{
  "best_metric": 4.340500354766846,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-only/lstm/0/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": 10.8202,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.547,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0571,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9934,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9468,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8982,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.725,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6126,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.5081,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.4463,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.4,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3472,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.2841,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.2292,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1797,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.1122,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0736,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.0188,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9675,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9339,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.896,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8561,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8321,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7888,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7709,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7445,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 5.7194,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.6909,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6583,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.648,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6244,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.6117,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5857,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.5673,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714926921053515e-05,
      "loss": 5.5563,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706540973542995e-05,
      "loss": 5.5398,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5212,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4894,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4849,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9673013562304644e-05,
      "loss": 5.4546,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9664627614794124e-05,
      "loss": 5.4559,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9656241667283604e-05,
      "loss": 5.4293,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4317,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.403,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963110020355577e-05,
      "loss": 5.3953,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962271425604525e-05,
      "loss": 5.3893,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3817,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3648,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959757279231743e-05,
      "loss": 5.3418,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958918684480691e-05,
      "loss": 5.3401,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958080089729639e-05,
      "loss": 5.342,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957241494978587e-05,
      "loss": 5.3304,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3095,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.2976,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954727348605804e-05,
      "loss": 5.2945,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953888753854752e-05,
      "loss": 5.2823,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2912,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2487,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2604,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.248,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2332,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948858823228813e-05,
      "loss": 5.2389,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.2182,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.2085,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94634467685603e-05,
      "loss": 5.2035,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945506082104978e-05,
      "loss": 5.2145,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9446674873539255e-05,
      "loss": 5.1901,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9438288926028735e-05,
      "loss": 5.1845,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9429902978518215e-05,
      "loss": 5.1701,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942153340981143e-05,
      "loss": 5.1618,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413147462300904e-05,
      "loss": 5.1668,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404761514790384e-05,
      "loss": 5.1691,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.939637556727987e-05,
      "loss": 5.1435,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938800599857309e-05,
      "loss": 5.1539,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937962005106256e-05,
      "loss": 5.1543,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1332,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.1173,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.1174,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934609263982421e-05,
      "loss": 5.1117,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933770669231369e-05,
      "loss": 5.0993,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932932074480317e-05,
      "loss": 5.0974,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932093479729265e-05,
      "loss": 5.0999,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931254884978213e-05,
      "loss": 5.0983,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0812,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0615,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287407386054305e-05,
      "loss": 5.0666,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279021438543785e-05,
      "loss": 5.0706,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270635491033264e-05,
      "loss": 5.0589,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262249543522744e-05,
      "loss": 5.0568,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253863596012224e-05,
      "loss": 5.0377,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0379,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.0309,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922872213228439e-05,
      "loss": 5.0307,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922033618477387e-05,
      "loss": 5.0305,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921195023726335e-05,
      "loss": 5.0128,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920358066855656e-05,
      "loss": 5.0292,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919519472104604e-05,
      "loss": 5.0132,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9943,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9899,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9894,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 4.9878,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915329774110091e-05,
      "loss": 4.9867,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.9799,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9712,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.9678,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9463,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911136800354831e-05,
      "loss": 4.9515,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910298205603779e-05,
