{
  "best_metric": 4.159811496734619,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-only/lstm/0/checkpoints/checkpoint-457920",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 457920,
  "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
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6244422198389246e-05,
      "loss": 4.3653,
      "step": 229376
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.623605262968246e-05,
      "loss": 4.3547,
      "step": 229888
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.622766668217194e-05,
      "loss": 4.373,
      "step": 230400
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.621928073466142e-05,
      "loss": 4.36,
      "step": 230912
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6210894787150895e-05,
      "loss": 4.3714,
      "step": 231424
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6202508839640374e-05,
      "loss": 4.3555,
      "step": 231936
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6194122892129854e-05,
      "loss": 4.3607,
      "step": 232448
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6185736944619334e-05,
      "loss": 4.3475,
      "step": 232960
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6177350997108814e-05,
      "loss": 4.3622,
      "step": 233472
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6168981428402023e-05,
      "loss": 4.3657,
      "step": 233984
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.616061185969524e-05,
      "loss": 4.3659,
      "step": 234496
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.615222591218472e-05,
      "loss": 4.3591,
      "step": 235008
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.614385634347793e-05,
      "loss": 4.3503,
      "step": 235520
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6135470395967415e-05,
      "loss": 4.352,
      "step": 236032
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6127084448456895e-05,
      "loss": 4.3458,
      "step": 236544
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.611869850094637e-05,
      "loss": 4.3501,
      "step": 237056
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.611031255343585e-05,
      "loss": 4.3545,
      "step": 237568
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.610192660592533e-05,
      "loss": 4.3442,
      "step": 238080
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.609354065841481e-05,
      "loss": 4.346,
      "step": 238592
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.608515471090429e-05,
      "loss": 4.3702,
      "step": 239104
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.607676876339377e-05,
      "loss": 4.353,
      "step": 239616
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.606839919468698e-05,
      "loss": 4.3528,
      "step": 240128
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.606001324717646e-05,
      "loss": 4.3461,
      "step": 240640
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.605162729966594e-05,
      "loss": 4.3607,
      "step": 241152
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.604324135215542e-05,
      "loss": 4.3416,
      "step": 241664
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.60348554046449e-05,
      "loss": 4.354,
      "step": 242176
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.602646945713438e-05,
      "loss": 4.3408,
      "step": 242688
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.601809988842759e-05,
      "loss": 4.3409,
      "step": 243200
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.600971394091707e-05,
      "loss": 4.3309,
      "step": 243712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.600132799340655e-05,
      "loss": 4.3408,
      "step": 244224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.599294204589603e-05,
      "loss": 4.3463,
      "step": 244736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.598457247718924e-05,
      "loss": 4.3497,
      "step": 245248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.597618652967872e-05,
      "loss": 4.3457,
      "step": 245760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.596781696097193e-05,
      "loss": 4.3475,
      "step": 246272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.595943101346141e-05,
      "loss": 4.3402,
      "step": 246784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.595104506595089e-05,
      "loss": 4.3481,
      "step": 247296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.594265911844037e-05,
      "loss": 4.3299,
      "step": 247808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.593427317092985e-05,
      "loss": 4.3399,
      "step": 248320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.592588722341933e-05,
      "loss": 4.325,
      "step": 248832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.591750127590882e-05,
      "loss": 4.3395,
      "step": 249344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.59091153283983e-05,
      "loss": 4.3285,
      "step": 249856
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.590072938088778e-05,
      "loss": 4.3388,
      "step": 250368
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5892359812180986e-05,
      "loss": 4.3326,
      "step": 250880
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5883973864670466e-05,
      "loss": 4.3254,
      "step": 251392
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5875587917159946e-05,
      "loss": 4.3392,
      "step": 251904
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5867201969649426e-05,
      "loss": 4.3305,
      "step": 252416
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5858816022138906e-05,
      "loss": 4.3305,
      "step": 252928
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5850446453432115e-05,
      "loss": 4.3183,
      "step": 253440
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5842060505921595e-05,
      "loss": 4.3206,
      "step": 253952
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5833674558411075e-05,
      "loss": 4.3339,
      "step": 254464
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5825288610900555e-05,
      "loss": 4.3288,
      "step": 254976
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.581691904219377e-05,
      "loss": 4.328,
      "step": 255488
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580854947348698e-05,
      "loss": 4.315,
      "step": 256000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580016352597646e-05,
      "loss": 4.3294,
      "step": 256512
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.579177757846594e-05,
