{
  "best_metric": 4.342503547668457,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/lstm/1/checkpoints/checkpoint-228960",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 228960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8184,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5501,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0545,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9925,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.945,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.8588,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7022,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6044,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5141,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4386,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.3794,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3101,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2455,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.1837,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988259673485272e-05,
      "loss": 6.1337,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98742107873422e-05,
      "loss": 6.0701,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986582483983168e-05,
      "loss": 6.0368,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.9885,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.948,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9079,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8763,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8376,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 5.8103,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.7674,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7556,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7158,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.6974,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.657,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.641,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6204,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.5951,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.5853,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.5601,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.5426,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5344,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5123,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5029,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.458,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 5.463,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 5.4249,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 5.4319,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.4097,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 5.4072,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.377,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 5.384,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 5.3734,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 5.3663,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3577,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3286,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3204,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3207,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3066,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.301,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.278,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.274,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.255,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2773,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2302,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2461,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.2351,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2166,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948858823228813e-05,
      "loss": 5.2139,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948020228477761e-05,
      "loss": 5.2009,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.1898,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94634467685603e-05,
      "loss": 5.1832,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945506082104978e-05,
      "loss": 5.1968,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9446674873539255e-05,
      "loss": 5.1724,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943830530483247e-05,
      "loss": 5.1618,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942991935732195e-05,
      "loss": 5.1519,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942153340981143e-05,
      "loss": 5.1495,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413147462300904e-05,
      "loss": 5.1569,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940477789359412e-05,
      "loss": 5.1505,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.126,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1407,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1351,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1067,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362880913648985e-05,
      "loss": 5.1056,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354494966138465e-05,
      "loss": 5.0965,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.0877,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.0835,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.0878,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0828,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.0852,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0608,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.043,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.928744014366177e-05,
      "loss": 5.0494,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927905419615125e-05,
      "loss": 5.0504,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927066824864073e-05,
      "loss": 5.049,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92622823011302e-05,
      "loss": 5.0319,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0311,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0306,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0214,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922875488989185e-05,
      "loss": 5.0126,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 5.0092,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 4.9969,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920359704736029e-05,
      "loss": 5.0083,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919521109984978e-05,
      "loss": 4.9975,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9754,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9812,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9784,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 4.9857,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915329774110091e-05,
      "loss": 4.9668,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.9652,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9499,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.9566,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9345,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911136800354831e-05,
      "loss": 4.9463,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9332,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9094612487330996e-05,
      "loss": 4.9359,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.9245,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907784059230996e-05,
      "loss": 4.9296,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906945464479944e-05,
      "loss": 4.9151,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9195,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9108,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.9047,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.9159,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.9011,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019155338540045e-05,
