{
  "best_metric": 4.027148246765137,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-only/transformer/4/checkpoints/checkpoint-152640",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 152640,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 11.1231,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.824,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.1929,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 5.9737,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.8118,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.7019,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.598,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5315,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.444,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.4024,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3529,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.3115,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 5.2496,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.2092,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.1764,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1286,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1127,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.0799,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0415,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0196,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.004,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 4.9743,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815558291179756e-05,
      "loss": 4.955,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807172343669236e-05,
      "loss": 4.9223,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798786396158716e-05,
      "loss": 4.9166,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 4.8945,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.8715,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8521,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.8266,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 4.8179,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.803,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 4.7952,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 4.7789,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 4.7661,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.7568,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7461,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 4.7349,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.7039,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.7035,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.6783,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.6836,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.6638,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.6661,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6447,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 4.643,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6326,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6258,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6149,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 4.5886,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95892359812181e-05,
      "loss": 4.5976,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958085003370758e-05,
      "loss": 4.5989,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957246408619706e-05,
      "loss": 4.5916,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956407813868654e-05,
      "loss": 4.5715,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.5587,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954732262246923e-05,
      "loss": 4.5602,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953893667495871e-05,
      "loss": 4.5471,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953055072744819e-05,
      "loss": 4.5613,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952216477993767e-05,
      "loss": 4.5225,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.5405,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5297,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5101,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5199,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.5046,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.4946,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.4899,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 4.4998,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 4.4796,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.4763,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4694,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 4.4607,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.4691,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940482703000532e-05,
      "loss": 4.4753,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396441082494794e-05,
      "loss": 4.4483,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388055134984274e-05,
      "loss": 4.4599,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379669187473754e-05,
      "loss": 4.463,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371283239963234e-05,
      "loss": 4.4457,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936291367125644e-05,
      "loss": 4.4344,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935452772374592e-05,
      "loss": 4.4367,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93461417762354e-05,
      "loss": 4.4313,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933775582872488e-05,
      "loss": 4.4215,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932936988121436e-05,
      "loss": 4.4167,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932100031250758e-05,
      "loss": 4.4271,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931261436499706e-05,
      "loss": 4.4261,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930422841748654e-05,
      "loss": 4.4082,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929584246997602e-05,
      "loss": 4.3877,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92874565224655e-05,
      "loss": 4.4003,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 4.4049,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927070100624819e-05,
      "loss": 4.3923,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.926231505873767e-05,
      "loss": 4.3974,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925392911122715e-05,
      "loss": 4.3791,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924554316371663e-05,
      "loss": 4.3776,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923715721620611e-05,
      "loss": 4.3726,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922878764749932e-05,
      "loss": 4.3707,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92204016999888e-05,
      "loss": 4.3775,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921201575247828e-05,
      "loss": 4.3662,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920362980496776e-05,
      "loss": 4.3823,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919524385745724e-05,
      "loss": 4.3652,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918685790994672e-05,
      "loss": 4.3479,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91784719624362e-05,
      "loss": 4.3519,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 4.3586,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916173282502262e-05,
      "loss": 4.3489,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91533468775121e-05,
      "loss": 4.3467,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914496093000158e-05,
      "loss": 4.3452,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913657498249106e-05,
      "loss": 4.3354,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912818903498054e-05,
      "loss": 4.3377,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911980308747002e-05,
      "loss": 4.316,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91114171399595e-05,
      "loss": 4.3228,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910304757125272e-05,
      "loss": 4.327,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90946616237422e-05,
      "loss": 4.3217,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908627567623168e-05,
      "loss": 4.315,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907788972872115e-05,
      "loss": 4.3248,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9069520160014366e-05,
      "loss": 4.2985,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9061134212503846e-05,
      "loss": 4.3105,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9052748264993326e-05,
      "loss": 4.3034,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90443623174828e-05,
      "loss": 4.3096,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9035992748776015e-05,
