{
  "best_metric": 3.986476421356201,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/det-adj-noun/lstm/4/checkpoints/checkpoint-1984320",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 1984320,
  "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": 10.8193,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5516,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0608,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9965,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9474,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9067,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7482,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6243,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5246,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4571,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.4095,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3562,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2988,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.2415,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1819,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.122,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.074,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.0331,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9837,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9423,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.907,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.86,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.838,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7998,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979875363855125e-05,
      "loss": 5.7797,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7431,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7159,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6918,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.97652262273129e-05,
      "loss": 5.674,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.975684027980238e-05,
      "loss": 5.6498,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974845433229186e-05,
      "loss": 5.6279,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.974008476358507e-05,
      "loss": 5.617,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.973169881607455e-05,
      "loss": 5.58,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.972331286856403e-05,
      "loss": 5.573,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714926921053515e-05,
      "loss": 5.5536,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5374,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5226,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.4817,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4874,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4534,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4561,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4345,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4366,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.4068,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.3918,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962273063484898e-05,
      "loss": 5.3965,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.961434468733847e-05,
      "loss": 5.3847,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960595873982795e-05,
      "loss": 5.3722,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3542,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.332,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3432,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95724313285896e-05,
      "loss": 5.3239,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956404538107908e-05,
      "loss": 5.3225,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955565943356856e-05,
      "loss": 5.2913,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954727348605804e-05,
      "loss": 5.2958,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953888753854752e-05,
      "loss": 5.273,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530501591037e-05,
      "loss": 5.2939,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952211564352648e-05,
      "loss": 5.258,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2597,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.2452,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2317,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948858823228813e-05,
      "loss": 5.2339,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948020228477761e-05,
      "loss": 5.2216,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9471816337267086e-05,
      "loss": 5.1945,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9463430389756566e-05,
      "loss": 5.2007,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9455044442246046e-05,
      "loss": 5.2104,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9446658494735526e-05,
      "loss": 5.1875,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9438272547225006e-05,
      "loss": 5.1765,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9429902978518215e-05,
      "loss": 5.159,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421517031007695e-05,
      "loss": 5.1617,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413131083497175e-05,
      "loss": 5.1777,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404745135986655e-05,
      "loss": 5.1614,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.939637556727987e-05,
      "loss": 5.1471,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.938798961976935e-05,
      "loss": 5.15,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937960367225883e-05,
      "loss": 5.1461,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.937121772474831e-05,
      "loss": 5.1278,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936283177723779e-05,
      "loss": 5.1132,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935444582972727e-05,
      "loss": 5.1184,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934605988221675e-05,
      "loss": 5.0984,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933769031350996e-05,
      "loss": 5.0998,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932930436599944e-05,
      "loss": 5.1044,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932091841848892e-05,
      "loss": 5.0902,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93125324709784e-05,
      "loss": 5.0931,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930416290227161e-05,
      "loss": 5.0813,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0611,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0642,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0609,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0672,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0484,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.0472,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0452,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.0372,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922872213228439e-05,