      "loss": 4.8993,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010769391029525e-05,
      "loss": 4.8892,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002383443519005e-05,
      "loss": 4.8935,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994013874812214e-05,
      "loss": 4.8894,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89856279273017e-05,
      "loss": 4.8819,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897724197979118e-05,
      "loss": 4.8858,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896885603228066e-05,
      "loss": 4.8724,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8733,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.8538,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8582,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8607,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926975431139254e-05,
      "loss": 4.8507,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918589483628734e-05,
      "loss": 4.8436,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910203536118214e-05,
      "loss": 4.8651,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890181758860769e-05,
      "loss": 4.847,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889343164109717e-05,
      "loss": 4.8399,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888504569358665e-05,
      "loss": 4.8397,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.834,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.8367,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885990422985882e-05,
      "loss": 4.8303,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.8271,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8299,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8286,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8168,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.8068,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88095885447957e-05,
      "loss": 4.8182,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880120259728518e-05,
      "loss": 4.8051,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879283302857839e-05,
      "loss": 4.8073,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878444708106787e-05,
      "loss": 4.7982,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877606113355735e-05,
      "loss": 4.7976,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876767518604683e-05,
      "loss": 4.8037,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875928923853632e-05,
      "loss": 4.7901,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87509032910258e-05,
      "loss": 4.786,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.754786968231201,
      "eval_runtime": 288.6597,
      "eval_samples_per_second": 1321.941,
      "eval_steps_per_second": 41.312,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874253372231901e-05,
      "loss": 4.7731,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873414777480849e-05,
      "loss": 4.7718,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872576182729797e-05,
      "loss": 4.7983,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871737587978745e-05,
      "loss": 4.7815,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870898993227693e-05,
      "loss": 4.7726,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87006039847664e-05,
      "loss": 4.7693,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869221803725588e-05,
      "loss": 4.7701,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868383208974536e-05,
      "loss": 4.7527,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867544614223484e-05,
      "loss": 4.7624,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667076573528056e-05,
      "loss": 4.7612,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658690626017536e-05,
      "loss": 4.7634,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8650304678507016e-05,
      "loss": 4.762,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8641935109800225e-05,
      "loss": 4.7437,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633549162289705e-05,
      "loss": 4.7441,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625163214779185e-05,
      "loss": 4.733,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616777267268665e-05,
      "loss": 4.7355,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608391319758145e-05,
      "loss": 4.7408,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600005372247625e-05,
      "loss": 4.7314,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591619424737105e-05,
      "loss": 4.7363,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583233477226585e-05,
      "loss": 4.7499,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574847529716065e-05,
      "loss": 4.7328,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566477961009274e-05,
      "loss": 4.7344,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558092013498754e-05,
      "loss": 4.7191,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854970606598824e-05,
      "loss": 4.7366,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854132011847772e-05,
      "loss": 4.7156,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853295054977093e-05,
      "loss": 4.718,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852456460226041e-05,
      "loss": 4.705,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851617865474989e-05,
      "loss": 4.7135,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85078090860431e-05,
      "loss": 4.7049,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849942313853258e-05,
      "loss": 4.7011,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849103719102206e-05,
      "loss": 4.7059,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848265124351154e-05,
      "loss": 4.7102,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847426529600102e-05,
      "loss": 4.7009,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84658793484905e-05,
      "loss": 4.703,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845749340097998e-05,
      "loss": 4.6937,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844910745346946e-05,
      "loss": 4.7033,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440737884762674e-05,
      "loss": 4.675,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432351937252154e-05,
      "loss": 4.6904,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423965989741634e-05,
      "loss": 4.6668,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415580042231114e-05,
      "loss": 4.6871,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840719409472059e-05,
      "loss": 4.6712,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839880814721007e-05,
      "loss": 4.6835,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839042219969955e-05,
      "loss": 4.6615,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838203625218903e-05,
      "loss": 4.6803,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837365030467851e-05,
      "loss": 4.6777,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836526435716799e-05,
      "loss": 4.675,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835687840965747e-05,
      "loss": 4.6805,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834849246214695e-05,
      "loss": 4.6515,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834010651463643e-05,
      "loss": 4.6524,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833172056712591e-05,
      "loss": 4.6678,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832333461961539e-05,
      "loss": 4.6598,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83149650509086e-05,
