 4.7029,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847426529600102e-05,
      "loss": 4.7069,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84658793484905e-05,
      "loss": 4.6969,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845750977978371e-05,
      "loss": 4.6981,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449123832273194e-05,
      "loss": 4.6981,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440737884762674e-05,
      "loss": 4.6748,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432351937252154e-05,
      "loss": 4.6861,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423965989741634e-05,
      "loss": 4.6701,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415580042231114e-05,
      "loss": 4.6805,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840719409472059e-05,
      "loss": 4.6811,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839880814721007e-05,
      "loss": 4.684,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839042219969955e-05,
      "loss": 4.6743,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838205263099276e-05,
      "loss": 4.6728,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373666683482236e-05,
      "loss": 4.68,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365280735971716e-05,
      "loss": 4.6741,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8356894788461196e-05,
      "loss": 4.6687,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834852521975441e-05,
      "loss": 4.6621,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834013927224389e-05,
      "loss": 4.6491,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833175332473337e-05,
      "loss": 4.6673,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832336737722285e-05,
      "loss": 4.6602,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831498142971233e-05,
      "loss": 4.6592,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830659548220181e-05,
      "loss": 4.6472,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829820953469129e-05,
      "loss": 4.6551,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82898399659845e-05,
      "loss": 4.639,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828145401847398e-05,
      "loss": 4.6596,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827306807096346e-05,
      "loss": 4.6403,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826468212345294e-05,
      "loss": 4.6485,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825629617594242e-05,
      "loss": 4.6394,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.824792660723563e-05,
      "loss": 4.6318,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823954065972511e-05,
      "loss": 4.6338,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82311547122146e-05,
      "loss": 4.6388,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222768764704077e-05,
      "loss": 4.6132,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214382817193557e-05,
      "loss": 4.6228,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8205996869683037e-05,
      "loss": 4.6354,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197610922172516e-05,
      "loss": 4.6222,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189224974661996e-05,
      "loss": 4.6159,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180855405955206e-05,
      "loss": 4.6117,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172485837248415e-05,
      "loss": 4.611,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164099889737895e-05,
      "loss": 4.6349,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155713942227375e-05,
      "loss": 4.6248,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147327994716854e-05,
      "loss": 4.6156,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138942047206334e-05,
      "loss": 4.6214,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8130556099695814e-05,
      "loss": 4.6215,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8122170152185294e-05,
      "loss": 4.6148,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113784204674774e-05,
      "loss": 4.5982,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105398257164254e-05,
      "loss": 4.6119,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8097012309653734e-05,
      "loss": 4.6027,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088626362143214e-05,
      "loss": 4.5993,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8080240414632694e-05,
      "loss": 4.6105,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80718708459259e-05,
      "loss": 4.601,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806348489841538e-05,
      "loss": 4.6082,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80551153297086e-05,
      "loss": 4.6023,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804672938219807e-05,
      "loss": 4.5807,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803834343468755e-05,
      "loss": 4.596,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802995748717703e-05,
      "loss": 4.5893,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802157153966652e-05,
      "loss": 4.6019,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013185592156e-05,
      "loss": 4.5912,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800479964464548e-05,
      "loss": 4.5936,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799641369713496e-05,
      "loss": 4.5934,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798802774962444e-05,
      "loss": 4.5871,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797965818091765e-05,
      "loss": 4.5908,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797127223340713e-05,
      "loss": 4.5842,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796288628589661e-05,
      "loss": 4.5716,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795451671718982e-05,
      "loss": 4.5831,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79461307696793e-05,
      "loss": 4.5824,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793774482216878e-05,
      "loss": 4.5725,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792935887465826e-05,
      "loss": 4.5788,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792097292714774e-05,
      "loss": 4.5766,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912586979637217e-05,
      "loss": 4.5869,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790421741093043e-05,
      "loss": 4.5712,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789583146341991e-05,
      "loss": 4.5642,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788744551590939e-05,
      "loss": 4.5657,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787905956839887e-05,
      "loss": 4.5637,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787067362088835e-05,
      "loss": 4.5453,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786228767337783e-05,
      "loss": 4.5597,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785391810467104e-05,
      "loss": 4.558,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784553215716052e-05,
      "loss": 4.5568,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783714620965e-05,
      "loss": 4.5518,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782876026213948e-05,
      "loss": 4.5443,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782037431462896e-05,
      "loss": 4.5504,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7811988367118434e-05,
      "loss": 4.5619,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780360241960792e-05,
      "loss": 4.555,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77952164720974e-05,
      "loss": 4.555,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778683052458688e-05,
      "loss": 4.5506,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777844457707636e-05,
      "loss": 4.5615,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777007500836957e-05,
      "loss": 4.5449,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776168906085905e-05,
      "loss": 4.5401,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775331949215226e-05,
      "loss": 4.548,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774493354464174e-05,
      "loss": 4.5506,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773654759713122e-05,
      "loss": 4.5301,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77281616496207e-05,
      "loss": 4.5322,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771977570211018e-05,
