{
  "best_metric": 4.484649181365967,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-only/lstm/2/checkpoints/checkpoint-152640",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 152640,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8185,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5488,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0593,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9949,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9472,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9233,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.764,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6598,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5464,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4721,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991617328250227e-05,
      "loss": 6.4071,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990778733499175e-05,
      "loss": 6.3339,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989940138748123e-05,
      "loss": 6.2532,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989101543997071e-05,
      "loss": 6.1933,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988262949246018e-05,
      "loss": 6.1447,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987424354494966e-05,
      "loss": 6.0824,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986585759743914e-05,
      "loss": 6.0491,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985747164992862e-05,
      "loss": 6.0018,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98490857024181e-05,
      "loss": 5.9554,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984069975490758e-05,
      "loss": 5.9253,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983231380739706e-05,
      "loss": 5.8907,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 5.8543,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 5.8311,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.7891,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7717,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7442,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7202,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6911,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6557,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6451,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.622,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.6095,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5842,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.565,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5561,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5396,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5203,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4888,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4847,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4545,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4576,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.431,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4339,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.4069,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.3993,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.393,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3838,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3682,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959757279231743e-05,
      "loss": 5.3472,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958918684480691e-05,
      "loss": 5.345,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3475,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.3356,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3154,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.3045,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.3004,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2884,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2985,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2552,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2682,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.254,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2409,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948858823228813e-05,
      "loss": 5.246,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.226,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.2144,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94634467685603e-05,
      "loss": 5.2113,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945506082104978e-05,
      "loss": 5.2228,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944669125234299e-05,
      "loss": 5.1998,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1937,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1776,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942153340981143e-05,
      "loss": 5.1706,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413147462300904e-05,
      "loss": 5.1754,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.1795,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93963919460836e-05,
      "loss": 5.1534,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938800599857309e-05,
      "loss": 5.1631,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937962005106256e-05,
      "loss": 5.1652,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937123410355204e-05,
      "loss": 5.1451,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936284815604152e-05,
      "loss": 5.1265,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.1288,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934609263982421e-05,
      "loss": 5.1227,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.1117,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.1101,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.113,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.1124,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0934,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0733,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287407386054305e-05,
      "loss": 5.0794,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279021438543785e-05,
      "loss": 5.0844,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270635491033264e-05,
      "loss": 5.0729,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262249543522744e-05,
      "loss": 5.0711,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.0518,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0505,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.0461,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922872213228439e-05,
      "loss": 5.0449,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922033618477387e-05,
      "loss": 5.0448,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921195023726335e-05,
      "loss": 5.029,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920356428975283e-05,
      "loss": 5.0444,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919517834224231e-05,
      "loss": 5.0288,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918680877353552e-05,
      "loss": 5.0083,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917842282602501e-05,
      "loss": 5.0058,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917003687851449e-05,
      "loss": 5.0055,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 5.0023,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915328136229718e-05,
      "loss": 5.0017,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914489541478666e-05,
      "loss": 4.9939,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913650946727614e-05,
      "loss": 4.9868,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912812351976562e-05,
      "loss": 4.983,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9621,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911136800354831e-05,
      "loss": 4.967,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910298205603779e-05,
      "loss": 4.9646,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909459610852727e-05,
      "loss": 4.9596,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908621016101675e-05,
      "loss": 4.9544,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907784059230996e-05,
      "loss": 4.9614,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906945464479944e-05,
      "loss": 4.9365,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906106869728892e-05,
      "loss": 4.9402,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9052682749778396e-05,
