{
  "best_metric": 4.347803592681885,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/rel-cl/lstm/2/checkpoints/checkpoint-228957",
  "epoch": 0.025000278439663435,
  "eval_steps": 10,
  "global_step": 228957,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8203,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5687,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0534,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9801,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9528,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8928,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994970069374061e-05,
      "loss": 6.7305,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994131474623009e-05,
      "loss": 6.6212,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.531,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.4382,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.3957,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3302,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.261,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.1908,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1396,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0984,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0549,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.997,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9588,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9206,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8828,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8553,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8258,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7928,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979875363855125e-05,
      "loss": 5.7648,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979036769104073e-05,
      "loss": 5.7474,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7141,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.6924,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6582,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6537,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6206,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 5.604,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.5936,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5759,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5452,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 5.5326,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969820416244367e-05,
      "loss": 5.5245,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 5.5017,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 5.4895,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 5.4689,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 5.4454,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.4333,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 5.4194,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.4124,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 5.4004,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 5.3951,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9614377444945934e-05,
      "loss": 5.3712,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9605991497435414e-05,
      "loss": 5.3707,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9597605549924894e-05,
      "loss": 5.3748,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 5.347,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 5.3368,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.3234,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3324,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.295,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 5.3127,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953893667495871e-05,
      "loss": 5.2827,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953055072744819e-05,
      "loss": 5.2793,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952216477993767e-05,
      "loss": 5.2744,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 5.2637,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 5.2516,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 5.2388,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 5.2241,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 5.2312,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.22,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.2093,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 5.2011,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944670763114672e-05,
      "loss": 5.1917,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94383216836362e-05,
      "loss": 5.1736,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942993573612568e-05,
      "loss": 5.1913,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 5.1683,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941318021990837e-05,
      "loss": 5.1615,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404794272397856e-05,
      "loss": 5.1706,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1465,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1592,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 5.1496,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 5.139,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 5.137,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 5.1119,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93461417762354e-05,
      "loss": 5.108,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933775582872488e-05,
      "loss": 5.1074,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932936988121436e-05,
      "loss": 5.1066,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932098393370384e-05,
      "loss": 5.0962,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931259798619332e-05,
      "loss": 5.0821,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930421203868281e-05,
      "loss": 5.0683,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 5.0755,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.928744014366177e-05,
      "loss": 5.0744,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927905419615125e-05,
      "loss": 5.0703,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927066824864073e-05,
      "loss": 5.0674,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92622823011302e-05,
      "loss": 5.0616,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0502,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92455267849129e-05,
      "loss": 5.0315,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923714083740238e-05,
      "loss": 5.0285,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922875488989185e-05,
      "loss": 5.0339,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922038532118507e-05,
      "loss": 5.0307,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921199937367455e-05,
      "loss": 5.0084,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 5.0133,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919522747865351e-05,
      "loss": 5.0087,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 5.0097,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91784719624362e-05,