      "loss": 5.0347,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922033618477387e-05,
      "loss": 5.0237,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921195023726335e-05,
      "loss": 5.0093,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920356428975283e-05,
      "loss": 5.0178,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919517834224231e-05,
      "loss": 5.0058,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918679239473179e-05,
      "loss": 5.0017,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917840644722127e-05,
      "loss": 4.9931,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917003687851449e-05,
      "loss": 4.9923,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916165093100397e-05,
      "loss": 4.9966,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915326498349345e-05,
      "loss": 4.9846,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914487903598292e-05,
      "loss": 4.9739,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91364930884724e-05,
      "loss": 4.9696,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912810714096188e-05,
      "loss": 4.9615,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911972119345136e-05,
      "loss": 4.9481,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911135162474457e-05,
      "loss": 4.9444,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910296567723405e-05,
      "loss": 4.9483,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909457972972353e-05,
      "loss": 4.9445,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908619378221301e-05,
      "loss": 4.9323,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907780783470249e-05,
      "loss": 4.9296,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906943826599571e-05,
      "loss": 4.9301,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906105231848519e-05,
      "loss": 4.9332,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905266637097467e-05,
      "loss": 4.9281,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9044280423464147e-05,
      "loss": 4.9207,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9035910854757356e-05,
      "loss": 4.9198,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027541286050565e-05,
      "loss": 4.9169,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019155338540045e-05,
      "loss": 4.9064,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010769391029525e-05,
      "loss": 4.8998,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002383443519005e-05,
      "loss": 4.9036,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8993997496008485e-05,
      "loss": 4.9031,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985611548497965e-05,
      "loss": 4.8806,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977225600987445e-05,
      "loss": 4.8813,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968839653476924e-05,
      "loss": 4.8789,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896047008477014e-05,
      "loss": 4.871,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895208413725962e-05,
      "loss": 4.8733,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89436981897491e-05,
      "loss": 4.8763,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893531224223858e-05,
      "loss": 4.8616,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892692629472806e-05,
      "loss": 4.8619,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891854034721754e-05,
      "loss": 4.8506,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891015439970702e-05,
      "loss": 4.8671,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.89017684521965e-05,
      "loss": 4.8578,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889338250468598e-05,
      "loss": 4.8493,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888501293597919e-05,
      "loss": 4.8461,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887662698846867e-05,
      "loss": 4.8456,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886824104095815e-05,
      "loss": 4.8457,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885985509344763e-05,
      "loss": 4.8345,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8851485524740845e-05,
      "loss": 4.8332,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8843099577230325e-05,
      "loss": 4.8331,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8834713629719805e-05,
      "loss": 4.8297,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882632768220928e-05,
      "loss": 4.8287,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881794173469876e-05,
      "loss": 4.809,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880955578718824e-05,
      "loss": 4.8195,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880116983967772e-05,
      "loss": 4.8045,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879281664977466e-05,
      "loss": 4.7945,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878443070226414e-05,
      "loss": 4.8131,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877604475475362e-05,
      "loss": 4.8018,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87676588072431e-05,
      "loss": 4.8021,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875927285973258e-05,
      "loss": 4.8048,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875088691222206e-05,
      "loss": 4.7824,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.7586212158203125,
      "eval_runtime": 284.5791,
      "eval_samples_per_second": 1340.896,
      "eval_steps_per_second": 41.904,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874250096471154e-05,
      "loss": 4.7858,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873411501720102e-05,
      "loss": 4.7817,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872574544849423e-05,
      "loss": 4.7995,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871735950098371e-05,
      "loss": 4.7787,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870897355347319e-05,
      "loss": 4.7864,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870058760596267e-05,
      "loss": 4.7659,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869220165845215e-05,
      "loss": 4.7714,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868381571094163e-05,
      "loss": 4.7606,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867544614223484e-05,
      "loss": 4.7651,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866706019472432e-05,
      "loss": 4.7658,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86586742472138e-05,
      "loss": 4.7685,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865028829970329e-05,
      "loss": 4.768,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8641918730996496e-05,