      "loss": 4.9341,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.938,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.9444,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.926,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019155338540045e-05,
      "loss": 4.924,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010769391029525e-05,
      "loss": 4.9146,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002383443519005e-05,
      "loss": 4.9209,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8993997496008485e-05,
      "loss": 4.9152,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89856279273017e-05,
      "loss": 4.9097,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897724197979118e-05,
      "loss": 4.9011,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896885603228066e-05,
      "loss": 4.8963,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896047008477014e-05,
      "loss": 4.8899,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895208413725962e-05,
      "loss": 4.8789,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.894371456855283e-05,
      "loss": 4.8843,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8793,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8745,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8676,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.8809,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.8799,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889341526229344e-05,
      "loss": 4.8697,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888502931478292e-05,
      "loss": 4.8581,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766433672724e-05,
      "loss": 4.8557,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886825741976188e-05,
      "loss": 4.8602,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859871472251365e-05,
      "loss": 4.855,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851485524740845e-05,
      "loss": 4.8533,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843099577230325e-05,
      "loss": 4.8483,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834730008523534e-05,
      "loss": 4.8446,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8374,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.835,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88095885447957e-05,
      "loss": 4.8366,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880120259728518e-05,
      "loss": 4.8252,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879283302857839e-05,
      "loss": 4.8196,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878444708106787e-05,
      "loss": 4.8268,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877606113355735e-05,
      "loss": 4.8124,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876767518604683e-05,
      "loss": 4.821,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.8112,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875091966982953e-05,
      "loss": 4.8029,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.7703471183776855,
      "eval_runtime": 300.3148,
      "eval_samples_per_second": 1270.637,
      "eval_steps_per_second": 39.708,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874253372231901e-05,
      "loss": 4.7957,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873414777480849e-05,
      "loss": 4.797,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872576182729797e-05,
      "loss": 4.8072,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871737587978745e-05,
      "loss": 4.7957,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870898993227693e-05,
      "loss": 4.798,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87006039847664e-05,
      "loss": 4.7853,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869221803725588e-05,
      "loss": 4.7871,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868383208974536e-05,
      "loss": 4.7716,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867544614223484e-05,
      "loss": 4.7842,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667092952331785e-05,
      "loss": 4.7803,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865870700482127e-05,
      "loss": 4.7877,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865032105731075e-05,
      "loss": 4.7673,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8641935109800225e-05,
      "loss": 4.7664,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633549162289705e-05,
      "loss": 4.7617,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625163214779185e-05,
      "loss": 4.7534,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616777267268665e-05,
      "loss": 4.755,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608391319758145e-05,
      "loss": 4.7637,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600021751051354e-05,
      "loss": 4.7468,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859165218234457e-05,
      "loss": 4.748,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858328261363778e-05,
      "loss": 4.7615,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857489666612726e-05,
      "loss": 4.7495,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856651071861674e-05,
      "loss": 4.7491,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558124771106226e-05,
      "loss": 4.7359,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85497388235957e-05,
      "loss": 4.7507,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854135287608518e-05,
      "loss": 4.7368,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853296692857466e-05,
      "loss": 4.7335,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852458098106414e-05,
      "loss": 4.7286,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851621141235735e-05,
      "loss": 4.7238,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850782546484683e-05,
      "loss": 4.7157,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849943951733631e-05,
      "loss": 4.7173,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849105356982579e-05,
      "loss": 4.7286,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848266762231527e-05,
      "loss": 4.722,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847429805360848e-05,
      "loss": 4.7144,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.7248,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8457526158587444e-05,
      "loss": 4.7138,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449140211076924e-05,
      "loss": 4.7121,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844077064237013e-05,
      "loss": 4.6988,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843238469485961e-05,
      "loss": 4.7018,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842399874734909e-05,
      "loss": 4.6893,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841561279983857e-05,
      "loss": 4.7006,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840724323113178e-05,
      "loss": 4.6834,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839885728362126e-05,
      "loss": 4.6984,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839047133611074e-05,
      "loss": 4.6811,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838208538860022e-05,
      "loss": 4.6893,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83736994410897e-05,
      "loss": 4.6858,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836531349357918e-05,
      "loss": 4.6849,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83569439248724e-05,
      "loss": 4.6803,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834855797736188e-05,
      "loss": 4.6683,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.6669,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833178608234084e-05,
      "loss": 4.6827,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832340013483032e-05,
      "loss": 4.6734,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83150141873198e-05,
      "loss": 4.6685,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830662823980928e-05,
      "loss": 4.657,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829824229229876e-05,
      "loss": 4.6672,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289889102395695e-05,
      "loss": 4.653,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281503154885175e-05,
      "loss": 4.6794,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273117207374655e-05,
      "loss": 4.6425,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264731259864135e-05,
      "loss": 4.6628,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256345312353615e-05,
      "loss": 4.6537,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247959364843095e-05,