      "loss": 5.0005,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 5.0025,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.9942,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.9846,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.9828,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136542224883596e-05,
      "loss": 4.9844,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128156277373076e-05,
      "loss": 4.9684,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119770329862556e-05,
      "loss": 4.9716,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111400761155765e-05,
      "loss": 4.9488,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9103014813645245e-05,
      "loss": 4.9452,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909462886613473e-05,
      "loss": 4.9431,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908624291862421e-05,
      "loss": 4.9489,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.941,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9176,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9233,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905271550738586e-05,
      "loss": 4.9261,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904432955987534e-05,
      "loss": 4.9335,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903594361236482e-05,
      "loss": 4.9202,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.9187,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.915,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.9177,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.9009,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.9012,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8918,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8958,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8777,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896048646357387e-05,
      "loss": 4.8901,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895210051606335e-05,
      "loss": 4.881,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.894371456855283e-05,
      "loss": 4.8882,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893532862104231e-05,
      "loss": 4.8592,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8706,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8804,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.853,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.8728,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889341526229344e-05,
      "loss": 4.8565,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888504569358665e-05,
      "loss": 4.8597,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8876659746076134e-05,
      "loss": 4.8493,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8868273798565614e-05,
      "loss": 4.8443,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859887851055094e-05,
      "loss": 4.8456,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.844,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.845,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8369,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8388,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.8373,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.8183,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.8198,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.8097,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8113,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.8115,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.8039,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.8016,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.8008,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.762693881988525,
      "eval_runtime": 547.5599,
      "eval_samples_per_second": 696.894,
      "eval_steps_per_second": 21.778,
      "step": 76319
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.8123,
      "step": 76800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.8018,
      "step": 77312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.7914,
      "step": 77824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7917,
      "step": 78336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8709006311080657e-05,
      "loss": 4.7865,
      "step": 78848
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8700620363570137e-05,
      "loss": 4.7638,
      "step": 79360
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.778,
      "step": 79872
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8683848468549096e-05,
      "loss": 4.7801,
      "step": 80384
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8675462521038576e-05,
      "loss": 4.7616,
      "step": 80896
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8667092952331785e-05,
      "loss": 4.7768,
      "step": 81408
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.865870700482127e-05,
      "loss": 4.7818,
      "step": 81920
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.865032105731075e-05,
      "loss": 4.7651,
      "step": 82432
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8641935109800225e-05,
      "loss": 4.7583,
      "step": 82944
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8633549162289705e-05,
      "loss": 4.7605,
      "step": 83456
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8625163214779185e-05,
      "loss": 4.7607,
      "step": 83968
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8616777267268665e-05,
      "loss": 4.7545,
      "step": 84480
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8608391319758145e-05,
      "loss": 4.75,
      "step": 84992
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8600005372247625e-05,
      "loss": 4.7488,
      "step": 85504
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8591619424737105e-05,
      "loss": 4.7441,
      "step": 86016
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8583233477226585e-05,
      "loss": 4.7474,
      "step": 86528
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8574847529716065e-05,
      "loss": 4.7475,
      "step": 87040
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8566461582205545e-05,
      "loss": 4.747,
      "step": 87552
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8558092013498754e-05,
      "loss": 4.7408,
      "step": 88064
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.854970606598824e-05,
      "loss": 4.7411,
      "step": 88576
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.854132011847772e-05,
      "loss": 4.7342,
      "step": 89088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.85329341709672e-05,
      "loss": 4.7323,
      "step": 89600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.852456460226041e-05,
      "loss": 4.7334,
      "step": 90112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.851617865474989e-05,
      "loss": 4.716,
      "step": 90624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.850779270723937e-05,
      "loss": 4.7295,