      "loss": 4.5385,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771138975459966e-05,
      "loss": 4.5306,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770300380708914e-05,
      "loss": 4.5345,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769461785957862e-05,
      "loss": 4.5406,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686231912068105e-05,
      "loss": 4.5222,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7677845964557585e-05,
      "loss": 4.5323,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669460017047065e-05,
      "loss": 4.5197,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661074069536545e-05,
      "loss": 4.5439,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652704500829754e-05,
      "loss": 4.5354,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644318553319234e-05,
      "loss": 4.5236,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7635932605808714e-05,
      "loss": 4.5264,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7627546658298194e-05,
      "loss": 4.5246,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7619160710787674e-05,
      "loss": 4.5323,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761077476327715e-05,
      "loss": 4.5264,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760238881576663e-05,
      "loss": 4.5218,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759400286825611e-05,
      "loss": 4.5206,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758563329954932e-05,
      "loss": 4.5192,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75772473520388e-05,
      "loss": 4.5218,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756887778333202e-05,
      "loss": 4.5086,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75604918358215e-05,
      "loss": 4.5199,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755210588831097e-05,
      "loss": 4.507,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754371994080045e-05,
      "loss": 4.4974,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753535037209367e-05,
      "loss": 4.5191,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752696442458315e-05,
      "loss": 4.5092,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751857847707262e-05,
      "loss": 4.5107,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75101925295621e-05,
      "loss": 4.5138,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750180658205158e-05,
      "loss": 4.4996,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.477122783660889,
      "eval_runtime": 284.0973,
      "eval_samples_per_second": 1343.17,
      "eval_steps_per_second": 41.975,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749342063454106e-05,
      "loss": 4.5034,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748503468703054e-05,
      "loss": 4.4938,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747664873952003e-05,
      "loss": 4.5195,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746826279200951e-05,
      "loss": 4.4957,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745987684449899e-05,
      "loss": 4.5104,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745149089698847e-05,
      "loss": 4.492,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744310494947795e-05,
      "loss": 4.4931,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743471900196743e-05,
      "loss": 4.4856,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742633305445691e-05,
      "loss": 4.4989,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741794710694639e-05,
      "loss": 4.4967,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740956115943586e-05,
      "loss": 4.4993,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740117521192534e-05,
      "loss": 4.502,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392805643218556e-05,
      "loss": 4.4836,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384419695708036e-05,
      "loss": 4.4876,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737603374819751e-05,
      "loss": 4.482,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736764780068699e-05,
      "loss": 4.4787,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7359278231980205e-05,
      "loss": 4.4878,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7350892284469685e-05,
      "loss": 4.4798,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7342506336959165e-05,
      "loss": 4.4774,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7334120389448645e-05,
      "loss": 4.4946,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7325734441938125e-05,
      "loss": 4.4796,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7317348494427605e-05,
      "loss": 4.4803,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7308962546917085e-05,
      "loss": 4.4793,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7300576599406565e-05,
      "loss": 4.4877,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292207030699774e-05,
      "loss": 4.4737,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7283821083189254e-05,
      "loss": 4.4718,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727545151448246e-05,
      "loss": 4.4699,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726706556697194e-05,
      "loss": 4.4722,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725867961946143e-05,
      "loss": 4.4626,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725029367195091e-05,
      "loss": 4.4676,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724190772444039e-05,
      "loss": 4.4688,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723352177692987e-05,
      "loss": 4.4666,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722513582941935e-05,
      "loss": 4.4743,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721674988190883e-05,
      "loss": 4.4613,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720838031320204e-05,
      "loss": 4.464,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719999436569152e-05,
      "loss": 4.4636,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7191608418181e-05,
      "loss": 4.4486,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718322247067048e-05,
      "loss": 4.4528,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717483652315996e-05,
      "loss": 4.4447,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716645057564944e-05,
      "loss": 4.4516,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715808100694265e-05,
      "loss": 4.4593,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714969505943213e-05,
      "loss": 4.4614,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7141309111921614e-05,
      "loss": 4.4513,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7132923164411094e-05,
      "loss": 4.451,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7124537216900574e-05,
      "loss": 4.4623,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711615126939005e-05,
      "loss": 4.4477,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710778170068326e-05,
      "loss": 4.4524,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709939575317274e-05,
      "loss": 4.4473,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709100980566222e-05,
      "loss": 4.4301,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082623858151696e-05,
      "loss": 4.4518,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074237910641176e-05,
      "loss": 4.4431,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7065851963130656e-05,
      "loss": 4.4462,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7057466015620136e-05,
      "loss": 4.4387,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7049080068109616e-05,
      "loss": 4.4435,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704071049940283e-05,
      "loss": 4.426,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703232455189231e-05,
      "loss": 4.4539,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702393860438179e-05,
      "loss": 4.4293,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701555265687127e-05,
      "loss": 4.4453,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700716670936075e-05,
      "loss": 4.4311,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699878076185023e-05,
      "loss": 4.4313,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699039481433971e-05,
      "loss": 4.4273,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698202524563292e-05,