      "loss": 4.7485,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633532783485976e-05,
      "loss": 4.7474,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625146835975456e-05,
      "loss": 4.7442,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616760888464936e-05,
      "loss": 4.7355,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608374940954416e-05,
      "loss": 4.7497,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8599988993443896e-05,
      "loss": 4.7337,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591603045933376e-05,
      "loss": 4.7324,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583217098422856e-05,
      "loss": 4.7522,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574847529716065e-05,
      "loss": 4.7305,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566461582205545e-05,
      "loss": 4.7309,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558075634695025e-05,
      "loss": 4.7267,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549689687184505e-05,
      "loss": 4.733,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541303739673985e-05,
      "loss": 4.7209,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532917792163465e-05,
      "loss": 4.7152,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524531844652945e-05,
      "loss": 4.7121,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8516145897142425e-05,
      "loss": 4.7203,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850779270723937e-05,
      "loss": 4.7009,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849940675972885e-05,
      "loss": 4.7072,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849103719102206e-05,
      "loss": 4.7081,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848265124351154e-05,
      "loss": 4.7029,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847426529600102e-05,
      "loss": 4.7069,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84658793484905e-05,
      "loss": 4.6969,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845750977978371e-05,
      "loss": 4.6981,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449123832273194e-05,
      "loss": 4.6981,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440737884762674e-05,
      "loss": 4.6748,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432351937252154e-05,
      "loss": 4.6861,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423965989741634e-05,
      "loss": 4.6701,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415580042231114e-05,
      "loss": 4.6805,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840719409472059e-05,
      "loss": 4.6811,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839880814721007e-05,
      "loss": 4.684,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839042219969955e-05,
      "loss": 4.6743,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838205263099276e-05,
      "loss": 4.6728,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373666683482236e-05,
      "loss": 4.68,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365280735971716e-05,
      "loss": 4.6741,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8356894788461196e-05,
      "loss": 4.6687,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834852521975441e-05,
      "loss": 4.6621,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834013927224389e-05,
      "loss": 4.6491,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833175332473337e-05,
      "loss": 4.6673,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832336737722285e-05,
      "loss": 4.6602,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831498142971233e-05,
      "loss": 4.6592,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830659548220181e-05,
      "loss": 4.6472,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829820953469129e-05,
      "loss": 4.6551,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82898399659845e-05,
      "loss": 4.639,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828145401847398e-05,
      "loss": 4.6596,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827306807096346e-05,
      "loss": 4.6403,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826468212345294e-05,
      "loss": 4.6485,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825629617594242e-05,
      "loss": 4.6394,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.824792660723563e-05,
      "loss": 4.6318,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823954065972511e-05,
      "loss": 4.6338,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82311547122146e-05,
      "loss": 4.6388,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222768764704077e-05,
      "loss": 4.6132,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214382817193557e-05,
      "loss": 4.6228,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8205996869683037e-05,
      "loss": 4.6354,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197610922172516e-05,
      "loss": 4.6222,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189224974661996e-05,
      "loss": 4.6159,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180855405955206e-05,
      "loss": 4.6117,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172485837248415e-05,
      "loss": 4.611,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164099889737895e-05,
      "loss": 4.6349,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155713942227375e-05,
      "loss": 4.6248,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147327994716854e-05,
      "loss": 4.6156,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138942047206334e-05,
      "loss": 4.6214,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8130556099695814e-05,
      "loss": 4.6215,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8122170152185294e-05,
      "loss": 4.6148,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113784204674774e-05,
      "loss": 4.5982,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105398257164254e-05,
      "loss": 4.6119,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8097012309653734e-05,
      "loss": 4.6027,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088626362143214e-05,
      "loss": 4.5993,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8080240414632694e-05,
      "loss": 4.6105,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80718708459259e-05,
      "loss": 4.601,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806348489841538e-05,
      "loss": 4.6082,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80551153297086e-05,
      "loss": 4.6023,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804672938219807e-05,
      "loss": 4.5807,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803834343468755e-05,
      "loss": 4.596,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802995748717703e-05,