      "loss": 4.6566,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830657910339808e-05,
      "loss": 4.6501,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829820953469129e-05,
      "loss": 4.6443,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828982358718077e-05,
      "loss": 4.6382,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828145401847398e-05,
      "loss": 4.6546,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827306807096346e-05,
      "loss": 4.6313,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826468212345294e-05,
      "loss": 4.6399,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825629617594242e-05,
      "loss": 4.6469,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82479102284319e-05,
      "loss": 4.6303,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823952428092138e-05,
      "loss": 4.6303,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823113833341086e-05,
      "loss": 4.6279,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822275238590034e-05,
      "loss": 4.6177,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214382817193557e-05,
      "loss": 4.6221,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8205996869683037e-05,
      "loss": 4.6382,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197610922172516e-05,
      "loss": 4.6211,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189224974661996e-05,
      "loss": 4.6147,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180839027151476e-05,
      "loss": 4.614,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817245307964095e-05,
      "loss": 4.6145,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816406713213043e-05,
      "loss": 4.6294,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155697563423645e-05,
      "loss": 4.6211,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147311615913125e-05,
      "loss": 4.611,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81389256684026e-05,
      "loss": 4.6251,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813053972089208e-05,
      "loss": 4.6229,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8122153773381565e-05,
      "loss": 4.6072,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113767825871045e-05,
      "loss": 4.6047,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105381878360525e-05,
      "loss": 4.604,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8096995930850005e-05,
      "loss": 4.6005,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088626362143214e-05,
      "loss": 4.5957,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8080240414632694e-05,
      "loss": 4.5998,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8071854467122174e-05,
      "loss": 4.6067,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063468519611654e-05,
      "loss": 4.6087,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055082572101134e-05,
      "loss": 4.5966,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046696624590614e-05,
      "loss": 4.5738,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803832705588382e-05,
      "loss": 4.5896,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80299411083733e-05,
      "loss": 4.5936,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802155516086278e-05,
      "loss": 4.5955,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801316921335226e-05,
      "loss": 4.5868,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800479964464548e-05,
      "loss": 4.5843,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799641369713496e-05,
      "loss": 4.5899,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798802774962444e-05,
      "loss": 4.5813,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797964180211392e-05,
      "loss": 4.5765,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797127223340713e-05,
      "loss": 4.583,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796290266470034e-05,
      "loss": 4.5686,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795451671718982e-05,
      "loss": 4.5871,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79461307696793e-05,
      "loss": 4.5782,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793774482216878e-05,
      "loss": 4.5585,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792935887465826e-05,
      "loss": 4.5701,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792098930595147e-05,
      "loss": 4.569,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791260335844095e-05,
      "loss": 4.5797,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790421741093043e-05,
      "loss": 4.5673,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789583146341991e-05,
      "loss": 4.5631,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788744551590939e-05,
      "loss": 4.5549,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787905956839887e-05,
      "loss": 4.5665,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787067362088835e-05,
      "loss": 4.548,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786230405218156e-05,
      "loss": 4.5558,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785391810467104e-05,
      "loss": 4.5531,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784553215716052e-05,
      "loss": 4.5542,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783714620965e-05,
      "loss": 4.5447,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782876026213948e-05,
      "loss": 4.5536,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782037431462896e-05,
      "loss": 4.5455,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781200474592217e-05,
      "loss": 4.5458,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780361879841165e-05,
      "loss": 4.5487,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779523285090114e-05,
      "loss": 4.5381,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778684690339061e-05,
      "loss": 4.5632,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777846095588009e-05,
      "loss": 4.5478,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777007500836957e-05,
      "loss": 4.5451,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776168906085905e-05,
      "loss": 4.5365,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775330311334853e-05,
      "loss": 4.543,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774491716583801e-05,
      "loss": 4.542,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773653121832749e-05,
      "loss": 4.5318,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772814527081697e-05,
      "loss": 4.5475,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771975932330645e-05,
      "loss": 4.5358,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771138975459966e-05,
      "loss": 4.5307,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7703020185892875e-05,
      "loss": 4.5256,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694634238382355e-05,
      "loss": 4.5272,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686248290871835e-05,
      "loss": 4.525,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7677862343361315e-05,
      "loss": 4.5277,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669476395850795e-05,
      "loss": 4.5143,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661090448340274e-05,
      "loss": 4.5431,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652704500829754e-05,
      "loss": 4.5322,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644318553319234e-05,
      "loss": 4.5226,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7635948984612444e-05,
      "loss": 4.5202,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7627563037101923e-05,
      "loss": 4.5157,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7619177089591403e-05,
      "loss": 4.529,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761079114208088e-05,