      "step": 91136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.849942313853258e-05,
      "loss": 4.7101,
      "step": 91648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.849103719102206e-05,
      "loss": 4.7153,
      "step": 92160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.848265124351154e-05,
      "loss": 4.7248,
      "step": 92672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.847426529600102e-05,
      "loss": 4.7218,
      "step": 93184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.84658793484905e-05,
      "loss": 4.708,
      "step": 93696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.845749340097998e-05,
      "loss": 4.7001,
      "step": 94208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.844910745346946e-05,
      "loss": 4.718,
      "step": 94720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.844072150595894e-05,
      "loss": 4.7053,
      "step": 95232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.843233555844842e-05,
      "loss": 4.7041,
      "step": 95744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.84239496109379e-05,
      "loss": 4.7015,
      "step": 96256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8415580042231114e-05,
      "loss": 4.6859,
      "step": 96768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.840719409472059e-05,
      "loss": 4.683,
      "step": 97280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.839880814721007e-05,
      "loss": 4.6846,
      "step": 97792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.839042219969955e-05,
      "loss": 4.6891,
      "step": 98304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.838203625218903e-05,
      "loss": 4.6821,
      "step": 98816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.837365030467851e-05,
      "loss": 4.6854,
      "step": 99328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.836526435716799e-05,
      "loss": 4.6781,
      "step": 99840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8356894788461196e-05,
      "loss": 4.6801,
      "step": 100352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8348508840950676e-05,
      "loss": 4.6916,
      "step": 100864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8340122893440156e-05,
      "loss": 4.6734,
      "step": 101376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.833173694592964e-05,
      "loss": 4.6745,
      "step": 101888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.832335099841912e-05,
      "loss": 4.6637,
      "step": 102400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.831498142971233e-05,
      "loss": 4.6853,
      "step": 102912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.830659548220181e-05,
      "loss": 4.6546,
      "step": 103424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.829820953469129e-05,
      "loss": 4.6767,
      "step": 103936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.828982358718077e-05,
      "loss": 4.6536,
      "step": 104448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.828143763967025e-05,
      "loss": 4.662,
      "step": 104960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.827305169215973e-05,
      "loss": 4.6628,
      "step": 105472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.826468212345294e-05,
      "loss": 4.6577,
      "step": 105984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.825629617594242e-05,
      "loss": 4.6493,
      "step": 106496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.82479102284319e-05,
      "loss": 4.6494,
      "step": 107008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.823952428092138e-05,
      "loss": 4.6365,
      "step": 107520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.823113833341086e-05,
      "loss": 4.6537,
      "step": 108032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.822275238590034e-05,
      "loss": 4.6452,
      "step": 108544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8214382817193557e-05,
      "loss": 4.6394,
      "step": 109056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8205996869683037e-05,
      "loss": 4.6377,
      "step": 109568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8197610922172516e-05,
      "loss": 4.6367,
      "step": 110080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8189224974661996e-05,
      "loss": 4.6197,
      "step": 110592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8180839027151476e-05,
      "loss": 4.6439,
      "step": 111104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.817245307964095e-05,
      "loss": 4.6283,
      "step": 111616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8164083510934165e-05,
      "loss": 4.6296,
      "step": 112128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8155697563423645e-05,
      "loss": 4.6379,
      "step": 112640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8147311615913125e-05,
      "loss": 4.6233,
      "step": 113152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8138942047206334e-05,
      "loss": 4.6383,
      "step": 113664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8130556099695814e-05,
      "loss": 4.6384,
      "step": 114176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8122170152185294e-05,
      "loss": 4.6325,
      "step": 114688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8113784204674774e-05,
      "loss": 4.6309,
      "step": 115200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8105398257164254e-05,
      "loss": 4.6124,
      "step": 115712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8097012309653734e-05,
      "loss": 4.6112,
      "step": 116224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8088626362143214e-05,
      "loss": 4.6203,
      "step": 116736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8080240414632694e-05,
      "loss": 4.6231,
      "step": 117248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8071854467122174e-05,
      "loss": 4.6138,
      "step": 117760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8063468519611654e-05,
      "loss": 4.6085,
      "step": 118272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.805509895090486e-05,
      "loss": 4.5931,
      "step": 118784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.804671300339434e-05,
      "loss": 4.6024,
      "step": 119296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.803832705588382e-05,
      "loss": 4.6073,
      "step": 119808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.80299411083733e-05,
      "loss": 4.6061,
      "step": 120320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.802155516086278e-05,
      "loss": 4.6146,
      "step": 120832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.8013185592156e-05,
      "loss": 4.6105,
      "step": 121344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.800479964464548e-05,
      "loss": 4.5996,
      "step": 121856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.799643007593869e-05,
      "loss": 4.5937,
      "step": 122368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.798804412842817e-05,