      "loss": 4.5893,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802157153966652e-05,
      "loss": 4.6019,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013185592156e-05,
      "loss": 4.5912,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800479964464548e-05,
      "loss": 4.5936,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799641369713496e-05,
      "loss": 4.5934,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798802774962444e-05,
      "loss": 4.5871,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797965818091765e-05,
      "loss": 4.5908,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797127223340713e-05,
      "loss": 4.5842,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796288628589661e-05,
      "loss": 4.5716,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795451671718982e-05,
      "loss": 4.5831,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79461307696793e-05,
      "loss": 4.5824,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793774482216878e-05,
      "loss": 4.5725,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792935887465826e-05,
      "loss": 4.5788,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792097292714774e-05,
      "loss": 4.5766,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912586979637217e-05,
      "loss": 4.5869,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790421741093043e-05,
      "loss": 4.5712,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789583146341991e-05,
      "loss": 4.5642,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788744551590939e-05,
      "loss": 4.5657,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787905956839887e-05,
      "loss": 4.5637,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787067362088835e-05,
      "loss": 4.5453,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786228767337783e-05,
      "loss": 4.5597,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785391810467104e-05,
      "loss": 4.558,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784553215716052e-05,
      "loss": 4.5568,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783714620965e-05,
      "loss": 4.5518,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782876026213948e-05,
      "loss": 4.5443,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782037431462896e-05,
      "loss": 4.5504,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7811988367118434e-05,
      "loss": 4.5619,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780360241960792e-05,
      "loss": 4.555,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77952164720974e-05,
      "loss": 4.555,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778683052458688e-05,
      "loss": 4.5506,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777844457707636e-05,
      "loss": 4.5615,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777007500836957e-05,
      "loss": 4.5449,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776168906085905e-05,
      "loss": 4.5401,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775331949215226e-05,
      "loss": 4.548,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774493354464174e-05,
      "loss": 4.5506,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773654759713122e-05,
      "loss": 4.5301,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77281616496207e-05,
      "loss": 4.5322,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771977570211018e-05,
      "loss": 4.5385,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771138975459966e-05,
      "loss": 4.5306,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770300380708914e-05,
      "loss": 4.5345,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769461785957862e-05,
      "loss": 4.5406,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686231912068105e-05,
      "loss": 4.5222,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7677845964557585e-05,
      "loss": 4.5323,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669460017047065e-05,
      "loss": 4.5197,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661074069536545e-05,
      "loss": 4.5439,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652704500829754e-05,
      "loss": 4.5354,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644318553319234e-05,
      "loss": 4.5236,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7635932605808714e-05,
      "loss": 4.5264,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7627546658298194e-05,
      "loss": 4.5246,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7619160710787674e-05,
      "loss": 4.5323,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761077476327715e-05,
      "loss": 4.5264,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760238881576663e-05,
      "loss": 4.5218,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759400286825611e-05,
      "loss": 4.5206,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758563329954932e-05,
      "loss": 4.5192,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75772473520388e-05,
      "loss": 4.5218,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756887778333202e-05,
      "loss": 4.5086,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75604918358215e-05,
      "loss": 4.5199,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755210588831097e-05,
      "loss": 4.507,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754371994080045e-05,
      "loss": 4.4974,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753535037209367e-05,
      "loss": 4.5191,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752696442458315e-05,
      "loss": 4.5092,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751857847707262e-05,
      "loss": 4.5107,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75101925295621e-05,
      "loss": 4.5138,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750180658205158e-05,
      "loss": 4.4996,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.477122783660889,
      "eval_runtime": 284.0973,
      "eval_samples_per_second": 1343.17,
      "eval_steps_per_second": 41.975,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749342063454106e-05,
      "loss": 4.5034,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748503468703054e-05,
      "loss": 4.4938,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747664873952003e-05,
      "loss": 4.5195,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746826279200951e-05,
      "loss": 4.4957,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745987684449899e-05,
      "loss": 4.5104,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745149089698847e-05,
      "loss": 4.492,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744310494947795e-05,
      "loss": 4.4931,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743471900196743e-05,
      "loss": 4.4856,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742633305445691e-05,