      "loss": 4.5234,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760240519457036e-05,
      "loss": 4.5189,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759401924705984e-05,
      "loss": 4.5249,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758563329954932e-05,
      "loss": 4.5213,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75772473520388e-05,
      "loss": 4.5196,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756887778333202e-05,
      "loss": 4.5094,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75604918358215e-05,
      "loss": 4.5231,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755210588831097e-05,
      "loss": 4.5106,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754373631960419e-05,
      "loss": 4.5135,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753535037209367e-05,
      "loss": 4.5085,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752696442458315e-05,
      "loss": 4.5097,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751857847707262e-05,
      "loss": 4.51,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75101925295621e-05,
      "loss": 4.4987,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7501839339659046e-05,
      "loss": 4.5038,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.476966381072998,
      "eval_runtime": 289.007,
      "eval_samples_per_second": 1320.352,
      "eval_steps_per_second": 41.262,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7493453392148526e-05,
      "loss": 4.4916,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748506744463801e-05,
      "loss": 4.4895,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747668149712749e-05,
      "loss": 4.5162,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746829554961697e-05,
      "loss": 4.5077,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7459909602106446e-05,
      "loss": 4.4957,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7451523654595926e-05,
      "loss": 4.494,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7443137707085406e-05,
      "loss": 4.4976,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7434751759574886e-05,
      "loss": 4.4819,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426365812064366e-05,
      "loss": 4.4996,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7417979864553846e-05,
      "loss": 4.4941,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7409593917043326e-05,
      "loss": 4.4959,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401207969532806e-05,
      "loss": 4.4974,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392838400826015e-05,
      "loss": 4.4802,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384452453315495e-05,
      "loss": 4.4842,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376066505804975e-05,
      "loss": 4.478,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736768055829446e-05,
      "loss": 4.4789,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735929461078394e-05,
      "loss": 4.483,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735090866327342e-05,
      "loss": 4.4751,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734253909456663e-05,
      "loss": 4.482,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733415314705611e-05,
      "loss": 4.5014,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732576719954559e-05,
      "loss": 4.4807,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.731738125203507e-05,
      "loss": 4.4834,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730899530452455e-05,
      "loss": 4.476,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730060935701403e-05,
      "loss": 4.4876,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729222340950351e-05,
      "loss": 4.4692,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728383746199298e-05,
      "loss": 4.4792,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.72754678932862e-05,
      "loss": 4.4634,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726708194577568e-05,
      "loss": 4.4682,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725869599826516e-05,
      "loss": 4.466,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725031005075464e-05,
      "loss": 4.4621,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724192410324412e-05,
      "loss": 4.468,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7233554534537335e-05,
      "loss": 4.4761,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722516858702681e-05,
      "loss": 4.4672,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721678263951629e-05,
      "loss": 4.4731,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720839669200577e-05,
      "loss": 4.4577,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720001074449525e-05,
      "loss": 4.4756,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719164117578846e-05,
      "loss": 4.4433,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718325522827794e-05,
      "loss": 4.4666,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717486928076742e-05,
      "loss": 4.4403,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71664833332569e-05,
      "loss": 4.4619,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7158097385746384e-05,
      "loss": 4.4489,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714972781703959e-05,
      "loss": 4.4628,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714134186952907e-05,
      "loss": 4.4434,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713295592201855e-05,
      "loss": 4.4561,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712456997450803e-05,
      "loss": 4.4566,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711618402699751e-05,
      "loss": 4.4581,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710779807948699e-05,
      "loss": 4.4588,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709941213197647e-05,
      "loss": 4.4383,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709102618446595e-05,
      "loss": 4.4343,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708265661575916e-05,
      "loss": 4.4567,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.707427066824864e-05,
      "loss": 4.4489,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706588472073812e-05,
      "loss": 4.4423,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70574987732276e-05,
      "loss": 4.4402,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704912920452082e-05,
      "loss": 4.4366,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7040759635814026e-05,
      "loss": 4.4269,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7032390067107235e-05,
      "loss": 4.4436,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7024004119596715e-05,
      "loss": 4.4251,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7015618172086195e-05,
      "loss": 4.435,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007232224575675e-05,
      "loss": 4.4452,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6998846277065155e-05,
      "loss": 4.4265,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990460329554635e-05,
      "loss": 4.4284,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982074382044115e-05,
      "loss": 4.431,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973688434533595e-05,
      "loss": 4.4159,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965302487023075e-05,
      "loss": 4.417,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956916539512555e-05,
      "loss": 4.4394,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948530592002035e-05,
      "loss": 4.4241,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694016102329525e-05,
      "loss": 4.4178,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693177507578473e-05,
      "loss": 4.4176,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.692338912827421e-05,