      "loss": 4.9491,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909459610852727e-05,
      "loss": 4.9446,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.9392,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907784059230996e-05,
      "loss": 4.9449,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906945464479944e-05,
      "loss": 4.9204,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906106869728892e-05,
      "loss": 4.9235,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9052682749778396e-05,
      "loss": 4.9176,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.9228,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.928,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.911,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019155338540045e-05,
      "loss": 4.9076,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.8984,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.9066,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994013874812214e-05,
      "loss": 4.8998,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89856279273017e-05,
      "loss": 4.8963,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8874,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8823,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8748,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.8645,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8692,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8651,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8591,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8513,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.8664,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.8667,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889343164109717e-05,
      "loss": 4.8561,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.8428,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.8431,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.8456,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885990422985882e-05,
      "loss": 4.8417,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.8373,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8329,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8291,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8232,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.8195,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88095885447957e-05,
      "loss": 4.8231,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.811,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879283302857839e-05,
      "loss": 4.8058,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878444708106787e-05,
      "loss": 4.8135,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877606113355735e-05,
      "loss": 4.799,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.8085,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.7975,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7901,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.755948543548584,
      "eval_runtime": 453.1859,
      "eval_samples_per_second": 842.019,
      "eval_steps_per_second": 26.314,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7832,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.7828,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.7934,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.781,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709006311080657e-05,
      "loss": 4.7841,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700620363570137e-05,
      "loss": 4.771,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.7726,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683848468549096e-05,
      "loss": 4.7595,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675462521038576e-05,
      "loss": 4.7702,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667076573528056e-05,
      "loss": 4.7681,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658690626017536e-05,
      "loss": 4.7749,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865033743611448e-05,
      "loss": 4.7534,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864195148860396e-05,
      "loss": 4.754,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863356554109344e-05,
      "loss": 4.7478,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862517959358292e-05,
      "loss": 4.7394,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86167936460724e-05,
      "loss": 4.743,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608407698561874e-05,
      "loss": 4.751,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600021751051354e-05,
      "loss": 4.7324,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591635803540834e-05,
      "loss": 4.7345,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583249856030314e-05,
      "loss": 4.7479,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574863908519794e-05,
      "loss": 4.7361,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856649433981301e-05,
      "loss": 4.7357,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855810839230249e-05,
      "loss": 4.723,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854972244479197e-05,
      "loss": 4.7371,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854133649728145e-05,
      "loss": 4.7234,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853296692857466e-05,
      "loss": 4.7203,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852458098106414e-05,
      "loss": 4.7167,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851619503355362e-05,
      "loss": 4.7093,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85078090860431e-05,
      "loss": 4.7032,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849942313853258e-05,
      "loss": 4.705,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849103719102206e-05,
      "loss": 4.7161,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848265124351154e-05,
      "loss": 4.7086,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847428167480475e-05,
      "loss": 4.7017,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846589572729423e-05,
      "loss": 4.7116,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845750977978371e-05,
      "loss": 4.7031,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449123832273194e-05,
      "loss": 4.6994,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440737884762674e-05,
      "loss": 4.6858,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432351937252154e-05,
      "loss": 4.6878,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423965989741634e-05,
      "loss": 4.6776,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415596421034843e-05,
      "loss": 4.688,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840721047352432e-05,
      "loss": 4.6721,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83988245260138e-05,
      "loss": 4.6872,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839043857850328e-05,
      "loss": 4.669,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838206900979649e-05,
      "loss": 4.6758,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837368306228597e-05,
      "loss": 4.6746,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836529711477545e-05,
      "loss": 4.6734,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835691116726493e-05,
      "loss": 4.6691,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834852521975441e-05,
      "loss": 4.6568,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834013927224389e-05,
      "loss": 4.6554,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833175332473337e-05,
      "loss": 4.6718,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832336737722285e-05,
      "loss": 4.6615,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831499780851606e-05,