      "loss": 4.4989,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741794710694639e-05,
      "loss": 4.4967,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740956115943586e-05,
      "loss": 4.4993,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740117521192534e-05,
      "loss": 4.502,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392805643218556e-05,
      "loss": 4.4836,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384419695708036e-05,
      "loss": 4.4876,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737603374819751e-05,
      "loss": 4.482,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736764780068699e-05,
      "loss": 4.4787,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7359278231980205e-05,
      "loss": 4.4878,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7350892284469685e-05,
      "loss": 4.4798,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7342506336959165e-05,
      "loss": 4.4774,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7334120389448645e-05,
      "loss": 4.4946,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7325734441938125e-05,
      "loss": 4.4796,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7317348494427605e-05,
      "loss": 4.4803,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7308962546917085e-05,
      "loss": 4.4793,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7300576599406565e-05,
      "loss": 4.4877,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292207030699774e-05,
      "loss": 4.4737,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7283821083189254e-05,
      "loss": 4.4718,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727545151448246e-05,
      "loss": 4.4699,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726706556697194e-05,
      "loss": 4.4722,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725867961946143e-05,
      "loss": 4.4626,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725029367195091e-05,
      "loss": 4.4676,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724190772444039e-05,
      "loss": 4.4688,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723352177692987e-05,
      "loss": 4.4666,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722513582941935e-05,
      "loss": 4.4743,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721674988190883e-05,
      "loss": 4.4613,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720838031320204e-05,
      "loss": 4.464,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719999436569152e-05,
      "loss": 4.4636,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7191608418181e-05,
      "loss": 4.4486,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718322247067048e-05,
      "loss": 4.4528,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717483652315996e-05,
      "loss": 4.4447,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716645057564944e-05,
      "loss": 4.4516,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715808100694265e-05,
      "loss": 4.4593,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714969505943213e-05,
      "loss": 4.4614,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7141309111921614e-05,
      "loss": 4.4513,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7132923164411094e-05,
      "loss": 4.451,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7124537216900574e-05,
      "loss": 4.4623,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711615126939005e-05,
      "loss": 4.4477,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710778170068326e-05,
      "loss": 4.4524,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709939575317274e-05,
      "loss": 4.4473,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709100980566222e-05,
      "loss": 4.4301,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082623858151696e-05,
      "loss": 4.4518,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074237910641176e-05,
      "loss": 4.4431,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7065851963130656e-05,
      "loss": 4.4462,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7057466015620136e-05,
      "loss": 4.4387,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7049080068109616e-05,
      "loss": 4.4435,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704071049940283e-05,
      "loss": 4.426,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703232455189231e-05,
      "loss": 4.4539,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702393860438179e-05,
      "loss": 4.4293,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701555265687127e-05,
      "loss": 4.4453,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700716670936075e-05,
      "loss": 4.4311,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699878076185023e-05,
      "loss": 4.4313,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699039481433971e-05,
      "loss": 4.4273,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698202524563292e-05,
      "loss": 4.4378,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69736392981224e-05,
      "loss": 4.4115,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696525335061188e-05,
      "loss": 4.4247,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695686740310136e-05,
      "loss": 4.4331,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694848145559084e-05,
      "loss": 4.4248,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694011188688405e-05,
      "loss": 4.4185,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693172593937353e-05,
      "loss": 4.4139,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923339991863017e-05,
      "loss": 4.416,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6914970423156226e-05,
      "loss": 4.4412,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906584475645706e-05,
      "loss": 4.4277,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898198528135186e-05,
      "loss": 4.4208,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6889812580624665e-05,
      "loss": 4.4292,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881426633114145e-05,
      "loss": 4.4301,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873040685603625e-05,
      "loss": 4.425,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864654738093105e-05,
      "loss": 4.4071,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856268790582585e-05,
      "loss": 4.4241,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684788284307206e-05,
      "loss": 4.4192,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683949689556154e-05,
      "loss": 4.4086,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683111094805102e-05,
      "loss": 4.4253,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68227250005405e-05,
      "loss": 4.4154,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6814355431833714e-05,
