{
  "best_metric": 4.342889308929443,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full2/lstm/3/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": 10.8203,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5515,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0524,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9874,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9458,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9008,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7316,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6255,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5262,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4501,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.3882,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3306,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2674,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.1911,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988259673485272e-05,
      "loss": 6.1403,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98742107873422e-05,
      "loss": 6.0839,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986582483983168e-05,
      "loss": 6.0407,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985743889232116e-05,
      "loss": 5.9963,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984905294481064e-05,
      "loss": 5.956,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9087,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8793,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8407,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8162,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.7746,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7653,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7338,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.707,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6743,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6541,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6398,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6092,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.6008,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.5742,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5575,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5558,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 5.5377,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 5.5144,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968980183612942e-05,
      "loss": 5.4868,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96814158886189e-05,
      "loss": 5.4778,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4545,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4529,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4299,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4233,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.3991,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.4005,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.3872,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3871,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3747,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.345,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3369,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3459,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3272,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3134,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.2966,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954727348605804e-05,
      "loss": 5.2988,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2791,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2903,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2554,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2576,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.254,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2366,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948860461109186e-05,
      "loss": 5.2444,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.2158,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.2118,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94634467685603e-05,
      "loss": 5.1997,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.2209,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1865,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1878,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1714,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942153340981143e-05,
      "loss": 5.1632,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413147462300904e-05,
      "loss": 5.1732,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404761514790384e-05,
      "loss": 5.1648,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.939637556727987e-05,
      "loss": 5.1567,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938800599857309e-05,
      "loss": 5.1557,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1571,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1266,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362880913648985e-05,
      "loss": 5.1356,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354494966138465e-05,
      "loss": 5.113,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.1073,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.1084,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.1081,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0975,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.1055,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0859,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0691,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287407386054305e-05,
      "loss": 5.0635,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279021438543785e-05,
      "loss": 5.0766,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0577,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0545,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0512,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0392,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0411,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0358,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 5.0335,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 5.0306,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920359704736029e-05,
      "loss": 5.0209,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919521109984978e-05,
      "loss": 5.0045,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 5.0058,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9956,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9938,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 5.002,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915329774110091e-05,
      "loss": 4.981,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.9847,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9805,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128156277373076e-05,
      "loss": 4.9681,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119770329862556e-05,
      "loss": 4.9596,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9597,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9439,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9094612487330996e-05,
      "loss": 4.9456,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.9433,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9419,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9353,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9316,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.924,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904432955987534e-05,
