{
  "best_metric": 4.345285892486572,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-sent-neg/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": 10.8194,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5555,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0629,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9948,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9484,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9278,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7608,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6496,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5525,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4704,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.4073,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3344,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.264,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.1874,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988259673485272e-05,
      "loss": 6.145,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98742107873422e-05,
      "loss": 6.0865,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986582483983168e-05,
      "loss": 6.046,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985743889232116e-05,
      "loss": 6.0009,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984905294481064e-05,
      "loss": 5.9637,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984066699730012e-05,
      "loss": 5.9208,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8886,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8482,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.815,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7832,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7639,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7265,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7032,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6747,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.97652262273129e-05,
      "loss": 5.6451,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6353,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6097,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.6019,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5645,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.5545,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714926921053515e-05,
      "loss": 5.543,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5257,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5043,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4693,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4704,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4375,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4397,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4237,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4131,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.3931,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 5.3855,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 5.3868,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 5.3685,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3637,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3361,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 5.3214,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 5.3302,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.3169,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3086,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 5.2772,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 5.2927,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 5.2669,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 5.2899,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 5.2405,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.2513,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.2317,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949699055860238e-05,
      "loss": 5.2205,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 5.2213,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 5.2083,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.1865,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.1881,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.2083,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1823,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1697,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1589,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942154978861516e-05,
      "loss": 5.1448,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941318021990837e-05,
      "loss": 5.1595,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404794272397856e-05,
      "loss": 5.1502,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1441,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1498,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 5.133,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 5.1283,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362880913648985e-05,
      "loss": 5.1067,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354494966138465e-05,
      "loss": 5.1071,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.0975,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 5.0885,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 5.09,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 5.0854,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931259798619332e-05,
      "loss": 5.0892,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930421203868281e-05,
      "loss": 5.0788,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 5.0532,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.928744014366177e-05,
      "loss": 5.0626,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927905419615125e-05,
      "loss": 5.0632,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927066824864073e-05,
      "loss": 5.0658,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92622823011302e-05,
      "loss": 5.0438,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.047,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92455267849129e-05,
      "loss": 5.0392,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923714083740238e-05,
      "loss": 5.0301,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922875488989185e-05,
      "loss": 5.0257,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 5.0213,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921199937367455e-05,
      "loss": 5.0103,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 5.0184,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919522747865351e-05,
      "loss": 4.9993,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 4.9913,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 4.9878,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 4.9927,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.9871,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.9716,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.9728,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.9606,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.9649,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.9415,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111400761155765e-05,
      "loss": 4.9502,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910303119244898e-05,
      "loss": 4.9441,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909464524493846e-05,
      "loss": 4.9394,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908625929742794e-05,
      "loss": 4.945,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907788972872115e-05,
      "loss": 4.9373,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906950378121063e-05,
      "loss": 4.9158,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906111783370011e-05,
      "loss": 4.9256,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905273188618959e-05,