      "loss": 4.4202,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6805969484323194e-05,
      "loss": 4.4221,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6797583536812674e-05,
      "loss": 4.4018,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6789197589302154e-05,
      "loss": 4.4128,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6780811641791634e-05,
      "loss": 4.4072,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677245845188858e-05,
      "loss": 4.417,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676407250437806e-05,
      "loss": 4.4162,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675568655686753e-05,
      "loss": 4.4116,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674730060935701e-05,
      "loss": 4.414,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673891466184649e-05,
      "loss": 4.41,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673052871433597e-05,
      "loss": 4.4087,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672214276682545e-05,
      "loss": 4.404,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671377319811867e-05,
      "loss": 4.4042,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670538725060815e-05,
      "loss": 4.4092,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669700130309763e-05,
      "loss": 4.4033,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668861535558711e-05,
      "loss": 4.3992,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668022940807659e-05,
      "loss": 4.407,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66718598393698e-05,
      "loss": 4.4057,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666347389185928e-05,
      "loss": 4.418,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665508794434876e-05,
      "loss": 4.3983,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664670199683824e-05,
      "loss": 4.3957,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663831604932772e-05,
      "loss": 4.3978,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6629930101817197e-05,
      "loss": 4.3955,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6621544154306677e-05,
      "loss": 4.3766,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661317458559989e-05,
      "loss": 4.3916,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660478863808937e-05,
      "loss": 4.3944,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659640269057885e-05,
      "loss": 4.3881,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658801674306833e-05,
      "loss": 4.3926,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657964717436154e-05,
      "loss": 4.3813,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657126122685102e-05,
      "loss": 4.3825,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65628752793405e-05,
      "loss": 4.4015,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655448933182998e-05,
      "loss": 4.3884,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654610338431946e-05,
      "loss": 4.397,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653773381561267e-05,
      "loss": 4.3886,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652934786810215e-05,
      "loss": 4.402,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652096192059163e-05,
      "loss": 4.387,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.651257597308111e-05,
      "loss": 4.3835,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504206404374326e-05,
      "loss": 4.3867,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495820456863806e-05,
      "loss": 4.3922,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487434509353286e-05,
      "loss": 4.3747,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479048561842766e-05,
      "loss": 4.3755,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6470662614332246e-05,
      "loss": 4.3776,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6462293045625455e-05,
      "loss": 4.3811,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6453907098114935e-05,
      "loss": 4.3804,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6445521150604415e-05,
      "loss": 4.3877,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6437151581897624e-05,
      "loss": 4.3674,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6428765634387104e-05,
      "loss": 4.3772,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6420379686876584e-05,
      "loss": 4.3676,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6411993739366064e-05,
      "loss": 4.3908,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6403607791855544e-05,
      "loss": 4.3882,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6395221844345024e-05,
      "loss": 4.3741,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6386835896834504e-05,
      "loss": 4.3715,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6378449949323984e-05,
      "loss": 4.3762,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6370064001813464e-05,
      "loss": 4.3806,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6361678054302944e-05,
      "loss": 4.3811,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6353292106792424e-05,
      "loss": 4.3708,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6344906159281903e-05,
      "loss": 4.3737,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633653659057511e-05,
      "loss": 4.3724,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632815064306459e-05,
      "loss": 4.378,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631976469555407e-05,
      "loss": 4.363,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631137874804355e-05,
      "loss": 4.3717,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630299280053303e-05,
      "loss": 4.366,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629462323182625e-05,
      "loss": 4.3498,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628623728431573e-05,
      "loss": 4.3764,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627785133680521e-05,
      "loss": 4.3654,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626946538929469e-05,
      "loss": 4.368,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.62610958205879e-05,
      "loss": 4.372,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625270987307738e-05,
      "loss": 4.3559,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.342620849609375,
      "eval_runtime": 328.6628,
      "eval_samples_per_second": 1161.041,
      "eval_steps_per_second": 36.283,
      "step": 228960
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6244340304370586e-05,
      "loss": 4.3637,
      "step": 229376
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6235954356860066e-05,
      "loss": 4.3516,
      "step": 229888
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6227568409349546e-05,
      "loss": 4.378,
      "step": 230400
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6219182461839026e-05,