      "loss": 4.3144,
      "step": 257024
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.578339163095542e-05,
      "loss": 4.3361,
      "step": 257536
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.57750056834449e-05,
      "loss": 4.3145,
      "step": 258048
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.576661973593438e-05,
      "loss": 4.3233,
      "step": 258560
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.575823378842385e-05,
      "loss": 4.3246,
      "step": 259072
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.574984784091333e-05,
      "loss": 4.3122,
      "step": 259584
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.574146189340281e-05,
      "loss": 4.3179,
      "step": 260096
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.573307594589229e-05,
      "loss": 4.3246,
      "step": 260608
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.572468999838177e-05,
      "loss": 4.3061,
      "step": 261120
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.571630405087125e-05,
      "loss": 4.3197,
      "step": 261632
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.570791810336074e-05,
      "loss": 4.3192,
      "step": 262144
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.569953215585022e-05,
      "loss": 4.321,
      "step": 262656
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.56911462083397e-05,
      "loss": 4.3095,
      "step": 263168
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.568277663963291e-05,
      "loss": 4.3158,
      "step": 263680
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.567439069212239e-05,
      "loss": 4.3026,
      "step": 264192
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.566600474461187e-05,
      "loss": 4.3155,
      "step": 264704
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.565761879710135e-05,
      "loss": 4.3278,
      "step": 265216
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.564923284959083e-05,
      "loss": 4.3121,
      "step": 265728
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.564084690208031e-05,
      "loss": 4.3261,
      "step": 266240
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.563246095456979e-05,
      "loss": 4.3263,
      "step": 266752
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5624091385863e-05,
      "loss": 4.3228,
      "step": 267264
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.561570543835248e-05,
      "loss": 4.2997,
      "step": 267776
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.560731949084196e-05,
      "loss": 4.3204,
      "step": 268288
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.559893354333144e-05,
      "loss": 4.3179,
      "step": 268800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.559056397462465e-05,
      "loss": 4.3058,
      "step": 269312
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.558217802711413e-05,
      "loss": 4.3109,
      "step": 269824
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.557379207960361e-05,
      "loss": 4.3124,
      "step": 270336
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.556540613209309e-05,
      "loss": 4.3276,
      "step": 270848
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5557020184582566e-05,
      "loss": 4.31,
      "step": 271360
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5548634237072046e-05,
      "loss": 4.2968,
      "step": 271872
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5540248289561526e-05,
      "loss": 4.3096,
      "step": 272384
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5531862342051006e-05,
      "loss": 4.3077,
      "step": 272896
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5523492773344215e-05,
      "loss": 4.3115,
      "step": 273408
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5515106825833695e-05,
      "loss": 4.315,
      "step": 273920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.550673725712691e-05,
      "loss": 4.3046,
      "step": 274432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.549835130961639e-05,
      "loss": 4.3028,
      "step": 274944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.548996536210587e-05,
      "loss": 4.2956,
      "step": 275456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.548157941459535e-05,
      "loss": 4.3011,
      "step": 275968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.547319346708483e-05,
      "loss": 4.3031,
      "step": 276480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.546480751957431e-05,
      "loss": 4.3114,
      "step": 276992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.545642157206379e-05,
      "loss": 4.322,
      "step": 277504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.544803562455327e-05,
      "loss": 4.3043,
      "step": 278016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.543964967704275e-05,
      "loss": 4.2951,
      "step": 278528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.543128010833596e-05,
      "loss": 4.2987,
      "step": 279040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.542289416082544e-05,
      "loss": 4.2991,
      "step": 279552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.541452459211865e-05,
      "loss": 4.3088,
      "step": 280064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.540613864460813e-05,
      "loss": 4.2984,
      "step": 280576
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.539775269709761e-05,
      "loss": 4.3031,
      "step": 281088
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5389366749587095e-05,
      "loss": 4.2973,
      "step": 281600
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5380980802076575e-05,
      "loss": 4.3062,
      "step": 282112
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5372594854566055e-05,
      "loss": 4.2785,
      "step": 282624
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5364225285859264e-05,
      "loss": 4.2948,
      "step": 283136
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5355839338348744e-05,
      "loss": 4.2869,
      "step": 283648
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5347453390838224e-05,
      "loss": 4.3008,
      "step": 284160
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5339067443327704e-05,
      "loss": 4.2916,
      "step": 284672
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.533069787462091e-05,
      "loss": 4.2993,
      "step": 285184
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.532231192711039e-05,
      "loss": 4.2828,
      "step": 285696
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.531392597959987e-05,