      "loss": 4.5871,
      "step": 122880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.797965818091765e-05,
      "loss": 4.5926,
      "step": 123392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.797127223340713e-05,
      "loss": 4.5957,
      "step": 123904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.796288628589661e-05,
      "loss": 4.5838,
      "step": 124416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.795450033838609e-05,
      "loss": 4.5833,
      "step": 124928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.794611439087557e-05,
      "loss": 4.5923,
      "step": 125440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.793772844336505e-05,
      "loss": 4.5951,
      "step": 125952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.792934249585453e-05,
      "loss": 4.5842,
      "step": 126464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.792097292714774e-05,
      "loss": 4.59,
      "step": 126976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7912586979637217e-05,
      "loss": 4.5911,
      "step": 127488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.790421741093043e-05,
      "loss": 4.5834,
      "step": 128000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.789583146341991e-05,
      "loss": 4.579,
      "step": 128512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.788744551590939e-05,
      "loss": 4.5886,
      "step": 129024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.787905956839887e-05,
      "loss": 4.5753,
      "step": 129536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.787067362088835e-05,
      "loss": 4.5868,
      "step": 130048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.786228767337783e-05,
      "loss": 4.5671,
      "step": 130560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.785390172586731e-05,
      "loss": 4.5559,
      "step": 131072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7845515778356785e-05,
      "loss": 4.5642,
      "step": 131584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7837129830846265e-05,
      "loss": 4.5728,
      "step": 132096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7828743883335745e-05,
      "loss": 4.5648,
      "step": 132608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7820357935825225e-05,
      "loss": 4.5506,
      "step": 133120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7811971988314705e-05,
      "loss": 4.5516,
      "step": 133632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.780360241960792e-05,
      "loss": 4.5602,
      "step": 134144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.779523285090114e-05,
      "loss": 4.5708,
      "step": 134656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.778684690339061e-05,
      "loss": 4.5587,
      "step": 135168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.777846095588009e-05,
      "loss": 4.5675,
      "step": 135680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.777007500836957e-05,
      "loss": 4.5612,
      "step": 136192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.776168906085905e-05,
      "loss": 4.5589,
      "step": 136704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7753335870955995e-05,
      "loss": 4.5543,
      "step": 137216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7744949923445475e-05,
      "loss": 4.5512,
      "step": 137728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7736563975934955e-05,
      "loss": 4.5459,
      "step": 138240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7728178028424435e-05,
      "loss": 4.5565,
      "step": 138752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.771979208091391e-05,
      "loss": 4.5422,
      "step": 139264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.771140613340339e-05,
      "loss": 4.5447,
      "step": 139776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7703020185892875e-05,
      "loss": 4.5482,
      "step": 140288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7694650617186084e-05,
      "loss": 4.5548,
      "step": 140800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7686264669675564e-05,
      "loss": 4.5325,
      "step": 141312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7677878722165044e-05,
      "loss": 4.5379,
      "step": 141824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7669492774654524e-05,
      "loss": 4.5592,
      "step": 142336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7661106827144004e-05,
      "loss": 4.5323,
      "step": 142848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.765273725843721e-05,
      "loss": 4.5536,
      "step": 143360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.764435131092669e-05,
      "loss": 4.5361,
      "step": 143872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.763596536341617e-05,
      "loss": 4.5473,
      "step": 144384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.762757941590565e-05,
      "loss": 4.5339,
      "step": 144896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.761919346839513e-05,
      "loss": 4.5286,
      "step": 145408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.761080752088461e-05,
      "loss": 4.5373,
      "step": 145920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.760242157337409e-05,
      "loss": 4.5353,
      "step": 146432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.759403562586357e-05,
      "loss": 4.5423,
      "step": 146944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.758564967835306e-05,
      "loss": 4.5374,
      "step": 147456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.757728010964627e-05,
      "loss": 4.5371,
      "step": 147968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.756889416213575e-05,
      "loss": 4.5334,
      "step": 148480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.756050821462523e-05,
      "loss": 4.5196,
      "step": 148992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.755212226711471e-05,
      "loss": 4.5304,
      "step": 149504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.754373631960419e-05,
      "loss": 4.5158,
      "step": 150016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.75353667508974e-05,
      "loss": 4.5221,
      "step": 150528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.752698080338688e-05,
      "loss": 4.5251,
      "step": 151040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.751859485587636e-05,
      "loss": 4.5122,
      "step": 151552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.751020890836584e-05,
      "loss": 4.5178,
      "step": 152064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.7501839339659046e-05,
      "loss": 4.5151,
      "step": 152576
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.481568813323975,
      "eval_runtime": 534.6878,
      "eval_samples_per_second": 713.671,
      "eval_steps_per_second": 22.303,
      "step": 152638
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7493453392148526e-05,
      "loss": 4.5286,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748506744463801e-05,