      "loss": 4.6421,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823957341733258e-05,
      "loss": 4.6557,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823118746982206e-05,
      "loss": 4.6499,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822281790111527e-05,
      "loss": 4.6324,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821443195360475e-05,
      "loss": 4.6391,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820604600609423e-05,
      "loss": 4.6534,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819766005858371e-05,
      "loss": 4.6376,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818929048987692e-05,
      "loss": 4.6343,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81809045423664e-05,
      "loss": 4.6329,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817251859485588e-05,
      "loss": 4.6258,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816413264734535e-05,
      "loss": 4.6362,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815576307863857e-05,
      "loss": 4.6483,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814737713112805e-05,
      "loss": 4.6238,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8139007562421265e-05,
      "loss": 4.6426,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8130621614910745e-05,
      "loss": 4.6416,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8122235667400224e-05,
      "loss": 4.6292,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113849719889704e-05,
      "loss": 4.6158,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105463772379184e-05,
      "loss": 4.628,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809707782486866e-05,
      "loss": 4.6224,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088708256161873e-05,
      "loss": 4.6148,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808032230865135e-05,
      "loss": 4.6155,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807195273994456e-05,
      "loss": 4.6194,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806356679243404e-05,
      "loss": 4.628,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805518084492352e-05,
      "loss": 4.615,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046794897413e-05,
      "loss": 4.5995,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803842532870622e-05,
      "loss": 4.61,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80300393811957e-05,
      "loss": 4.611,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802165343368518e-05,
      "loss": 4.6075,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801326748617466e-05,
      "loss": 4.6127,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800489791746787e-05,
      "loss": 4.5991,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799651196995735e-05,
      "loss": 4.5955,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798812602244683e-05,
      "loss": 4.5938,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79797400749363e-05,
      "loss": 4.6004,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797135412742578e-05,
      "loss": 4.5993,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7962984558718996e-05,
      "loss": 4.5907,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7954598611208476e-05,
      "loss": 4.6104,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7946212663697956e-05,
      "loss": 4.5978,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7937826716187436e-05,
      "loss": 4.5801,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7929440768676916e-05,
      "loss": 4.5843,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7921054821166396e-05,
      "loss": 4.5871,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912668873655876e-05,
      "loss": 4.5876,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7904282926145356e-05,
      "loss": 4.587,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895896978634836e-05,
      "loss": 4.5844,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887511031124316e-05,
      "loss": 4.5797,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879125083613796e-05,
      "loss": 4.5826,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870739136103276e-05,
      "loss": 4.5597,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862369567396485e-05,
      "loss": 4.575,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853983619885965e-05,
      "loss": 4.5677,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845597672375445e-05,
      "loss": 4.5726,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783722810366866e-05,
      "loss": 4.5643,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782884215615814e-05,
      "loss": 4.5759,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782045620864762e-05,
      "loss": 4.5585,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78120702611371e-05,
      "loss": 4.56,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780368431362658e-05,
      "loss": 4.5619,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779529836611606e-05,
      "loss": 4.5614,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778691241860554e-05,
      "loss": 4.5767,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777852647109502e-05,
      "loss": 4.5643,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777017328119196e-05,
      "loss": 4.5623,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776178733368144e-05,
      "loss": 4.554,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775340138617092e-05,
      "loss": 4.5605,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77450154386604e-05,
      "loss": 4.5591,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773662949114988e-05,
      "loss": 4.5521,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772824354363936e-05,
      "loss": 4.5522,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719857596128845e-05,
      "loss": 4.5492,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771147164861832e-05,
      "loss": 4.547,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77030857011078e-05,
      "loss": 4.5364,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769469975359728e-05,
      "loss": 4.5435,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768631380608676e-05,
      "loss": 4.538,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767792785857624e-05,
      "loss": 4.5444,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766955828986945e-05,
      "loss": 4.5296,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766118872116266e-05,
      "loss": 4.5519,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652802773652136e-05,
      "loss": 4.5571,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644416826141616e-05,
      "loss": 4.5417,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636030878631096e-05,
      "loss": 4.535,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762764493112058e-05,
      "loss": 4.5361,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761925898361006e-05,
      "loss": 4.5387,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761087303609954e-05,
      "loss": 4.54,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760248708858902e-05,
      "loss": 4.5364,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75941011410785e-05,
      "loss": 4.5344,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758571519356798e-05,
      "loss": 4.5298,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757732924605746e-05,
      "loss": 4.5334,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756894329854694e-05,
      "loss": 4.5238,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756057372984015e-05,
      "loss": 4.5359,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755218778232963e-05,
      "loss": 4.5257,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754380183481911e-05,
      "loss": 4.5179,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753541588730859e-05,
      "loss": 4.5303,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75270463186018e-05,
      "loss": 4.5196,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518676749895016e-05,
      "loss": 4.5268,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510290802384496e-05,
      "loss": 4.5159,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7501904854873976e-05,
      "loss": 4.5143,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.484649181365967,
      "eval_runtime": 294.8109,
      "eval_samples_per_second": 1294.359,
      "eval_steps_per_second": 40.45,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 6.318492144496669e+16,
  "trial_name": null,
  "trial_params": null
}