      "loss": 4.9265,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903594361236482e-05,
      "loss": 4.928,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.92,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.9204,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901080214863699e-05,
      "loss": 4.9076,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900241620112647e-05,
      "loss": 4.9106,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.9063,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8915,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977274737398646e-05,
      "loss": 4.8894,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968888789888125e-05,
      "loss": 4.8865,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8877,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.8737,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8803,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8732,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8662,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8576,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910219914921943e-05,
      "loss": 4.8762,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8901833967411423e-05,
      "loss": 4.866,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.857,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.857,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.8543,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.8517,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.8532,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885153466115204e-05,
      "loss": 4.844,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.8376,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883477914493473e-05,
      "loss": 4.8346,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.8232,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881800724991369e-05,
      "loss": 4.8211,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880962130240316e-05,
      "loss": 4.8199,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.823,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.8154,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8173,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.811,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.8121,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.817,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7949,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.762335300445557,
      "eval_runtime": 293.5975,
      "eval_samples_per_second": 1299.708,
      "eval_steps_per_second": 40.617,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7876,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.785,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.8037,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717408637394906e-05,
      "loss": 4.7866,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709022689884386e-05,
      "loss": 4.7921,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700636742373866e-05,
      "loss": 4.7832,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692250794863346e-05,
      "loss": 4.7782,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683881226156555e-05,
      "loss": 4.7653,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867549527864604e-05,
      "loss": 4.7706,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.7674,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7878,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865035381491821e-05,
      "loss": 4.7707,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864198424621142e-05,
      "loss": 4.755,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86335982987009e-05,
      "loss": 4.7553,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862521235119038e-05,
      "loss": 4.7559,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861682640367986e-05,
      "loss": 4.7459,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860845683497307e-05,
      "loss": 4.7588,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860007088746255e-05,
      "loss": 4.751,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.7415,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.7574,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.7421,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566527097420475e-05,
      "loss": 4.743,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558141149909955e-05,
      "loss": 4.7292,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.7512,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541385633692644e-05,
      "loss": 4.7294,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532999686182124e-05,
      "loss": 4.727,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524613738671604e-05,
      "loss": 4.718,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8516227791161084e-05,
      "loss": 4.7207,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8507841843650564e-05,
      "loss": 4.7171,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8499455896140044e-05,
      "loss": 4.7123,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8491069948629524e-05,
      "loss": 4.7186,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848270037992273e-05,
      "loss": 4.715,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847431443241221e-05,
      "loss": 4.7105,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.721,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845755891619491e-05,
      "loss": 4.7124,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844917296868439e-05,
      "loss": 4.7078,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844078702117387e-05,
      "loss": 4.6942,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843240107366335e-05,
      "loss": 4.6968,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842401512615282e-05,
      "loss": 4.688,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841564555744604e-05,
      "loss": 4.6952,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840725960993552e-05,
      "loss": 4.6832,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8398873662425e-05,
      "loss": 4.6922,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839048771491447e-05,
      "loss": 4.6729,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838210176740395e-05,
      "loss": 4.6952,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837371581989343e-05,
      "loss": 4.6819,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836532987238292e-05,
      "loss": 4.6876,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83569439248724e-05,
      "loss": 4.6826,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834855797736188e-05,
      "loss": 4.6664,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.666,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8331802461144566e-05,
      "loss": 4.6812,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8323432892437775e-05,
      "loss": 4.6662,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8315046944927255e-05,
      "loss": 4.6673,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8306660997416735e-05,
      "loss": 4.6542,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298275049906215e-05,
      "loss": 4.6648,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289889102395695e-05,
      "loss": 4.6465,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281503154885175e-05,
      "loss": 4.6723,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273117207374655e-05,
      "loss": 4.6371,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264731259864135e-05,