      "loss": 4.524,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747668149712749e-05,
      "loss": 4.5131,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746829554961697e-05,
      "loss": 4.5121,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7459909602106446e-05,
      "loss": 4.5118,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7451523654595926e-05,
      "loss": 4.4943,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7443137707085406e-05,
      "loss": 4.5092,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7434751759574886e-05,
      "loss": 4.5075,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426365812064366e-05,
      "loss": 4.4931,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7417979864553846e-05,
      "loss": 4.5115,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7409593917043326e-05,
      "loss": 4.5183,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401207969532806e-05,
      "loss": 4.5013,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392838400826015e-05,
      "loss": 4.4945,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384452453315495e-05,
      "loss": 4.4981,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376066505804975e-05,
      "loss": 4.5022,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736768055829446e-05,
      "loss": 4.4961,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735929461078394e-05,
      "loss": 4.4938,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735090866327342e-05,
      "loss": 4.4925,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73425227157629e-05,
      "loss": 4.4939,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733413676825238e-05,
      "loss": 4.4959,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732576719954559e-05,
      "loss": 4.496,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.731738125203507e-05,
      "loss": 4.4993,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730899530452455e-05,
      "loss": 4.4975,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730060935701403e-05,
      "loss": 4.4898,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729222340950351e-05,
      "loss": 4.4924,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728385384079672e-05,
      "loss": 4.4882,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.72754678932862e-05,
      "loss": 4.4922,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726708194577568e-05,
      "loss": 4.4761,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725869599826516e-05,
      "loss": 4.4876,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250326429558375e-05,
      "loss": 4.4727,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7241940482047855e-05,
      "loss": 4.479,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7233554534537335e-05,
      "loss": 4.4911,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722516858702681e-05,
      "loss": 4.4863,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721678263951629e-05,
      "loss": 4.4757,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7208413070809504e-05,
      "loss": 4.4699,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7200027123298984e-05,
      "loss": 4.4908,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719164117578846e-05,
      "loss": 4.472,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718325522827794e-05,
      "loss": 4.4798,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717486928076742e-05,
      "loss": 4.4762,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71664833332569e-05,
      "loss": 4.4604,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7158097385746384e-05,
      "loss": 4.4584,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7149711438235864e-05,
      "loss": 4.4602,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714134186952907e-05,
      "loss": 4.4739,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713295592201855e-05,
      "loss": 4.4638,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712456997450803e-05,
      "loss": 4.4654,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711620040580124e-05,
      "loss": 4.4579,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710781445829072e-05,
      "loss": 4.4622,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70994285107802e-05,
      "loss": 4.4778,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709104256326968e-05,
      "loss": 4.4569,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708265661575916e-05,
      "loss": 4.4575,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.707427066824864e-05,
      "loss": 4.4525,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706590109954185e-05,
      "loss": 4.4724,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705751515203134e-05,
      "loss": 4.4442,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704912920452082e-05,
      "loss": 4.4654,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70407432570103e-05,
      "loss": 4.4453,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703235730949978e-05,
      "loss": 4.4569,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702397136198926e-05,
      "loss": 4.4589,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701558541447874e-05,
      "loss": 4.4463,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007215845771946e-05,
      "loss": 4.453,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6998829898261426e-05,
      "loss": 4.4456,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990443950750906e-05,
      "loss": 4.4299,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982058003240386e-05,
      "loss": 4.4503,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973688434533595e-05,
      "loss": 4.447,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965302487023075e-05,
      "loss": 4.4395,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956916539512555e-05,
      "loss": 4.44,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948530592002035e-05,
      "loss": 4.4398,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940144644491515e-05,
      "loss": 4.4225,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931758696980995e-05,
      "loss": 4.4486,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923372749470475e-05,
      "loss": 4.4301,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691500318076369e-05,
      "loss": 4.4349,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690661723325317e-05,
      "loss": 4.4451,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898231285742644e-05,
      "loss": 4.4332,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6889845338232124e-05,
      "loss": 4.4422,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881459390721604e-05,
      "loss": 4.4515,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687308982201481e-05,
      "loss": 4.4383,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686470387450429e-05,
      "loss": 4.4457,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.685631792699377e-05,