      "loss": 4.2854,
      "step": 286208
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.530554003208935e-05,
      "loss": 4.291,
      "step": 286720
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.529715408457883e-05,
      "loss": 4.2929,
      "step": 287232
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528876813706831e-05,
      "loss": 4.3036,
      "step": 287744
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528038218955779e-05,
      "loss": 4.3007,
      "step": 288256
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.527199624204728e-05,
      "loss": 4.2951,
      "step": 288768
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.526361029453675e-05,
      "loss": 4.2882,
      "step": 289280
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.525522434702623e-05,
      "loss": 4.2908,
      "step": 289792
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.524683839951571e-05,
      "loss": 4.2953,
      "step": 290304
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.523845245200519e-05,
      "loss": 4.2848,
      "step": 290816
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.52300828832984e-05,
      "loss": 4.2917,
      "step": 291328
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.522171331459162e-05,
      "loss": 4.2876,
      "step": 291840
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.52133273670811e-05,
      "loss": 4.2811,
      "step": 292352
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.520494141957057e-05,
      "loss": 4.2809,
      "step": 292864
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.519655547206005e-05,
      "loss": 4.2827,
      "step": 293376
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.518816952454953e-05,
      "loss": 4.2761,
      "step": 293888
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.517981633464648e-05,
      "loss": 4.2879,
      "step": 294400
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.517143038713596e-05,
      "loss": 4.2717,
      "step": 294912
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.516304443962544e-05,
      "loss": 4.2961,
      "step": 295424
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.515465849211492e-05,
      "loss": 4.2949,
      "step": 295936
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.51462725446044e-05,
      "loss": 4.2958,
      "step": 296448
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5137886597093876e-05,
      "loss": 4.2718,
      "step": 296960
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5129500649583356e-05,
      "loss": 4.2863,
      "step": 297472
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5121114702072836e-05,
      "loss": 4.2915,
      "step": 297984
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5112745133366045e-05,
      "loss": 4.2878,
      "step": 298496
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5104359185855525e-05,
      "loss": 4.2823,
      "step": 299008
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5095973238345005e-05,
      "loss": 4.2817,
      "step": 299520
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5087587290834485e-05,
      "loss": 4.2801,
      "step": 300032
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.507920134332397e-05,
      "loss": 4.2871,
      "step": 300544
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.507081539581345e-05,
      "loss": 4.2778,
      "step": 301056
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.506242944830293e-05,
      "loss": 4.2834,
      "step": 301568
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.505404350079241e-05,
      "loss": 4.2802,
      "step": 302080
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.504567393208562e-05,
      "loss": 4.2705,
      "step": 302592
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.503730436337883e-05,
      "loss": 4.2855,
      "step": 303104
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.502891841586831e-05,
      "loss": 4.282,
      "step": 303616
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.502053246835779e-05,
      "loss": 4.278,
      "step": 304128
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.501214652084727e-05,
      "loss": 4.2705,
      "step": 304640
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.500376057333675e-05,
      "loss": 4.2698,
      "step": 305152
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.257569789886475,
      "eval_runtime": 294.7413,
      "eval_samples_per_second": 1294.664,
      "eval_steps_per_second": 40.459,
      "step": 305280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.499537462582623e-05,
      "loss": 4.2726,
      "step": 305664
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.498698867831571e-05,
      "loss": 4.2643,
      "step": 306176
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.497860273080519e-05,
      "loss": 4.2795,
      "step": 306688
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.497021678329467e-05,
      "loss": 4.2716,
      "step": 307200
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.496183083578415e-05,
      "loss": 4.2861,
      "step": 307712
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4953444888273636e-05,
      "loss": 4.2656,
      "step": 308224
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4945058940763116e-05,
      "loss": 4.2701,
      "step": 308736
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.493667299325259e-05,
      "loss": 4.2567,
      "step": 309248
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.492828704574207e-05,
      "loss": 4.2786,
      "step": 309760
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.491990109823155e-05,
      "loss": 4.276,
      "step": 310272
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.491151515072103e-05,
      "loss": 4.2751,
      "step": 310784
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.490312920321051e-05,
      "loss": 4.2722,
      "step": 311296
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.489474325569999e-05,
      "loss": 4.2612,
      "step": 311808
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.488635730818947e-05,
      "loss": 4.2707,
      "step": 312320
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.487797136067895e-05,
      "loss": 4.258,
      "step": 312832
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.486958541316843e-05,
      "loss": 4.2665,
      "step": 313344
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.486121584446164e-05,
      "loss": 4.2639,
      "step": 313856
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.485282989695112e-05,