      "loss": 4.316,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027606801265495e-05,
      "loss": 4.3044,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019220853754975e-05,
      "loss": 4.298,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010834906244455e-05,
      "loss": 4.292,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900246533753767e-05,
      "loss": 4.2983,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899407939002715e-05,
      "loss": 4.2913,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985693442516624e-05,
      "loss": 4.2913,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977307495006104e-05,
      "loss": 4.2818,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896893792629932e-05,
      "loss": 4.2804,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89605519787888e-05,
      "loss": 4.2816,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895216603127827e-05,
      "loss": 4.2698,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.894378008376775e-05,
      "loss": 4.2752,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893539413625723e-05,
      "loss": 4.2736,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892702456755045e-05,
      "loss": 4.2692,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891863862003992e-05,
      "loss": 4.2598,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891025267252941e-05,
      "loss": 4.2771,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890186672501889e-05,
      "loss": 4.2773,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934971563121e-05,
      "loss": 4.2651,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888511120880158e-05,
      "loss": 4.2555,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887672526129106e-05,
      "loss": 4.2629,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886833931378054e-05,
      "loss": 4.2624,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859969745073747e-05,
      "loss": 4.2621,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851583797563226e-05,
      "loss": 4.257,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843197850052706e-05,
      "loss": 4.2564,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834811902542186e-05,
      "loss": 4.2518,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8826442333835395e-05,
      "loss": 4.2488,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8818056386324875e-05,
      "loss": 4.2436,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880967043881436e-05,
      "loss": 4.2466,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880128449130384e-05,
      "loss": 4.2379,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879291492259705e-05,
      "loss": 4.2334,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878452897508653e-05,
      "loss": 4.2444,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877614302757601e-05,
      "loss": 4.2327,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876775708006549e-05,
      "loss": 4.2407,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875937113255497e-05,
      "loss": 4.2297,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875100156384818e-05,
      "loss": 4.2235,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.195745468139648,
      "eval_runtime": 307.7836,
      "eval_samples_per_second": 1239.803,
      "eval_steps_per_second": 38.745,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874261561633766e-05,
      "loss": 4.2157,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873422966882714e-05,
      "loss": 4.2142,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872584372131662e-05,
      "loss": 4.2256,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871747415260983e-05,
      "loss": 4.2177,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709088205099316e-05,
      "loss": 4.2193,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700702257588796e-05,
      "loss": 4.2085,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692316310078276e-05,
      "loss": 4.2066,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683946741371485e-05,
      "loss": 4.1983,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675560793860965e-05,
      "loss": 4.2119,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667174846350445e-05,
      "loss": 4.2045,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658788898839925e-05,
      "loss": 4.2167,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8650419330133134e-05,
      "loss": 4.1982,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864204976142634e-05,
      "loss": 4.1934,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863366381391582e-05,
      "loss": 4.188,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86252778664053e-05,
      "loss": 4.1854,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861689191889478e-05,
      "loss": 4.1827,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860850597138426e-05,
      "loss": 4.1927,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860012002387375e-05,
      "loss": 4.1782,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859173407636323e-05,
      "loss": 4.1782,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858334812885271e-05,
      "loss": 4.1981,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857497856014592e-05,
      "loss": 4.1831,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85665926126354e-05,
      "loss": 4.1831,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855820666512488e-05,
      "loss": 4.173,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854982071761436e-05,
      "loss": 4.1847,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854145114890757e-05,
      "loss": 4.1698,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853306520139705e-05,
      "loss": 4.1711,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852467925388653e-05,
      "loss": 4.1689,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851629330637601e-05,
      "loss": 4.1623,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8507923737669216e-05,
      "loss": 4.1563,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84995377901587e-05,
      "loss": 4.1566,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849115184264818e-05,
      "loss": 4.1653,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848276589513766e-05,
      "loss": 4.1647,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847439632643087e-05,
      "loss": 4.159,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846601037892035e-05,
      "loss": 4.1649,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845762443140983e-05,
      "loss": 4.1598,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844923848389931e-05,
      "loss": 4.158,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844086891519252e-05,
      "loss": 4.1461,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432482967682e-05,
      "loss": 4.1469,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842409702017148e-05,
      "loss": 4.1362,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841571107266096e-05,
      "loss": 4.1488,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840734150395417e-05,
      "loss": 4.1345,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839895555644366e-05,
      "loss": 4.1483,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839056960893314e-05,
      "loss": 4.1345,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382200040226346e-05,
      "loss": 4.138,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373814092715826e-05,
      "loss": 4.1386,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365428145205306e-05,
      "loss": 4.1348,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8357042197694786e-05,
      "loss": 4.1334,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8348672628987995e-05,
      "loss": 4.1186,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8340286681477475e-05,
      "loss": 4.1242,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8331900733966955e-05,
      "loss": 4.1353,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8323514786456435e-05,
      "loss": 4.1308,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8315145217749644e-05,
      "loss": 4.1231,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8306759270239124e-05,
      "loss": 4.1126,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829837332272861e-05,
      "loss": 4.1226,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828998737521809e-05,
      "loss": 4.1088,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82816178065113e-05,
      "loss": 4.1323,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827323185900078e-05,
      "loss": 4.1026,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826484591149026e-05,