      "loss": 4.9286,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904434593867907e-05,
      "loss": 4.9201,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903595999116855e-05,
      "loss": 4.9189,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902757404365803e-05,
      "loss": 4.9105,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901918809614751e-05,
      "loss": 4.9065,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901081852744072e-05,
      "loss": 4.8963,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90024325799302e-05,
      "loss": 4.903,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994046632419686e-05,
      "loss": 4.8981,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985660684909166e-05,
      "loss": 4.8815,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977274737398646e-05,
      "loss": 4.884,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968888789888125e-05,
      "loss": 4.8807,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960519221181335e-05,
      "loss": 4.8879,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952133273670815e-05,
      "loss": 4.8639,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.8696,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.8574,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926991809942984e-05,
      "loss": 4.8635,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918605862432463e-05,
      "loss": 4.8539,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910219914921943e-05,
      "loss": 4.868,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8901833967411423e-05,
      "loss": 4.8553,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.8486,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.8468,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.8417,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.849,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.8482,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885153466115204e-05,
      "loss": 4.8386,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.8341,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883477914493473e-05,
      "loss": 4.8345,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.8244,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881800724991369e-05,
      "loss": 4.8123,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880962130240316e-05,
      "loss": 4.8194,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880125173369638e-05,
      "loss": 4.8107,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879286578618586e-05,
      "loss": 4.8027,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878447983867534e-05,
      "loss": 4.8034,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877609389116482e-05,
      "loss": 4.802,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.8046,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.8055,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7935,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.759880065917969,
      "eval_runtime": 288.7522,
      "eval_samples_per_second": 1321.517,
      "eval_steps_per_second": 41.298,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7832,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8734180532415946e-05,
      "loss": 4.78,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8725794584905426e-05,
      "loss": 4.7903,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717408637394906e-05,
      "loss": 4.7835,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709022689884386e-05,
      "loss": 4.7866,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700636742373866e-05,
      "loss": 4.7688,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692250794863346e-05,
      "loss": 4.7701,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683864847352826e-05,
      "loss": 4.7637,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675478899842306e-05,
      "loss": 4.7685,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.7684,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7641,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865033743611448e-05,
      "loss": 4.7716,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864196786740769e-05,
      "loss": 4.7404,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863358191989717e-05,
      "loss": 4.75,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862519597238665e-05,
      "loss": 4.7438,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861682640367986e-05,
      "loss": 4.7366,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860844045616934e-05,
      "loss": 4.7479,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860005450865882e-05,
      "loss": 4.7386,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85916685611483e-05,
      "loss": 4.7363,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858328261363778e-05,
      "loss": 4.7561,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.7368,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566527097420475e-05,
      "loss": 4.7311,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558141149909955e-05,
      "loss": 4.7315,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549755202399435e-05,
      "loss": 4.745,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541369254888915e-05,
      "loss": 4.7232,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532983307378395e-05,
      "loss": 4.7195,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524597359867875e-05,
      "loss": 4.7189,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851621141235735e-05,
      "loss": 4.7161,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850782546484683e-05,
      "loss": 4.7149,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849943951733631e-05,
      "loss": 4.7084,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8491069948629524e-05,
      "loss": 4.7274,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8482684001119e-05,
      "loss": 4.7084,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847429805360848e-05,
      "loss": 4.7063,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.7086,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845754253739117e-05,
      "loss": 4.7068,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844915658988065e-05,
      "loss": 4.7016,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844077064237013e-05,
      "loss": 4.6822,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843238469485961e-05,
      "loss": 4.6885,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842399874734909e-05,
      "loss": 4.6758,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841561279983857e-05,
      "loss": 4.692,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840724323113178e-05,
      "loss": 4.6806,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839885728362126e-05,
      "loss": 4.6865,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839047133611074e-05,
      "loss": 4.6758,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838208538860022e-05,
      "loss": 4.6808,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83736994410897e-05,
      "loss": 4.6842,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836531349357918e-05,
      "loss": 4.6768,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835692754606866e-05,
      "loss": 4.6789,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834854159855814e-05,
      "loss": 4.6673,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834015565104763e-05,
      "loss": 4.6592,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833176970353711e-05,
      "loss": 4.6713,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832338375602659e-05,
      "loss": 4.6646,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831499780851606e-05,