      "loss": 4.2592,
      "step": 314368
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4844443949440604e-05,
      "loss": 4.2656,
      "step": 314880
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4836058001930084e-05,
      "loss": 4.2795,
      "step": 315392
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4827672054419564e-05,
      "loss": 4.2622,
      "step": 315904
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4819286106909044e-05,
      "loss": 4.2675,
      "step": 316416
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4810900159398524e-05,
      "loss": 4.2617,
      "step": 316928
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4802514211888004e-05,
      "loss": 4.2732,
      "step": 317440
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.479412826437748e-05,
      "loss": 4.2589,
      "step": 317952
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.478575869567069e-05,
      "loss": 4.2666,
      "step": 318464
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.477737274816017e-05,
      "loss": 4.2576,
      "step": 318976
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.476898680064965e-05,
      "loss": 4.2545,
      "step": 319488
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4760600853139126e-05,
      "loss": 4.249,
      "step": 320000
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.475223128443234e-05,
      "loss": 4.2568,
      "step": 320512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.474384533692182e-05,
      "loss": 4.2605,
      "step": 321024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.47354593894113e-05,
      "loss": 4.2665,
      "step": 321536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.472707344190078e-05,
      "loss": 4.2626,
      "step": 322048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4718703873194e-05,
      "loss": 4.2627,
      "step": 322560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.471031792568348e-05,
      "loss": 4.2591,
      "step": 323072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.470193197817295e-05,
      "loss": 4.2633,
      "step": 323584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.469356240946617e-05,
      "loss": 4.2534,
      "step": 324096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.468517646195565e-05,
      "loss": 4.2545,
      "step": 324608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.467679051444512e-05,
      "loss": 4.2412,
      "step": 325120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.46684045669346e-05,
      "loss": 4.2527,
      "step": 325632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.466001861942408e-05,
      "loss": 4.2504,
      "step": 326144
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.465163267191356e-05,
      "loss": 4.2579,
      "step": 326656
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.464324672440304e-05,
      "loss": 4.2528,
      "step": 327168
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4634877155696256e-05,
      "loss": 4.2431,
      "step": 327680
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4626491208185736e-05,
      "loss": 4.2595,
      "step": 328192
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4618105260675215e-05,
      "loss": 4.2496,
      "step": 328704
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4609719313164695e-05,
      "loss": 4.2486,
      "step": 329216
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4601333365654175e-05,
      "loss": 4.2402,
      "step": 329728
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4592947418143655e-05,
      "loss": 4.2367,
      "step": 330240
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4584561470633135e-05,
      "loss": 4.2522,
      "step": 330752
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4576175523122615e-05,
      "loss": 4.2487,
      "step": 331264
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4567789575612095e-05,
      "loss": 4.2476,
      "step": 331776
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4559420006905304e-05,
      "loss": 4.2385,
      "step": 332288
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4551034059394784e-05,
      "loss": 4.2506,
      "step": 332800
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4542648111884264e-05,
      "loss": 4.2359,
      "step": 333312
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4534262164373744e-05,
      "loss": 4.2544,
      "step": 333824
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.452589259566696e-05,
      "loss": 4.233,
      "step": 334336
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.451750664815644e-05,
      "loss": 4.2469,
      "step": 334848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.450913707944965e-05,
      "loss": 4.2449,
      "step": 335360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.450075113193913e-05,
      "loss": 4.2405,
      "step": 335872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.449236518442861e-05,
      "loss": 4.2387,
      "step": 336384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.448397923691809e-05,
      "loss": 4.2453,
      "step": 336896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.447559328940757e-05,
      "loss": 4.2276,
      "step": 337408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.446720734189705e-05,
      "loss": 4.2422,
      "step": 337920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.445882139438653e-05,
      "loss": 4.2443,
      "step": 338432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.445043544687601e-05,
      "loss": 4.2417,
      "step": 338944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.444208225697295e-05,
      "loss": 4.235,
      "step": 339456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.443369630946243e-05,
      "loss": 4.2372,
      "step": 339968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4425310361951914e-05,
      "loss": 4.2302,
      "step": 340480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4416924414441394e-05,
      "loss": 4.2339,
      "step": 340992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4408538466930874e-05,
      "loss": 4.2489,
      "step": 341504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.440016889822408e-05,
      "loss": 4.2395,
      "step": 342016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.439178295071356e-05,
      "loss": 4.2499,
      "step": 342528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.438339700320304e-05,
      "loss": 4.2498,
      "step": 343040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.437501105569252e-05,