      "loss": 4.4378,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69736392981224e-05,
      "loss": 4.4115,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696525335061188e-05,
      "loss": 4.4247,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695686740310136e-05,
      "loss": 4.4331,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694848145559084e-05,
      "loss": 4.4248,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694011188688405e-05,
      "loss": 4.4185,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693172593937353e-05,
      "loss": 4.4139,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923339991863017e-05,
      "loss": 4.416,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6914970423156226e-05,
      "loss": 4.4412,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906584475645706e-05,
      "loss": 4.4277,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898198528135186e-05,
      "loss": 4.4208,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6889812580624665e-05,
      "loss": 4.4292,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881426633114145e-05,
      "loss": 4.4301,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873040685603625e-05,
      "loss": 4.425,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864654738093105e-05,
      "loss": 4.4071,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856268790582585e-05,
      "loss": 4.4241,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684788284307206e-05,
      "loss": 4.4192,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683949689556154e-05,
      "loss": 4.4086,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683111094805102e-05,
      "loss": 4.4253,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68227250005405e-05,
      "loss": 4.4154,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6814355431833714e-05,
      "loss": 4.4202,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6805969484323194e-05,
      "loss": 4.4221,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6797583536812674e-05,
      "loss": 4.4018,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6789197589302154e-05,
      "loss": 4.4128,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6780811641791634e-05,
      "loss": 4.4072,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677245845188858e-05,
      "loss": 4.417,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676407250437806e-05,
      "loss": 4.4162,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675568655686753e-05,
      "loss": 4.4116,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674730060935701e-05,
      "loss": 4.414,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673891466184649e-05,
      "loss": 4.41,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673052871433597e-05,
      "loss": 4.4087,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672214276682545e-05,
      "loss": 4.404,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671377319811867e-05,
      "loss": 4.4042,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670538725060815e-05,
      "loss": 4.4092,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669700130309763e-05,
      "loss": 4.4033,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668861535558711e-05,
      "loss": 4.3992,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668022940807659e-05,
      "loss": 4.407,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66718598393698e-05,
      "loss": 4.4057,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666347389185928e-05,
      "loss": 4.418,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665508794434876e-05,
      "loss": 4.3983,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664670199683824e-05,
      "loss": 4.3957,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663831604932772e-05,
      "loss": 4.3978,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6629930101817197e-05,
      "loss": 4.3955,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6621544154306677e-05,
      "loss": 4.3766,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661317458559989e-05,
      "loss": 4.3916,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660478863808937e-05,
      "loss": 4.3944,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659640269057885e-05,
      "loss": 4.3881,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658801674306833e-05,
      "loss": 4.3926,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657964717436154e-05,
      "loss": 4.3813,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657126122685102e-05,
      "loss": 4.3825,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65628752793405e-05,
      "loss": 4.4015,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655448933182998e-05,
      "loss": 4.3884,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654610338431946e-05,
      "loss": 4.397,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653773381561267e-05,
      "loss": 4.3886,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652934786810215e-05,
      "loss": 4.402,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652096192059163e-05,
      "loss": 4.387,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.651257597308111e-05,
      "loss": 4.3835,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504206404374326e-05,
      "loss": 4.3867,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495820456863806e-05,
      "loss": 4.3922,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487434509353286e-05,
      "loss": 4.3747,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479048561842766e-05,
      "loss": 4.3755,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6470662614332246e-05,
      "loss": 4.3776,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6462293045625455e-05,
      "loss": 4.3811,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6453907098114935e-05,
      "loss": 4.3804,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6445521150604415e-05,
      "loss": 4.3877,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6437151581897624e-05,
      "loss": 4.3674,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6428765634387104e-05,
      "loss": 4.3772,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6420379686876584e-05,
      "loss": 4.3676,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6411993739366064e-05,
      "loss": 4.3908,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6403607791855544e-05,
      "loss": 4.3882,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6395221844345024e-05,
      "loss": 4.3741,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6386835896834504e-05,
      "loss": 4.3715,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6378449949323984e-05,
      "loss": 4.3762,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6370064001813464e-05,
      "loss": 4.3806,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6361678054302944e-05,
      "loss": 4.3811,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6353292106792424e-05,
      "loss": 4.3708,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6344906159281903e-05,
      "loss": 4.3737,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633653659057511e-05,
      "loss": 4.3724,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632815064306459e-05,
      "loss": 4.378,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631976469555407e-05,
      "loss": 4.363,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631137874804355e-05,
      "loss": 4.3717,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630299280053303e-05,
      "loss": 4.366,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629462323182625e-05,
      "loss": 4.3498,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628623728431573e-05,
      "loss": 4.3764,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627785133680521e-05,
      "loss": 4.3654,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626946538929469e-05,
      "loss": 4.368,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.62610958205879e-05,
      "loss": 4.372,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625270987307738e-05,
      "loss": 4.3559,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.342620849609375,
      "eval_runtime": 328.6628,
      "eval_samples_per_second": 1161.041,
      "eval_steps_per_second": 36.283,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.40925261274002e+16,
  "trial_name": null,
  "trial_params": null
}