      "loss": 4.6707,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830662823980928e-05,
      "loss": 4.6418,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298258671102486e-05,
      "loss": 4.6645,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289872723591966e-05,
      "loss": 4.6422,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281486776081446e-05,
      "loss": 4.6748,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273100828570926e-05,
      "loss": 4.639,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264714881060406e-05,
      "loss": 4.6544,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256328933549886e-05,
      "loss": 4.6413,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247942986039366e-05,
      "loss": 4.6347,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239557038528846e-05,
      "loss": 4.6404,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231171091018326e-05,
      "loss": 4.6396,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222785143507806e-05,
      "loss": 4.6207,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214399195997286e-05,
      "loss": 4.6273,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206029627290495e-05,
      "loss": 4.6431,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819766005858371e-05,
      "loss": 4.6295,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189274111073184e-05,
      "loss": 4.6252,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180888163562664e-05,
      "loss": 4.6219,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172502216052144e-05,
      "loss": 4.611,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164116268541624e-05,
      "loss": 4.6308,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155730321031104e-05,
      "loss": 4.6249,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147344373520584e-05,
      "loss": 4.6255,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138958426010064e-05,
      "loss": 4.6324,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813057247849955e-05,
      "loss": 4.626,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812218653098903e-05,
      "loss": 4.6253,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811381696228224e-05,
      "loss": 4.6037,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810543101477172e-05,
      "loss": 4.6155,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80970450672612e-05,
      "loss": 4.6094,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808865911975068e-05,
      "loss": 4.5979,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808027317224016e-05,
      "loss": 4.6072,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807190360353337e-05,
      "loss": 4.6086,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806353403482658e-05,
      "loss": 4.6179,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805514808731606e-05,
      "loss": 4.6098,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804677851860927e-05,
      "loss": 4.5887,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803839257109875e-05,
      "loss": 4.5977,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803000662358823e-05,
      "loss": 4.6053,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802162067607771e-05,
      "loss": 4.6099,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801323472856719e-05,
      "loss": 4.5981,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800484878105667e-05,
      "loss": 4.6026,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799646283354615e-05,
      "loss": 4.5991,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798807688603563e-05,
      "loss": 4.5874,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797969093852511e-05,
      "loss": 4.5933,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797130499101459e-05,
      "loss": 4.5928,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796291904350407e-05,
      "loss": 4.5811,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795454947479728e-05,
      "loss": 4.5972,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794616352728676e-05,
      "loss": 4.5794,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793777757977624e-05,
      "loss": 4.5767,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792939163226572e-05,
      "loss": 4.5797,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792102206355894e-05,
      "loss": 4.5849,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791263611604842e-05,
      "loss": 4.5841,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79042501685379e-05,
      "loss": 4.5715,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789586422102737e-05,
      "loss": 4.571,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788747827351685e-05,
      "loss": 4.566,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787909232600633e-05,
      "loss": 4.576,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870722757299546e-05,
      "loss": 4.5517,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786233680978902e-05,
      "loss": 4.5601,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78539508622785e-05,
      "loss": 4.5677,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784556491476798e-05,
      "loss": 4.5615,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783717896725746e-05,
      "loss": 4.5632,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782879301974694e-05,
      "loss": 4.5613,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782040707223642e-05,
      "loss": 4.5502,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812037503529635e-05,
      "loss": 4.5603,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803651556019115e-05,
      "loss": 4.5575,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795265608508595e-05,
      "loss": 4.5573,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786896039801804e-05,
      "loss": 4.5603,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778510092291284e-05,
      "loss": 4.561,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777014052358449e-05,
      "loss": 4.5512,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776175457607397e-05,
      "loss": 4.5452,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775336862856345e-05,
      "loss": 4.5537,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774498268105293e-05,
      "loss": 4.5489,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773659673354241e-05,
      "loss": 4.5363,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772821078603189e-05,
      "loss": 4.5463,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771982483852137e-05,
      "loss": 4.5401,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771143889101086e-05,
      "loss": 4.5505,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770305294350034e-05,
      "loss": 4.5322,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769466699598982e-05,
      "loss": 4.5409,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768629742728303e-05,
      "loss": 4.5198,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767791147977251e-05,
      "loss": 4.5419,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766952553226199e-05,
      "loss": 4.5292,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766113958475147e-05,
      "loss": 4.5462,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765275363724095e-05,
      "loss": 4.5375,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764436768973043e-05,
      "loss": 4.5346,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76359817422199e-05,
      "loss": 4.5253,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762759579470938e-05,
      "loss": 4.5271,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761920984719886e-05,
      "loss": 4.5408,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761082389968834e-05,