      "loss": 4.6564,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830661186100554e-05,
      "loss": 4.6453,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829822591349502e-05,
      "loss": 4.6553,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82898399659845e-05,
      "loss": 4.6414,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828147039727771e-05,
      "loss": 4.6689,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827308444976719e-05,
      "loss": 4.6305,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826469850225667e-05,
      "loss": 4.6516,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825631255474615e-05,
      "loss": 4.6405,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247942986039366e-05,
      "loss": 4.6323,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823957341733258e-05,
      "loss": 4.643,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823118746982206e-05,
      "loss": 4.6381,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222801522311535e-05,
      "loss": 4.6212,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214415574801015e-05,
      "loss": 4.6286,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820604600609423e-05,
      "loss": 4.6411,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819766005858371e-05,
      "loss": 4.626,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189274111073184e-05,
      "loss": 4.6233,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180888163562664e-05,
      "loss": 4.6206,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817251859485588e-05,
      "loss": 4.614,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816413264734535e-05,
      "loss": 4.6246,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815574669983483e-05,
      "loss": 4.6361,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814736075232432e-05,
      "loss": 4.612,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81389748048138e-05,
      "loss": 4.6304,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813058885730328e-05,
      "loss": 4.6298,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812220290979276e-05,
      "loss": 4.6175,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811381696228224e-05,
      "loss": 4.6053,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810544739357545e-05,
      "loss": 4.6156,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809707782486866e-05,
      "loss": 4.6112,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088708256161873e-05,
      "loss": 4.6024,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808032230865135e-05,
      "loss": 4.6035,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8071936361140827e-05,
      "loss": 4.6077,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063550413630307e-05,
      "loss": 4.6164,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055164466119786e-05,
      "loss": 4.6043,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804677851860927e-05,
      "loss": 4.5882,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803839257109875e-05,
      "loss": 4.5983,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803000662358823e-05,
      "loss": 4.6012,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802162067607771e-05,
      "loss": 4.5964,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801325110737092e-05,
      "loss": 4.6027,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80048651598604e-05,
      "loss": 4.588,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799647921234988e-05,
      "loss": 4.584,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798809326483936e-05,
      "loss": 4.5847,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797972369613257e-05,
      "loss": 4.5888,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797133774862205e-05,
      "loss": 4.5875,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796295180111153e-05,
      "loss": 4.5809,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795456585360101e-05,
      "loss": 4.5988,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794617990609049e-05,
      "loss": 4.5878,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793781033738371e-05,
      "loss": 4.5695,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792942438987319e-05,
      "loss": 4.5736,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792103844236267e-05,
      "loss": 4.5765,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791265249485215e-05,
      "loss": 4.5761,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7904282926145356e-05,
      "loss": 4.5754,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895896978634836e-05,
      "loss": 4.5741,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887511031124316e-05,
      "loss": 4.5675,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879125083613796e-05,
      "loss": 4.5721,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870739136103276e-05,
      "loss": 4.548,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862353188592756e-05,
      "loss": 4.5639,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853983619885965e-05,
      "loss": 4.5581,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845597672375445e-05,
      "loss": 4.5614,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837211724864925e-05,
      "loss": 4.5569,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782882577735441e-05,
      "loss": 4.5659,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782045620864762e-05,
      "loss": 4.5463,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78120702611371e-05,
      "loss": 4.549,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780368431362658e-05,
      "loss": 4.5516,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779529836611606e-05,
      "loss": 4.55,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778692879740927e-05,
      "loss": 4.5656,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777854284989875e-05,
      "loss": 4.5528,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777015690238823e-05,
      "loss": 4.5511,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776177095487771e-05,
      "loss": 4.5444,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775338500736719e-05,
      "loss": 4.5511,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77450154386604e-05,
      "loss": 4.5496,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773662949114988e-05,
      "loss": 4.5425,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772824354363936e-05,
      "loss": 4.5425,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719857596128845e-05,
      "loss": 4.5379,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7711488027422054e-05,
      "loss": 4.5356,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7703102079911534e-05,
      "loss": 4.5273,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694716132401014e-05,
      "loss": 4.5327,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686330184890494e-05,
      "loss": 4.5291,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767794423737997e-05,
      "loss": 4.5336,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766955828986945e-05,
      "loss": 4.5197,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766117234235893e-05,
      "loss": 4.5416,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652802773652136e-05,
      "loss": 4.5471,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644416826141616e-05,
      "loss": 4.5332,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636030878631096e-05,
      "loss": 4.5263,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762764493112058e-05,
      "loss": 4.5273,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761925898361006e-05,
      "loss": 4.5314,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761088941490327e-05,