      "loss": 4.2478,
      "step": 343552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.436664148698573e-05,
      "loss": 4.2273,
      "step": 344064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.435825553947521e-05,
      "loss": 4.2383,
      "step": 344576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.434986959196469e-05,
      "loss": 4.2464,
      "step": 345088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.434148364445417e-05,
      "loss": 4.231,
      "step": 345600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.433309769694365e-05,
      "loss": 4.2346,
      "step": 346112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.432471174943313e-05,
      "loss": 4.235,
      "step": 346624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.431632580192261e-05,
      "loss": 4.2536,
      "step": 347136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.430793985441209e-05,
      "loss": 4.2387,
      "step": 347648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.429957028570531e-05,
      "loss": 4.2183,
      "step": 348160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.429118433819479e-05,
      "loss": 4.2361,
      "step": 348672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.428279839068426e-05,
      "loss": 4.2345,
      "step": 349184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.427441244317374e-05,
      "loss": 4.2387,
      "step": 349696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.426602649566322e-05,
      "loss": 4.2407,
      "step": 350208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.425765692695643e-05,
      "loss": 4.2287,
      "step": 350720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.424927097944591e-05,
      "loss": 4.2288,
      "step": 351232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.424088503193539e-05,
      "loss": 4.2279,
      "step": 351744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.423249908442487e-05,
      "loss": 4.2225,
      "step": 352256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4224129515718085e-05,
      "loss": 4.2328,
      "step": 352768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.42157599470113e-05,
      "loss": 4.2358,
      "step": 353280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.420737399950078e-05,
      "loss": 4.2487,
      "step": 353792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.419898805199026e-05,
      "loss": 4.2371,
      "step": 354304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4190602104479734e-05,
      "loss": 4.221,
      "step": 354816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.418223253577295e-05,
      "loss": 4.2273,
      "step": 355328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.417384658826243e-05,
      "loss": 4.2259,
      "step": 355840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.41654606407519e-05,
      "loss": 4.2363,
      "step": 356352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.415707469324138e-05,
      "loss": 4.2302,
      "step": 356864
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.414868874573086e-05,
      "loss": 4.2309,
      "step": 357376
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.414030279822034e-05,
      "loss": 4.2241,
      "step": 357888
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.413193322951356e-05,
      "loss": 4.2347,
      "step": 358400
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.412354728200304e-05,
      "loss": 4.2088,
      "step": 358912
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.411516133449252e-05,
      "loss": 4.2269,
      "step": 359424
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4106775386982e-05,
      "loss": 4.2117,
      "step": 359936
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.409838943947148e-05,
      "loss": 4.2312,
      "step": 360448
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.409000349196096e-05,
      "loss": 4.2243,
      "step": 360960
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.408161754445044e-05,
      "loss": 4.2265,
      "step": 361472
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.407323159693992e-05,
      "loss": 4.2139,
      "step": 361984
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.406486202823313e-05,
      "loss": 4.2171,
      "step": 362496
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.405647608072261e-05,
      "loss": 4.224,
      "step": 363008
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.404810651201582e-05,
      "loss": 4.2209,
      "step": 363520
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.40397205645053e-05,
      "loss": 4.2331,
      "step": 364032
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4031334616994777e-05,
      "loss": 4.2312,
      "step": 364544
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4022948669484257e-05,
      "loss": 4.2265,
      "step": 365056
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.401456272197374e-05,
      "loss": 4.2188,
      "step": 365568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.400617677446322e-05,
      "loss": 4.22,
      "step": 366080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.399780720575643e-05,
      "loss": 4.231,
      "step": 366592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.398942125824591e-05,
      "loss": 4.2101,
      "step": 367104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.398103531073539e-05,
      "loss": 4.2286,
      "step": 367616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.397264936322487e-05,
      "loss": 4.2137,
      "step": 368128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.396426341571435e-05,
      "loss": 4.2163,
      "step": 368640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.395587746820383e-05,
      "loss": 4.2157,
      "step": 369152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.394749152069331e-05,
      "loss": 4.215,
      "step": 369664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.393912195198652e-05,
      "loss": 4.2055,
      "step": 370176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3930736004476e-05,
      "loss": 4.224,
      "step": 370688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.392235005696548e-05,
      "loss": 4.2062,
      "step": 371200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.391396410945496e-05,
      "loss": 4.2295,
      "step": 371712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.390557816194444e-05,
      "loss": 4.2229,
      "step": 372224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.389719221443392e-05,