      "loss": 4.6563,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825636169115735e-05,
      "loss": 4.6518,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.824797574364683e-05,
      "loss": 4.6436,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823958979613631e-05,
      "loss": 4.6523,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823120384862579e-05,
      "loss": 4.6415,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822281790111527e-05,
      "loss": 4.6332,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821444833240848e-05,
      "loss": 4.6295,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820607876370169e-05,
      "loss": 4.6542,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81977091949949e-05,
      "loss": 4.6244,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818932324748438e-05,
      "loss": 4.6368,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818093729997386e-05,
      "loss": 4.6186,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817255135246334e-05,
      "loss": 4.6255,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816416540495282e-05,
      "loss": 4.6358,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155779457442305e-05,
      "loss": 4.6325,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147393509931785e-05,
      "loss": 4.6262,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8139007562421265e-05,
      "loss": 4.6335,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8130621614910745e-05,
      "loss": 4.634,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8122235667400224e-05,
      "loss": 4.6193,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113849719889704e-05,
      "loss": 4.6257,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105463772379184e-05,
      "loss": 4.612,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809707782486866e-05,
      "loss": 4.6132,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80887246349656e-05,
      "loss": 4.6085,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808033868745508e-05,
      "loss": 4.6172,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807195273994456e-05,
      "loss": 4.6159,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806358317123777e-05,
      "loss": 4.6225,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805519722372726e-05,
      "loss": 4.6094,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804681127621674e-05,
      "loss": 4.5917,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803842532870622e-05,
      "loss": 4.593,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803005575999943e-05,
      "loss": 4.6067,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802166981248891e-05,
      "loss": 4.6003,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801328386497839e-05,
      "loss": 4.5967,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800489791746787e-05,
      "loss": 4.6022,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799651196995735e-05,
      "loss": 4.5877,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798812602244683e-05,
      "loss": 4.5936,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79797400749363e-05,
      "loss": 4.5935,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797135412742578e-05,
      "loss": 4.5985,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796296817991526e-05,
      "loss": 4.5953,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795458223240474e-05,
      "loss": 4.5944,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794619628489423e-05,
      "loss": 4.5761,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793781033738371e-05,
      "loss": 4.5821,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792942438987319e-05,
      "loss": 4.5807,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792103844236267e-05,
      "loss": 4.5793,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912668873655876e-05,
      "loss": 4.5942,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7904282926145356e-05,
      "loss": 4.5713,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895896978634836e-05,
      "loss": 4.5793,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887511031124316e-05,
      "loss": 4.5778,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879125083613796e-05,
      "loss": 4.5722,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870739136103276e-05,
      "loss": 4.5623,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862353188592756e-05,
      "loss": 4.5718,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853967241082236e-05,
      "loss": 4.5585,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845581293571715e-05,
      "loss": 4.5593,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837195346061195e-05,
      "loss": 4.5608,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828809398550675e-05,
      "loss": 4.5598,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820423451040155e-05,
      "loss": 4.5556,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781205388233337e-05,
      "loss": 4.5633,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780368431362658e-05,
      "loss": 4.5495,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779529836611606e-05,
      "loss": 4.5591,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778692879740927e-05,
      "loss": 4.5673,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777854284989875e-05,
      "loss": 4.5585,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777017328119196e-05,
      "loss": 4.5619,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776178733368144e-05,
      "loss": 4.5531,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775340138617092e-05,
      "loss": 4.5524,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77450154386604e-05,
      "loss": 4.5545,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773662949114988e-05,
      "loss": 4.544,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772824354363936e-05,
      "loss": 4.5423,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719857596128845e-05,
      "loss": 4.5435,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771147164861832e-05,
      "loss": 4.5455,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77030857011078e-05,
      "loss": 4.5407,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769469975359728e-05,
      "loss": 4.5419,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768631380608676e-05,
      "loss": 4.5402,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767792785857624e-05,
      "loss": 4.5333,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766954191106572e-05,
      "loss": 4.5284,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76611559635552e-05,
      "loss": 4.5512,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765277001604468e-05,
      "loss": 4.5436,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764440044733789e-05,
      "loss": 4.5396,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763601449982737e-05,
      "loss": 4.5349,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762762855231685e-05,
      "loss": 4.5354,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761924260480633e-05,
      "loss": 4.5341,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610856657295813e-05,
      "loss": 4.5434,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760248708858902e-05,
      "loss": 4.5329,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75941011410785e-05,
      "loss": 4.5264,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758571519356798e-05,
      "loss": 4.5265,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757732924605746e-05,