      "loss": 4.5303,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760250346739275e-05,
      "loss": 4.5255,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759411751988223e-05,
      "loss": 4.5241,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758573157237171e-05,
      "loss": 4.5206,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757734562486119e-05,
      "loss": 4.5247,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756895967735067e-05,
      "loss": 4.5151,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756057372984015e-05,
      "loss": 4.5245,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755218778232963e-05,
      "loss": 4.5156,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754381821362284e-05,
      "loss": 4.5082,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753543226611232e-05,
      "loss": 4.5212,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75270463186018e-05,
      "loss": 4.5093,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518676749895016e-05,
      "loss": 4.5172,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510290802384496e-05,
      "loss": 4.506,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7501904854873976e-05,
      "loss": 4.504,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.474734306335449,
      "eval_runtime": 450.1395,
      "eval_samples_per_second": 847.717,
      "eval_steps_per_second": 26.492,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7493518907363456e-05,
      "loss": 4.5017,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7485132959852936e-05,
      "loss": 4.4944,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7476747012342416e-05,
      "loss": 4.5149,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7468361064831896e-05,
      "loss": 4.5004,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7459975117321376e-05,
      "loss": 4.5093,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745158916981085e-05,
      "loss": 4.4932,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744320322230033e-05,
      "loss": 4.5019,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743481727478981e-05,
      "loss": 4.4837,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742643132727929e-05,
      "loss": 4.4998,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741804537976877e-05,
      "loss": 4.5034,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740965943225825e-05,
      "loss": 4.5046,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401273484747736e-05,
      "loss": 4.4937,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392887537237216e-05,
      "loss": 4.4848,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384501589726696e-05,
      "loss": 4.4867,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376115642216176e-05,
      "loss": 4.4806,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7367746073509385e-05,
      "loss": 4.484,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7359376504802594e-05,
      "loss": 4.4907,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7350990557292074e-05,
      "loss": 4.4765,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7342604609781554e-05,
      "loss": 4.4808,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7334218662271034e-05,
      "loss": 4.4995,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7325832714760514e-05,
      "loss": 4.4844,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7317446767249994e-05,
      "loss": 4.487,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7309060819739474e-05,
      "loss": 4.476,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7300674872228953e-05,
      "loss": 4.4953,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292288924718433e-05,
      "loss": 4.4713,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728390297720791e-05,
      "loss": 4.4774,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727551702969739e-05,
      "loss": 4.4721,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726713108218687e-05,
      "loss": 4.4716,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725876151348009e-05,
      "loss": 4.4608,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725037556596957e-05,
      "loss": 4.4672,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724198961845904e-05,
      "loss": 4.4777,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723360367094852e-05,
      "loss": 4.4759,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7225217723438e-05,
      "loss": 4.4701,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721684815473121e-05,
      "loss": 4.4724,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720846220722069e-05,
      "loss": 4.4696,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720007625971017e-05,
      "loss": 4.4705,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719170669100339e-05,
      "loss": 4.4585,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718332074349287e-05,
      "loss": 4.4602,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717493479598235e-05,
      "loss": 4.4505,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716654884847183e-05,
      "loss": 4.465,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715816290096131e-05,
      "loss": 4.4491,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714977695345079e-05,
      "loss": 4.4617,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714139100594027e-05,
      "loss": 4.4515,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713300505842975e-05,
      "loss": 4.4509,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7124635489722956e-05,
      "loss": 4.46,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7116249542212436e-05,
      "loss": 4.4507,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7107863594701916e-05,
      "loss": 4.454,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7099477647191396e-05,
      "loss": 4.4362,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7091091699680876e-05,
      "loss": 4.4414,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082705752170356e-05,
      "loss": 4.4557,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074319804659836e-05,
      "loss": 4.4469,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706593385714932e-05,
      "loss": 4.4487,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705756428844253e-05,
      "loss": 4.4322,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704919471973574e-05,
      "loss": 4.4456,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704080877222522e-05,
      "loss": 4.4328,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70324228247147e-05,
      "loss": 4.4586,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702403687720418e-05,
      "loss": 4.4261,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701565092969366e-05,
      "loss": 4.4453,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700726498218314e-05,
      "loss": 4.4406,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699887903467262e-05,
      "loss": 4.4265,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69904930871621e-05,
      "loss": 4.4346,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698212351845531e-05,
      "loss": 4.4399,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.697373757094479e-05,
      "loss": 4.4212,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965351623434276e-05,
      "loss": 4.4296,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956965675923756e-05,
      "loss": 4.4392,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694857972841323e-05,
      "loss": 4.4325,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940210159706445e-05,
      "loss": 4.4225,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931824212195925e-05,
      "loss": 4.4265,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69234382646854e-05,