      "loss": 4.1241,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825645996397974e-05,
      "loss": 4.1165,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.824809039527295e-05,
      "loss": 4.1038,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823970444776243e-05,
      "loss": 4.1143,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823131850025191e-05,
      "loss": 4.1111,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822293255274139e-05,
      "loss": 4.0983,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82145629840346e-05,
      "loss": 4.0998,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820617703652408e-05,
      "loss": 4.114,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197791089013564e-05,
      "loss": 4.1013,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189405141503044e-05,
      "loss": 4.0995,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818103557279625e-05,
      "loss": 4.096,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817264962528573e-05,
      "loss": 4.0922,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816426367777521e-05,
      "loss": 4.1017,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815587773026469e-05,
      "loss": 4.1133,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81475081615579e-05,
      "loss": 4.0929,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813912221404738e-05,
      "loss": 4.1094,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813073626653686e-05,
      "loss": 4.1092,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812235031902634e-05,
      "loss": 4.0954,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811398075031955e-05,
      "loss": 4.0906,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810559480280903e-05,
      "loss": 4.0995,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809720885529852e-05,
      "loss": 4.0949,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088822907788e-05,
      "loss": 4.0859,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808045333908121e-05,
      "loss": 4.0888,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807206739157069e-05,
      "loss": 4.0938,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806368144406017e-05,
      "loss": 4.101,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805529549654965e-05,
      "loss": 4.0883,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046925927842856e-05,
      "loss": 4.0733,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8038539980332336e-05,
      "loss": 4.0826,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030154032821816e-05,
      "loss": 4.0863,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021768085311296e-05,
      "loss": 4.0839,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013398516604505e-05,
      "loss": 4.0899,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8005012569093985e-05,
      "loss": 4.0756,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799662662158347e-05,
      "loss": 4.0714,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7988240674072945e-05,
      "loss": 4.0713,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797987110536616e-05,
      "loss": 4.0738,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797148515785564e-05,
      "loss": 4.0777,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796309921034512e-05,
      "loss": 4.0739,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7954713262834594e-05,
      "loss": 4.0909,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794634369412781e-05,
      "loss": 4.0787,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793795774661729e-05,
      "loss": 4.06,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792957179910677e-05,
      "loss": 4.0717,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792118585159624e-05,
      "loss": 4.0721,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791279990408572e-05,
      "loss": 4.0722,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790441395657521e-05,
      "loss": 4.0694,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789604438786842e-05,
      "loss": 4.0671,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78876584403579e-05,
      "loss": 4.0618,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787927249284738e-05,
      "loss": 4.0715,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787088654533686e-05,
      "loss": 4.0471,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786250059782634e-05,
      "loss": 4.0601,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785413102911955e-05,
      "loss": 4.0593,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784574508160903e-05,
      "loss": 4.0614,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783735913409851e-05,
      "loss": 4.0509,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782897318658799e-05,
      "loss": 4.0616,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782058723907747e-05,
      "loss": 4.0475,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781220129156695e-05,
      "loss": 4.0539,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780383172286016e-05,
      "loss": 4.0513,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779544577534964e-05,
      "loss": 4.052,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778705982783912e-05,
      "loss": 4.0673,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77786738803286e-05,
      "loss": 4.058,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777030431162181e-05,
      "loss": 4.051,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776191836411129e-05,
      "loss": 4.0467,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775353241660077e-05,
      "loss": 4.0513,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774514646909025e-05,
      "loss": 4.0494,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773676052157973e-05,
      "loss": 4.0472,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772837457406921e-05,
      "loss": 4.0433,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771998862655869e-05,
      "loss": 4.0426,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771160267904817e-05,
      "loss": 4.0466,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770323311034138e-05,
      "loss": 4.0342,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769484716283086e-05,
      "loss": 4.0414,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768646121532034e-05,
      "loss": 4.0402,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767807526780983e-05,
      "loss": 4.0405,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669705699103036e-05,
      "loss": 4.0289,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661319751592516e-05,
      "loss": 4.0511,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652933804081996e-05,
      "loss": 4.0546,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644547856571476e-05,
      "loss": 4.0387,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636194666668415e-05,
      "loss": 4.0316,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7627808719157894e-05,
      "loss": 4.039,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7619422771647374e-05,
      "loss": 4.0425,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7611036824136854e-05,
      "loss": 4.042,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602650876626334e-05,
      "loss": 4.0368,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759428130791955e-05,
      "loss": 4.0373,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758589536040903e-05,
      "loss": 4.0337,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757750941289851e-05,
      "loss": 4.0385,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756912346538799e-05,
      "loss": 4.0265,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756073751787747e-05,
      "loss": 4.0361,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755235157036695e-05,
      "loss": 4.0328,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754398200166016e-05,
      "loss": 4.0204,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753559605414964e-05,
      "loss": 4.034,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752721010663912e-05,
      "loss": 4.0251,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75188241591286e-05,
      "loss": 4.0333,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751043821161808e-05,
      "loss": 4.0217,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750205226410755e-05,
      "loss": 4.019,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.027148246765137,
      "eval_runtime": 308.4603,
      "eval_samples_per_second": 1237.083,
      "eval_steps_per_second": 38.66,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.0644055367242752e+17,
  "trial_name": null,
  "trial_params": null
}