      "loss": 4.4227,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691500318076369e-05,
      "loss": 4.4338,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690661723325317e-05,
      "loss": 4.4301,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898231285742644e-05,
      "loss": 4.4183,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6889845338232124e-05,
      "loss": 4.4327,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881459390721604e-05,
      "loss": 4.4284,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687308982201481e-05,
      "loss": 4.4168,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686470387450429e-05,
      "loss": 4.4143,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.685631792699377e-05,
      "loss": 4.4209,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684794835828699e-05,
      "loss": 4.414,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683956241077647e-05,
      "loss": 4.4137,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683117646326595e-05,
      "loss": 4.413,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682279051575543e-05,
      "loss": 4.4203,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681440456824491e-05,
      "loss": 4.4261,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680601862073439e-05,
      "loss": 4.4115,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679763267322387e-05,
      "loss": 4.3964,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678926310451708e-05,
      "loss": 4.4063,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678087715700656e-05,
      "loss": 4.4083,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677249120949604e-05,
      "loss": 4.416,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676410526198552e-05,
      "loss": 4.4068,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755735693278726e-05,
      "loss": 4.4057,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6747349745768206e-05,
      "loss": 4.416,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673896379825769e-05,
      "loss": 4.4005,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673057785074717e-05,
      "loss": 4.4007,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672220828204038e-05,
      "loss": 4.4055,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671382233452986e-05,
      "loss": 4.3978,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670543638701934e-05,
      "loss": 4.4146,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669705043950882e-05,
      "loss": 4.4001,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66886644919983e-05,
      "loss": 4.3875,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668027854448778e-05,
      "loss": 4.3986,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667189259697726e-05,
      "loss": 4.3965,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666350664946674e-05,
      "loss": 4.4123,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665513708075995e-05,
      "loss": 4.4021,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664675113324943e-05,
      "loss": 4.3926,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663836518573891e-05,
      "loss": 4.3844,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662997923822839e-05,
      "loss": 4.3963,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662160966952161e-05,
      "loss": 4.3812,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661322372201109e-05,
      "loss": 4.3917,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6604837774500567e-05,
      "loss": 4.3866,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6596451826990047e-05,
      "loss": 4.3894,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6588065879479526e-05,
      "loss": 4.3777,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6579679931969006e-05,
      "loss": 4.3965,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657129398445848e-05,
      "loss": 4.3806,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6562924415751695e-05,
      "loss": 4.3812,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6554538468241175e-05,
      "loss": 4.389,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654615252073065e-05,
      "loss": 4.3771,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653776657322013e-05,
      "loss": 4.4001,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529380625709615e-05,
      "loss": 4.3877,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652101105700283e-05,
      "loss": 4.3878,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512625109492304e-05,
      "loss": 4.3784,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504239161981784e-05,
      "loss": 4.3837,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495853214471264e-05,
      "loss": 4.3846,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487467266960744e-05,
      "loss": 4.3729,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479081319450224e-05,
      "loss": 4.3886,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647071175074343e-05,
      "loss": 4.3812,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646232580323291e-05,
      "loss": 4.3755,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645393985572239e-05,
      "loss": 4.3751,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644555390821187e-05,
      "loss": 4.3748,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643716796070135e-05,
      "loss": 4.3652,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642879839199457e-05,
      "loss": 4.379,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642041244448405e-05,
      "loss": 4.3636,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641202649697353e-05,
      "loss": 4.3863,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640364054946301e-05,
      "loss": 4.3833,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639525460195249e-05,
      "loss": 4.37,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638686865444197e-05,
      "loss": 4.3683,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637849908573518e-05,
      "loss": 4.3696,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637011313822466e-05,
      "loss": 4.379,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636172719071414e-05,
      "loss": 4.3738,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635334124320362e-05,
      "loss": 4.3712,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63449552956931e-05,
      "loss": 4.3744,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633656934818258e-05,
      "loss": 4.3739,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632818340067206e-05,
      "loss": 4.3746,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631979745316154e-05,
      "loss": 4.362,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631142788445475e-05,
      "loss": 4.38,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630304193694423e-05,
      "loss": 4.367,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629467236823744e-05,
      "loss": 4.3644,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628628642072692e-05,
      "loss": 4.3667,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627791685202013e-05,
      "loss": 4.3679,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626953090450961e-05,
      "loss": 4.3644,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626114495699909e-05,
      "loss": 4.3539,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625275900948857e-05,
      "loss": 4.3578,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.342503547668457,
      "eval_runtime": 298.1181,
      "eval_samples_per_second": 1279.999,
      "eval_steps_per_second": 40.001,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.460672397103816e+16,
  "trial_name": null,
  "trial_params": null
}