      "loss": 4.4191,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691505231717488e-05,
      "loss": 4.4311,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906682748468094e-05,
      "loss": 4.4438,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898296800957574e-05,
      "loss": 4.422,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688991085344705e-05,
      "loss": 4.4387,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688152490593653e-05,
      "loss": 4.4378,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873138958426014e-05,
      "loss": 4.429,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864753010915494e-05,
      "loss": 4.4154,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68563834422087e-05,
      "loss": 4.4318,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684799749469818e-05,
      "loss": 4.4221,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683961154718766e-05,
      "loss": 4.4199,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683124197848087e-05,
      "loss": 4.4167,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682285603097035e-05,
      "loss": 4.4215,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681447008345983e-05,
      "loss": 4.4349,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680608413594931e-05,
      "loss": 4.4196,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679769818843879e-05,
      "loss": 4.4044,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678931224092827e-05,
      "loss": 4.4152,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678092629341775e-05,
      "loss": 4.4205,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677254034590723e-05,
      "loss": 4.4187,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676415439839671e-05,
      "loss": 4.4245,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675578482968993e-05,
      "loss": 4.4084,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674739888217941e-05,
      "loss": 4.4079,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673901293466889e-05,
      "loss": 4.4025,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673062698715837e-05,
      "loss": 4.4105,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6722257418451576e-05,
      "loss": 4.4122,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6713887849744785e-05,
      "loss": 4.4103,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705501902234265e-05,
      "loss": 4.4272,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6697115954723745e-05,
      "loss": 4.4078,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6688730007213225e-05,
      "loss": 4.4018,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6680344059702705e-05,
      "loss": 4.4008,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6671958112192185e-05,
      "loss": 4.4045,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66635885434854e-05,
      "loss": 4.4094,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665520259597488e-05,
      "loss": 4.4072,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664681664846436e-05,
      "loss": 4.4034,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663843070095384e-05,
      "loss": 4.3993,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663004475344332e-05,
      "loss": 4.4053,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662167518473653e-05,
      "loss": 4.3807,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661328923722601e-05,
      "loss": 4.3985,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660490328971549e-05,
      "loss": 4.3901,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659651734220497e-05,
      "loss": 4.3978,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658813139469445e-05,
      "loss": 4.3929,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657974544718393e-05,
      "loss": 4.4025,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657135949967341e-05,
      "loss": 4.3838,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656297355216288e-05,
      "loss": 4.3843,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655458760465237e-05,
      "loss": 4.3959,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654620165714185e-05,
      "loss": 4.3866,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653781570963133e-05,
      "loss": 4.4063,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652942976212081e-05,
      "loss": 4.3966,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6521076572217755e-05,
      "loss": 4.3955,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512690624707234e-05,
      "loss": 4.3855,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650430467719671e-05,
      "loss": 4.3906,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649591872968619e-05,
      "loss": 4.3908,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648753278217567e-05,
      "loss": 4.387,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647914683466515e-05,
      "loss": 4.3895,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647076088715463e-05,
      "loss": 4.3789,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646237493964411e-05,
      "loss": 4.3834,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645400537093732e-05,
      "loss": 4.3719,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644563580223053e-05,
      "loss": 4.3811,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643724985472001e-05,
      "loss": 4.3753,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642886390720949e-05,
      "loss": 4.3796,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642047795969897e-05,
      "loss": 4.3684,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641209201218845e-05,
      "loss": 4.3922,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640370606467793e-05,
      "loss": 4.3945,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639532011716741e-05,
      "loss": 4.3857,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638693416965689e-05,
      "loss": 4.3719,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637854822214637e-05,
      "loss": 4.3799,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637017865343958e-05,
      "loss": 4.3799,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636179270592906e-05,
      "loss": 4.3844,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635340675841854e-05,
      "loss": 4.3755,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634502081090802e-05,
      "loss": 4.3751,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633663486339751e-05,
      "loss": 4.3737,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632824891588699e-05,
      "loss": 4.3796,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631986296837647e-05,
      "loss": 4.3703,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631147702086595e-05,
      "loss": 4.3775,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630310745215916e-05,
      "loss": 4.3704,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629472150464864e-05,
      "loss": 4.3654,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628633555713812e-05,
      "loss": 4.3776,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277965988431326e-05,
      "loss": 4.3717,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269580040920806e-05,
      "loss": 4.3676,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261194093410286e-05,
      "loss": 4.3636,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6252808145899766e-05,
      "loss": 4.3606,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.340500354766846,
      "eval_runtime": 299.1295,
      "eval_samples_per_second": 1275.671,
      "eval_steps_per_second": 39.866,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.477738216745003e+16,
  "trial_name": null,
  "trial_params": null
}