      "loss": 4.2335,
      "step": 372736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.388882264572714e-05,
      "loss": 4.2022,
      "step": 373248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.388043669821662e-05,
      "loss": 4.2228,
      "step": 373760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.387205075070609e-05,
      "loss": 4.2234,
      "step": 374272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.386366480319557e-05,
      "loss": 4.2216,
      "step": 374784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.385527885568505e-05,
      "loss": 4.2163,
      "step": 375296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3846909286978266e-05,
      "loss": 4.2198,
      "step": 375808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.383852333946774e-05,
      "loss": 4.2131,
      "step": 376320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.383013739195722e-05,
      "loss": 4.2225,
      "step": 376832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.38217514444467e-05,
      "loss": 4.2049,
      "step": 377344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.381336549693618e-05,
      "loss": 4.2189,
      "step": 377856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.380497954942566e-05,
      "loss": 4.2134,
      "step": 378368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3796609980718875e-05,
      "loss": 4.2088,
      "step": 378880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3788224033208355e-05,
      "loss": 4.2168,
      "step": 379392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3779838085697835e-05,
      "loss": 4.2195,
      "step": 379904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3771452138187314e-05,
      "loss": 4.2121,
      "step": 380416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3763066190676794e-05,
      "loss": 4.2025,
      "step": 380928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3754696621970004e-05,
      "loss": 4.2099,
      "step": 381440
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.200775146484375,
      "eval_runtime": 294.0479,
      "eval_samples_per_second": 1297.717,
      "eval_steps_per_second": 40.555,
      "step": 381600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3746310674459483e-05,
      "loss": 4.215,
      "step": 381952
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3737924726948963e-05,
      "loss": 4.199,
      "step": 382464
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3729538779438443e-05,
      "loss": 4.213,
      "step": 382976
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.372115283192792e-05,
      "loss": 4.2107,
      "step": 383488
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.37127668844174e-05,
      "loss": 4.2199,
      "step": 384000
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.370439731571061e-05,
      "loss": 4.2059,
      "step": 384512
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.36960113682001e-05,
      "loss": 4.2023,
      "step": 385024
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.368762542068958e-05,
      "loss": 4.1938,
      "step": 385536
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.367925585198279e-05,
      "loss": 4.2115,
      "step": 386048
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.367086990447227e-05,
      "loss": 4.2161,
      "step": 386560
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.366248395696175e-05,
      "loss": 4.2101,
      "step": 387072
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.365409800945123e-05,
      "loss": 4.2093,
      "step": 387584
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.364572844074444e-05,
      "loss": 4.1983,
      "step": 388096
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.363734249323392e-05,
      "loss": 4.2079,
      "step": 388608
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.36289565457234e-05,
      "loss": 4.1994,
      "step": 389120
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.362057059821288e-05,
      "loss": 4.2001,
      "step": 389632
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.361218465070236e-05,
      "loss": 4.1977,
      "step": 390144
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.360379870319184e-05,
      "loss": 4.203,
      "step": 390656
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.359541275568132e-05,
      "loss": 4.1976,
      "step": 391168
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.358704318697453e-05,
      "loss": 4.2149,
      "step": 391680
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.357865723946401e-05,
      "loss": 4.2044,
      "step": 392192
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.357027129195349e-05,
      "loss": 4.2101,
      "step": 392704
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.35619017232467e-05,
      "loss": 4.1982,
      "step": 393216
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.355351577573618e-05,
      "loss": 4.2116,
      "step": 393728
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.354512982822566e-05,
      "loss": 4.1953,
      "step": 394240
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.353674388071514e-05,
      "loss": 4.2057,
      "step": 394752
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.352835793320462e-05,
      "loss": 4.2011,
      "step": 395264
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.35199719856941e-05,
      "loss": 4.1899,
      "step": 395776
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3511586038183575e-05,
      "loss": 4.1852,
      "step": 396288
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3503200090673055e-05,
      "loss": 4.1961,
      "step": 396800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3494814143162535e-05,
      "loss": 4.2027,
      "step": 397312
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.348642819565202e-05,
      "loss": 4.2037,
      "step": 397824
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.34780422481415e-05,
      "loss": 4.2022,
      "step": 398336
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.346965630063098e-05,
      "loss": 4.2055,
      "step": 398848
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.346127035312046e-05,
      "loss": 4.1982,
      "step": 399360
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.345290078441367e-05,
      "loss": 4.2019,
      "step": 399872
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.344451483690315e-05,
      "loss": 4.1956,
      "step": 400384
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.343612888939263e-05,
      "loss": 4.1923,