      "loss": 4.5423,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760243795217783e-05,
      "loss": 4.531,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759406838347104e-05,
      "loss": 4.5288,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758568243596052e-05,
      "loss": 4.5248,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757729648845e-05,
      "loss": 4.5235,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756891054093948e-05,
      "loss": 4.5167,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756052459342896e-05,
      "loss": 4.523,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7552155024722166e-05,
      "loss": 4.5166,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543769077211646e-05,
      "loss": 4.511,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535383129701126e-05,
      "loss": 4.5155,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7526997182190606e-05,
      "loss": 4.5132,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518627613483815e-05,
      "loss": 4.516,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510241665973295e-05,
      "loss": 4.5166,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750185571846278e-05,
      "loss": 4.5123,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.480078220367432,
      "eval_runtime": 293.8973,
      "eval_samples_per_second": 1298.382,
      "eval_steps_per_second": 40.575,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749346977095226e-05,
      "loss": 4.4999,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748508382344174e-05,
      "loss": 4.5012,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747669787593122e-05,
      "loss": 4.5111,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.74683119284207e-05,
      "loss": 4.5087,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745992598091018e-05,
      "loss": 4.5131,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745154003339966e-05,
      "loss": 4.491,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744315408588914e-05,
      "loss": 4.5024,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743476813837862e-05,
      "loss": 4.4928,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426382190868095e-05,
      "loss": 4.503,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7417996243357575e-05,
      "loss": 4.501,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7409610295847055e-05,
      "loss": 4.5012,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401224348336535e-05,
      "loss": 4.5044,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392838400826015e-05,
      "loss": 4.4784,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384452453315495e-05,
      "loss": 4.4877,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376066505804975e-05,
      "loss": 4.4891,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736769693709819e-05,
      "loss": 4.481,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735931098958767e-05,
      "loss": 4.4907,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735092504207715e-05,
      "loss": 4.4839,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734253909456663e-05,
      "loss": 4.4846,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733416952585984e-05,
      "loss": 4.5107,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732578357834932e-05,
      "loss": 4.4839,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73173976308388e-05,
      "loss": 4.4864,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730901168332828e-05,
      "loss": 4.4832,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730062573581776e-05,
      "loss": 4.5003,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729223978830724e-05,
      "loss": 4.4769,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728385384079672e-05,
      "loss": 4.4766,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.72754678932862e-05,
      "loss": 4.4794,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726708194577568e-05,
      "loss": 4.4796,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725869599826516e-05,
      "loss": 4.4702,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725031005075464e-05,
      "loss": 4.4744,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724192410324412e-05,
      "loss": 4.4883,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7233554534537335e-05,
      "loss": 4.4762,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7225184965830544e-05,
      "loss": 4.473,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7216799018320024e-05,
      "loss": 4.4788,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7208413070809504e-05,
      "loss": 4.4739,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7200027123298984e-05,
      "loss": 4.473,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719164117578846e-05,
      "loss": 4.4539,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718325522827794e-05,
      "loss": 4.4694,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717486928076742e-05,
      "loss": 4.4522,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71664833332569e-05,
      "loss": 4.4638,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715811376455011e-05,
      "loss": 4.4591,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714972781703959e-05,
      "loss": 4.4654,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714134186952907e-05,
      "loss": 4.4572,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713297230082228e-05,
      "loss": 4.4586,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712458635331176e-05,
      "loss": 4.4646,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711620040580124e-05,
      "loss": 4.461,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710781445829072e-05,
      "loss": 4.4642,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70994285107802e-05,
      "loss": 4.4502,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709104256326968e-05,
      "loss": 4.4419,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708265661575916e-05,
      "loss": 4.4615,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.707427066824864e-05,
      "loss": 4.4524,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706590109954185e-05,
      "loss": 4.456,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7057531530835066e-05,
      "loss": 4.436,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7049145583324546e-05,
      "loss": 4.4601,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7040759635814026e-05,
      "loss": 4.4338,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7032373688303506e-05,
      "loss": 4.4647,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7023987740792986e-05,
      "loss": 4.4263,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7015618172086195e-05,
      "loss": 4.4501,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007232224575675e-05,
      "loss": 4.4392,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6998846277065155e-05,
      "loss": 4.4375,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990460329554635e-05,
      "loss": 4.4361,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982074382044115e-05,
      "loss": 4.4392,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973704813337324e-05,
      "loss": 4.4219,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696533524463054e-05,
      "loss": 4.4283,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695694929712002e-05,
      "loss": 4.4393,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69485633496095e-05,
      "loss": 4.4355,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694017740209898e-05,
      "loss": 4.4275,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693179145458846e-05,
      "loss": 4.4275,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.692340550707794e-05,