      "loss": 4.521,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756895967735067e-05,
      "loss": 4.5161,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756057372984015e-05,
      "loss": 4.521,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755218778232963e-05,
      "loss": 4.5266,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754380183481911e-05,
      "loss": 4.5146,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753543226611232e-05,
      "loss": 4.5259,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75270463186018e-05,
      "loss": 4.521,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751866037109128e-05,
      "loss": 4.5169,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751027442358077e-05,
      "loss": 4.5213,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750188847607025e-05,
      "loss": 4.5075,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.479409217834473,
      "eval_runtime": 293.7436,
      "eval_samples_per_second": 1299.062,
      "eval_steps_per_second": 40.597,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749350252855973e-05,
      "loss": 4.4995,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748511658104921e-05,
      "loss": 4.4983,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747673063353868e-05,
      "loss": 4.5189,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746834468602816e-05,
      "loss": 4.5077,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745995873851764e-05,
      "loss": 4.5145,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745157279100712e-05,
      "loss": 4.5022,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.74431868434966e-05,
      "loss": 4.5065,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743480089598608e-05,
      "loss": 4.4911,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742641494847556e-05,
      "loss": 4.501,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741802900096504e-05,
      "loss": 4.4962,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740964305345452e-05,
      "loss": 4.5181,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401257105944e-05,
      "loss": 4.5074,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392887537237216e-05,
      "loss": 4.4853,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384501589726696e-05,
      "loss": 4.4923,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376115642216176e-05,
      "loss": 4.4998,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7367729694705656e-05,
      "loss": 4.4809,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7359343747195135e-05,
      "loss": 4.4962,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7350957799684615e-05,
      "loss": 4.4922,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7342571852174095e-05,
      "loss": 4.4851,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7334202283467305e-05,
      "loss": 4.5085,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7325816335956784e-05,
      "loss": 4.4866,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7317430388446264e-05,
      "loss": 4.4923,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730904444093574e-05,
      "loss": 4.4783,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730065849342522e-05,
      "loss": 4.5014,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292272545914704e-05,
      "loss": 4.4772,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728390297720791e-05,
      "loss": 4.4817,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727553340850113e-05,
      "loss": 4.4704,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726714746099061e-05,
      "loss": 4.481,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725876151348009e-05,
      "loss": 4.4726,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725037556596957e-05,
      "loss": 4.4733,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724198961845904e-05,
      "loss": 4.4795,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723360367094852e-05,
      "loss": 4.4801,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7225217723438e-05,
      "loss": 4.471,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721683177592748e-05,
      "loss": 4.4862,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720846220722069e-05,
      "loss": 4.4734,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720007625971017e-05,
      "loss": 4.4768,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719169031219965e-05,
      "loss": 4.463,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718330436468914e-05,
      "loss": 4.4676,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717493479598235e-05,
      "loss": 4.4547,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716654884847183e-05,
      "loss": 4.4695,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715816290096131e-05,
      "loss": 4.4553,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714977695345079e-05,
      "loss": 4.466,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714139100594027e-05,
      "loss": 4.4528,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713300505842975e-05,
      "loss": 4.4672,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712461911091923e-05,
      "loss": 4.466,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711623316340871e-05,
      "loss": 4.4621,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7107863594701916e-05,
      "loss": 4.4664,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7099477647191396e-05,
      "loss": 4.447,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7091091699680876e-05,
      "loss": 4.4469,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082705752170356e-05,
      "loss": 4.463,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074319804659836e-05,
      "loss": 4.4524,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706595023595305e-05,
      "loss": 4.4531,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705758066724626e-05,
      "loss": 4.4395,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704919471973574e-05,
      "loss": 4.4542,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704080877222522e-05,
      "loss": 4.4302,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70324228247147e-05,
      "loss": 4.4629,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702403687720418e-05,
      "loss": 4.4251,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701566730849739e-05,
      "loss": 4.4481,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700728136098687e-05,
      "loss": 4.4474,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699889541347635e-05,
      "loss": 4.4356,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699050946596583e-05,
      "loss": 4.4509,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698212351845531e-05,
      "loss": 4.4358,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973753949748525e-05,
      "loss": 4.4288,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965368002238005e-05,
      "loss": 4.4265,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956982054727485e-05,
      "loss": 4.4507,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948596107216965e-05,
      "loss": 4.4261,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940210159706445e-05,
      "loss": 4.436,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931824212195925e-05,
      "loss": 4.423,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69234382646854e-05,
      "loss": 4.425,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691505231717488e-05,
      "loss": 4.438,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906682748468094e-05,