      "step": 400896
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.342774294188211e-05,
      "loss": 4.1811,
      "step": 401408
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.341935699437159e-05,
      "loss": 4.1954,
      "step": 401920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.341097104686107e-05,
      "loss": 4.1885,
      "step": 402432
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.340258509935055e-05,
      "loss": 4.1988,
      "step": 402944
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.339419915184003e-05,
      "loss": 4.1912,
      "step": 403456
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.338582958313324e-05,
      "loss": 4.1879,
      "step": 403968
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.337744363562272e-05,
      "loss": 4.2011,
      "step": 404480
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.33690576881122e-05,
      "loss": 4.1895,
      "step": 404992
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3360688119405415e-05,
      "loss": 4.1913,
      "step": 405504
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3352302171894895e-05,
      "loss": 4.1814,
      "step": 406016
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3343916224384375e-05,
      "loss": 4.1788,
      "step": 406528
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3335530276873855e-05,
      "loss": 4.1926,
      "step": 407040
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3327160708167064e-05,
      "loss": 4.191,
      "step": 407552
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3318774760656544e-05,
      "loss": 4.1878,
      "step": 408064
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.331040519194975e-05,
      "loss": 4.1811,
      "step": 408576
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.330201924443923e-05,
      "loss": 4.1928,
      "step": 409088
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.329363329692871e-05,
      "loss": 4.1729,
      "step": 409600
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.328524734941819e-05,
      "loss": 4.1981,
      "step": 410112
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.327686140190767e-05,
      "loss": 4.1751,
      "step": 410624
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.326847545439715e-05,
      "loss": 4.1884,
      "step": 411136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.326008950688664e-05,
      "loss": 4.184,
      "step": 411648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.325170355937611e-05,
      "loss": 4.1832,
      "step": 412160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.324331761186559e-05,
      "loss": 4.179,
      "step": 412672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.323494804315881e-05,
      "loss": 4.1879,
      "step": 413184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.322656209564829e-05,
      "loss": 4.1774,
      "step": 413696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.321817614813776e-05,
      "loss": 4.1811,
      "step": 414208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.320979020062724e-05,
      "loss": 4.1864,
      "step": 414720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.320140425311672e-05,
      "loss": 4.1849,
      "step": 415232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.31930183056062e-05,
      "loss": 4.1732,
      "step": 415744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.318463235809568e-05,
      "loss": 4.1862,
      "step": 416256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.317624641058516e-05,
      "loss": 4.1695,
      "step": 416768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.316786046307464e-05,
      "loss": 4.1776,
      "step": 417280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.315949089436786e-05,
      "loss": 4.1923,
      "step": 417792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.315110494685734e-05,
      "loss": 4.1852,
      "step": 418304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3142735378150546e-05,
      "loss": 4.1897,
      "step": 418816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3134349430640026e-05,
      "loss": 4.1903,
      "step": 419328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3125963483129506e-05,
      "loss": 4.1982,
      "step": 419840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3117577535618986e-05,
      "loss": 4.1757,
      "step": 420352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3109207966912195e-05,
      "loss": 4.1804,
      "step": 420864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3100822019401675e-05,
      "loss": 4.1901,
      "step": 421376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3092436071891155e-05,
      "loss": 4.1731,
      "step": 421888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3084050124380635e-05,
      "loss": 4.1824,
      "step": 422400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3075680555673844e-05,
      "loss": 4.1832,
      "step": 422912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.306729460816333e-05,
      "loss": 4.1915,
      "step": 423424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.305890866065281e-05,
      "loss": 4.1846,
      "step": 423936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.305052271314229e-05,
      "loss": 4.1691,
      "step": 424448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.304213676563177e-05,
      "loss": 4.1794,
      "step": 424960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.303375081812125e-05,
      "loss": 4.1783,
      "step": 425472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.302536487061073e-05,
      "loss": 4.1844,
      "step": 425984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.301697892310021e-05,
      "loss": 4.1822,
      "step": 426496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3008642112000885e-05,
      "loss": 4.1782,
      "step": 427008
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.300025616449036e-05,
      "loss": 4.1753,
      "step": 427520
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.299187021697984e-05,
      "loss": 4.1708,
      "step": 428032
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.298348426946932e-05,
      "loss": 4.1682,
      "step": 428544
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.29750983219588e-05,
      "loss": 4.1752,
      "step": 429056
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2966712374448285e-05,
      "loss": 4.1844,
      "step": 429568
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2958326426937765e-05,
      "loss": 4.1958,