      "loss": 4.426,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684793197948325e-05,
      "loss": 4.4218,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683956241077647e-05,
      "loss": 4.4309,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683117646326595e-05,
      "loss": 4.4415,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682279051575543e-05,
      "loss": 4.4291,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681440456824491e-05,
      "loss": 4.4234,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680601862073439e-05,
      "loss": 4.4144,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679763267322387e-05,
      "loss": 4.4172,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678924672571335e-05,
      "loss": 4.4204,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678086077820283e-05,
      "loss": 4.4256,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677247483069231e-05,
      "loss": 4.4337,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676408888318179e-05,
      "loss": 4.4311,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675570293567127e-05,
      "loss": 4.4213,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674733336696448e-05,
      "loss": 4.4167,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673894741945396e-05,
      "loss": 4.4112,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673056147194344e-05,
      "loss": 4.4156,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6722175524432924e-05,
      "loss": 4.4159,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671380595572613e-05,
      "loss": 4.4119,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670542000821561e-05,
      "loss": 4.4094,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669703406070509e-05,
      "loss": 4.4203,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668864811319457e-05,
      "loss": 4.4191,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668026216568405e-05,
      "loss": 4.4166,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667187621817353e-05,
      "loss": 4.4169,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6663490270663006e-05,
      "loss": 4.4195,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655104323152486e-05,
      "loss": 4.4143,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66467347544457e-05,
      "loss": 4.4072,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6638348806935175e-05,
      "loss": 4.4234,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662997923822839e-05,
      "loss": 4.4043,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662159329071788e-05,
      "loss": 4.4241,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661320734320736e-05,
      "loss": 4.3968,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660482139569683e-05,
      "loss": 4.391,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659643544818631e-05,
      "loss": 4.3983,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658804950067579e-05,
      "loss": 4.4121,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657966355316527e-05,
      "loss": 4.4031,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657129398445848e-05,
      "loss": 4.3874,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6562924415751695e-05,
      "loss": 4.3895,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6554538468241175e-05,
      "loss": 4.3963,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654615252073065e-05,
      "loss": 4.4109,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653776657322013e-05,
      "loss": 4.3982,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529397004513344e-05,
      "loss": 4.4072,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652101105700283e-05,
      "loss": 4.4018,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512625109492304e-05,
      "loss": 4.4,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504239161981784e-05,
      "loss": 4.402,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495853214471264e-05,
      "loss": 4.3896,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487467266960744e-05,
      "loss": 4.391,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647909769825395e-05,
      "loss": 4.4013,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647071175074343e-05,
      "loss": 4.3849,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646232580323291e-05,
      "loss": 4.3919,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645393985572239e-05,
      "loss": 4.3944,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64455702870156e-05,
      "loss": 4.3988,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643718433950508e-05,
      "loss": 4.3828,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642879839199457e-05,
      "loss": 4.3812,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642041244448405e-05,
      "loss": 4.4068,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641202649697353e-05,
      "loss": 4.385,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640364054946301e-05,
      "loss": 4.3975,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639525460195249e-05,
      "loss": 4.3889,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638686865444197e-05,
      "loss": 4.3955,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637848270693145e-05,
      "loss": 4.385,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637009675942093e-05,
      "loss": 4.3822,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636172719071414e-05,
      "loss": 4.3916,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635334124320362e-05,
      "loss": 4.384,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63449552956931e-05,
      "loss": 4.396,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633656934818258e-05,
      "loss": 4.3905,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632818340067206e-05,
      "loss": 4.3905,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631979745316154e-05,
      "loss": 4.3852,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631142788445475e-05,
      "loss": 4.3766,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630304193694423e-05,
      "loss": 4.3845,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629465598943371e-05,
      "loss": 4.3738,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628627004192319e-05,
      "loss": 4.378,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277884094412666e-05,
      "loss": 4.3789,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269498146902146e-05,
      "loss": 4.3709,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261112199391626e-05,
      "loss": 4.3752,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6252742630684835e-05,
      "loss": 4.374,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.347803592681885,
      "eval_runtime": 537.7927,
      "eval_samples_per_second": 709.55,
      "eval_steps_per_second": 22.174,
      "step": 228957
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.271601361142906e+16,
  "trial_name": null,
  "trial_params": null
}