      "loss": 4.422,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691501955956742e-05,
      "loss": 4.4306,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69066336120569e-05,
      "loss": 4.4322,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689824766454638e-05,
      "loss": 4.431,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688987809583959e-05,
      "loss": 4.4411,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688149214832907e-05,
      "loss": 4.4415,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687310620081855e-05,
      "loss": 4.4346,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686472025330803e-05,
      "loss": 4.4154,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.685633430579751e-05,
      "loss": 4.4257,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6847964737090725e-05,
      "loss": 4.4296,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6839578789580204e-05,
      "loss": 4.4112,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6831192842069684e-05,
      "loss": 4.4243,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6822823273362894e-05,
      "loss": 4.4227,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6814437325852373e-05,
      "loss": 4.4338,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6806051378341853e-05,
      "loss": 4.4263,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6797665430831333e-05,
      "loss": 4.4071,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678927948332081e-05,
      "loss": 4.4194,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6780893535810287e-05,
      "loss": 4.4231,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6772507588299766e-05,
      "loss": 4.4282,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6764121640789246e-05,
      "loss": 4.4195,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755735693278726e-05,
      "loss": 4.4247,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6747349745768206e-05,
      "loss": 4.4231,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673896379825769e-05,
      "loss": 4.4074,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.67305942295509e-05,
      "loss": 4.4211,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672220828204038e-05,
      "loss": 4.415,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671383871333359e-05,
      "loss": 4.406,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670545276582307e-05,
      "loss": 4.4267,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669706681831255e-05,
      "loss": 4.4087,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668868087080203e-05,
      "loss": 4.4014,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668029492329151e-05,
      "loss": 4.4094,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667190897578099e-05,
      "loss": 4.415,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666352302827047e-05,
      "loss": 4.4123,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665513708075995e-05,
      "loss": 4.4033,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664676751205316e-05,
      "loss": 4.399,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663838156454265e-05,
      "loss": 4.3981,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662999561703213e-05,
      "loss": 4.4104,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662160966952161e-05,
      "loss": 4.3852,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6613240100814816e-05,
      "loss": 4.3977,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6604854153304296e-05,
      "loss": 4.4034,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6596468205793776e-05,
      "loss": 4.396,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6588082258283256e-05,
      "loss": 4.3973,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6579712689576465e-05,
      "loss": 4.4002,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6571326742065945e-05,
      "loss": 4.3905,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6562940794555425e-05,
      "loss": 4.3972,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6554554847044905e-05,
      "loss": 4.3957,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6546185278338114e-05,
      "loss": 4.3951,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65377993308276e-05,
      "loss": 4.4017,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652941338331708e-05,
      "loss": 4.4006,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652102743580656e-05,
      "loss": 4.3979,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.651264148829604e-05,
      "loss": 4.3855,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650425554078552e-05,
      "loss": 4.3925,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495869593275e-05,
      "loss": 4.3896,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648748364576447e-05,
      "loss": 4.3782,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647909769825395e-05,
      "loss": 4.395,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64707445083509e-05,
      "loss": 4.38,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646235856084038e-05,
      "loss": 4.3947,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645397261332986e-05,
      "loss": 4.3772,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644558666581934e-05,
      "loss": 4.3878,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643720071830882e-05,
      "loss": 4.3712,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64288147707983e-05,
      "loss": 4.3888,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642042882328778e-05,
      "loss": 4.375,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641204287577726e-05,
      "loss": 4.3969,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640365692826674e-05,
      "loss": 4.3902,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639527098075622e-05,
      "loss": 4.3834,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63868850332457e-05,
      "loss": 4.3727,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637849908573518e-05,
      "loss": 4.3781,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637014589583212e-05,
      "loss": 4.3909,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6361759948321596e-05,
      "loss": 4.3954,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6353374000811076e-05,
      "loss": 4.3807,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6344988053300556e-05,
      "loss": 4.3839,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633661848459377e-05,
      "loss": 4.3791,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632823253708325e-05,
      "loss": 4.3786,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631984658957273e-05,
      "loss": 4.3708,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631146064206221e-05,
      "loss": 4.3751,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630307469455169e-05,
      "loss": 4.3764,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629468874704117e-05,
      "loss": 4.3685,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628630279953065e-05,
      "loss": 4.3698,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627791685202013e-05,
      "loss": 4.3756,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626954728331334e-05,
      "loss": 4.3685,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626116133580282e-05,
      "loss": 4.374,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.62527753882923e-05,
      "loss": 4.3681,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.345285892486572,
      "eval_runtime": 293.9389,
      "eval_samples_per_second": 1298.198,
      "eval_steps_per_second": 40.57,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.39826519428338e+16,
  "trial_name": null,
  "trial_params": null
}