      "step": 430080
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2949940479427244e-05,
      "loss": 4.1849,
      "step": 430592
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2941554531916724e-05,
      "loss": 4.1634,
      "step": 431104
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2933168584406204e-05,
      "loss": 4.1757,
      "step": 431616
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2924782636895684e-05,
      "loss": 4.1747,
      "step": 432128
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2916396689385164e-05,
      "loss": 4.1817,
      "step": 432640
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.290802712067837e-05,
      "loss": 4.1751,
      "step": 433152
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.289964117316785e-05,
      "loss": 4.1771,
      "step": 433664
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.289127160446106e-05,
      "loss": 4.1728,
      "step": 434176
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.288288565695054e-05,
      "loss": 4.1761,
      "step": 434688
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.287449970944002e-05,
      "loss": 4.1622,
      "step": 435200
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.28661137619295e-05,
      "loss": 4.1732,
      "step": 435712
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.285772781441898e-05,
      "loss": 4.158,
      "step": 436224
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.284934186690847e-05,
      "loss": 4.1821,
      "step": 436736
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.284095591939795e-05,
      "loss": 4.1692,
      "step": 437248
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.283256997188742e-05,
      "loss": 4.1701,
      "step": 437760
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.28241840243769e-05,
      "loss": 4.159,
      "step": 438272
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.281581445567012e-05,
      "loss": 4.1698,
      "step": 438784
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.28074285081596e-05,
      "loss": 4.169,
      "step": 439296
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.279905893945281e-05,
      "loss": 4.169,
      "step": 439808
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2790689370746016e-05,
      "loss": 4.1826,
      "step": 440320
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2782303423235496e-05,
      "loss": 4.1817,
      "step": 440832
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2773917475724976e-05,
      "loss": 4.1765,
      "step": 441344
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2765531528214456e-05,
      "loss": 4.1705,
      "step": 441856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2757145580703936e-05,
      "loss": 4.1648,
      "step": 442368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2748759633193416e-05,
      "loss": 4.1793,
      "step": 442880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.274039006448663e-05,
      "loss": 4.162,
      "step": 443392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.273200411697611e-05,
      "loss": 4.1753,
      "step": 443904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.272361816946559e-05,
      "loss": 4.1593,
      "step": 444416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.271523222195507e-05,
      "loss": 4.17,
      "step": 444928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2706846274444545e-05,
      "loss": 4.1568,
      "step": 445440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2698460326934025e-05,
      "loss": 4.1675,
      "step": 445952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2690074379423505e-05,
      "loss": 4.1599,
      "step": 446464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2681688431912985e-05,
      "loss": 4.1667,
      "step": 446976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2673302484402465e-05,
      "loss": 4.16,
      "step": 447488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2664916536891945e-05,
      "loss": 4.1758,
      "step": 448000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2656530589381425e-05,
      "loss": 4.1764,
      "step": 448512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2648144641870904e-05,
      "loss": 4.179,
      "step": 449024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.263977507316412e-05,
      "loss": 4.1526,
      "step": 449536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.26313891256536e-05,
      "loss": 4.1728,
      "step": 450048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.262300317814308e-05,
      "loss": 4.1732,
      "step": 450560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.261461723063256e-05,
      "loss": 4.1714,
      "step": 451072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.260624766192577e-05,
      "loss": 4.1637,
      "step": 451584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.259786171441525e-05,
      "loss": 4.1702,
      "step": 452096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.258947576690473e-05,
      "loss": 4.1664,
      "step": 452608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.258108981939421e-05,
      "loss": 4.1745,
      "step": 453120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.257270387188369e-05,
      "loss": 4.1558,
      "step": 453632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.25643343031769e-05,
      "loss": 4.171,
      "step": 454144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.255594835566638e-05,
      "loss": 4.1622,
      "step": 454656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.254756240815586e-05,
      "loss": 4.1618,
      "step": 455168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.253917646064534e-05,
      "loss": 4.1673,
      "step": 455680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2530806891938554e-05,
      "loss": 4.1643,
      "step": 456192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.252243732323176e-05,
      "loss": 4.1696,
      "step": 456704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.251405137572124e-05,
      "loss": 4.1523,
      "step": 457216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.250566542821072e-05,
      "loss": 4.1617,
      "step": 457728
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.159811496734619,
      "eval_runtime": 298.1859,
      "eval_samples_per_second": 1279.708,
      "eval_steps_per_second": 39.992,
      "step": 457920
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.8955476433490003e+17,
  "trial_name": null,
  "trial_params": null
}