      "loss": 4.4416,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68983131797613e-05,
      "loss": 4.4261,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688992723225078e-05,
      "loss": 4.4427,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688154128474026e-05,
      "loss": 4.44,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687317171603348e-05,
      "loss": 4.431,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686478576852296e-05,
      "loss": 4.4286,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.685639982101244e-05,
      "loss": 4.4276,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684801387350192e-05,
      "loss": 4.4228,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68396279259914e-05,
      "loss": 4.421,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683124197848087e-05,
      "loss": 4.428,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682285603097035e-05,
      "loss": 4.4319,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681447008345983e-05,
      "loss": 4.4375,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680608413594931e-05,
      "loss": 4.4245,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679769818843879e-05,
      "loss": 4.4046,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678931224092827e-05,
      "loss": 4.409,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678092629341775e-05,
      "loss": 4.4231,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677255672471097e-05,
      "loss": 4.4183,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676417077720045e-05,
      "loss": 4.4149,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675578482968993e-05,
      "loss": 4.42,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6747415260983136e-05,
      "loss": 4.4123,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6739029313472616e-05,
      "loss": 4.4079,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6730643365962096e-05,
      "loss": 4.416,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6722257418451576e-05,
      "loss": 4.4207,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6713871470941056e-05,
      "loss": 4.4168,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705485523430536e-05,
      "loss": 4.4179,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6697115954723745e-05,
      "loss": 4.4057,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6688730007213225e-05,
      "loss": 4.4083,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6680344059702705e-05,
      "loss": 4.4055,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6671958112192185e-05,
      "loss": 4.4087,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6663572164681665e-05,
      "loss": 4.42,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655186217171145e-05,
      "loss": 4.4019,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664680026966063e-05,
      "loss": 4.405,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663841432215011e-05,
      "loss": 4.407,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663002837463959e-05,
      "loss": 4.4075,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66216588059328e-05,
      "loss": 4.3919,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661327285842228e-05,
      "loss": 4.399,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660488691091176e-05,
      "loss": 4.3967,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6596500963401234e-05,
      "loss": 4.3928,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658813139469445e-05,
      "loss": 4.394,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657974544718393e-05,
      "loss": 4.3934,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657135949967341e-05,
      "loss": 4.3959,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656297355216288e-05,
      "loss": 4.3965,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65546039834561e-05,
      "loss": 4.386,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6546218035945586e-05,
      "loss": 4.3946,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6537848467238795e-05,
      "loss": 4.404,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529462519728275e-05,
      "loss": 4.4022,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6521076572217755e-05,
      "loss": 4.397,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512690624707234e-05,
      "loss": 4.3936,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650430467719671e-05,
      "loss": 4.3925,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649591872968619e-05,
      "loss": 4.3954,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648753278217567e-05,
      "loss": 4.3808,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647914683466515e-05,
      "loss": 4.3832,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647076088715463e-05,
      "loss": 4.3846,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646237493964411e-05,
      "loss": 4.3924,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645400537093732e-05,
      "loss": 4.3836,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64456194234268e-05,
      "loss": 4.388,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643723347591628e-05,
      "loss": 4.3851,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642884752840576e-05,
      "loss": 4.3784,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642047795969897e-05,
      "loss": 4.3727,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641209201218845e-05,
      "loss": 4.3952,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640370606467793e-05,
      "loss": 4.3928,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639532011716741e-05,
      "loss": 4.3842,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638695054846062e-05,
      "loss": 4.3811,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63785646009501e-05,
      "loss": 4.3815,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637017865343958e-05,
      "loss": 4.3857,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636179270592906e-05,
      "loss": 4.3976,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635342313722228e-05,
      "loss": 4.3803,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634503718971176e-05,
      "loss": 4.3745,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633665124220124e-05,
      "loss": 4.3756,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632826529469072e-05,
      "loss": 4.3725,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6319895725983926e-05,
      "loss": 4.3729,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6311509778473406e-05,
      "loss": 4.3733,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6303123830962886e-05,
      "loss": 4.3787,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6294737883452366e-05,
      "loss": 4.372,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6286351935941846e-05,
      "loss": 4.3767,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277998746038784e-05,
      "loss": 4.3788,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269612798528264e-05,
      "loss": 4.3694,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261226851017744e-05,
      "loss": 4.3768,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625284090350723e-05,
      "loss": 4.3602,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.342889308929443,
      "eval_runtime": 288.3332,
      "eval_samples_per_second": 1323.438,
      "eval_steps_per_second": 41.358,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.459309247762675e+16,
  "trial_name": null,
  "trial_params": null
}