      "loss": 4.3557,
      "step": 230912
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6210796514328506e-05,
      "loss": 4.3777,
      "step": 231424
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6202410566817986e-05,
      "loss": 4.3486,
      "step": 231936
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6194024619307466e-05,
      "loss": 4.3571,
      "step": 232448
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6185638671796946e-05,
      "loss": 4.3431,
      "step": 232960
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.617726910309016e-05,
      "loss": 4.364,
      "step": 233472
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.616888315557964e-05,
      "loss": 4.3589,
      "step": 233984
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.616049720806912e-05,
      "loss": 4.3591,
      "step": 234496
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.61521112605586e-05,
      "loss": 4.3727,
      "step": 235008
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.614374169185181e-05,
      "loss": 4.3465,
      "step": 235520
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.613535574434129e-05,
      "loss": 4.3538,
      "step": 236032
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.612696979683077e-05,
      "loss": 4.3477,
      "step": 236544
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.611858384932025e-05,
      "loss": 4.3456,
      "step": 237056
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6110197901809724e-05,
      "loss": 4.3508,
      "step": 237568
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6101811954299204e-05,
      "loss": 4.3495,
      "step": 238080
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6093426006788684e-05,
      "loss": 4.3448,
      "step": 238592
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.608504005927817e-05,
      "loss": 4.3603,
      "step": 239104
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.607667049057138e-05,
      "loss": 4.3495,
      "step": 239616
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.606828454306086e-05,
      "loss": 4.3477,
      "step": 240128
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.605989859555034e-05,
      "loss": 4.3514,
      "step": 240640
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.605151264803982e-05,
      "loss": 4.3551,
      "step": 241152
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.60431267005293e-05,
      "loss": 4.3433,
      "step": 241664
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.603475713182251e-05,
      "loss": 4.3427,
      "step": 242176
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.602637118431199e-05,
      "loss": 4.3431,
      "step": 242688
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.601798523680147e-05,
      "loss": 4.3384,
      "step": 243200
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.600959928929095e-05,
      "loss": 4.3377,
      "step": 243712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.600121334178043e-05,
      "loss": 4.337,
      "step": 244224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.599282739426991e-05,
      "loss": 4.3473,
      "step": 244736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.598444144675939e-05,
      "loss": 4.3387,
      "step": 245248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.597605549924887e-05,
      "loss": 4.3484,
      "step": 245760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.596766955173835e-05,
      "loss": 4.3374,
      "step": 246272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5959299983031564e-05,
      "loss": 4.3368,
      "step": 246784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.595093041432477e-05,
      "loss": 4.3366,
      "step": 247296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.594254446681425e-05,
      "loss": 4.3289,
      "step": 247808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.593415851930373e-05,
      "loss": 4.3267,
      "step": 248320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.592577257179321e-05,
      "loss": 4.3212,
      "step": 248832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.591740300308642e-05,
      "loss": 4.3275,
      "step": 249344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.59090170555759e-05,
      "loss": 4.3352,
      "step": 249856
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.590063110806538e-05,
      "loss": 4.3358,
      "step": 250368
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.589224516055486e-05,
      "loss": 4.3318,
      "step": 250880
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.588385921304434e-05,
      "loss": 4.3309,
      "step": 251392
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.587547326553382e-05,
      "loss": 4.3412,
      "step": 251904
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.58670873180233e-05,
      "loss": 4.327,
      "step": 252416
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.585870137051279e-05,
      "loss": 4.331,
      "step": 252928
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.585031542300227e-05,
      "loss": 4.3247,
      "step": 253440
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.584194585429548e-05,
      "loss": 4.3122,
      "step": 253952
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.583355990678496e-05,
      "loss": 4.3368,
      "step": 254464
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.582517395927444e-05,
      "loss": 4.3211,
      "step": 254976
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.581678801176391e-05,
      "loss": 4.3279,
      "step": 255488
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580840206425339e-05,
      "loss": 4.3179,
      "step": 256000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580001611674287e-05,
      "loss": 4.3295,
      "step": 256512
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.579163016923235e-05,
      "loss": 4.3061,
      "step": 257024
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.578324422172183e-05,
      "loss": 4.3323,
      "step": 257536
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.577487465301504e-05,
      "loss": 4.3158,
      "step": 258048
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5766488705504526e-05,
      "loss": 4.3255,
      "step": 258560
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5758102757994006e-05,
      "loss": 4.3204,
      "step": 259072
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5749716810483486e-05,
      "loss": 4.3113,
      "step": 259584
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5741347241776695e-05,
      "loss": 4.3115,
      "step": 260096
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.573297767306991e-05,
      "loss": 4.3288,
      "step": 260608
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5724591725559384e-05,
      "loss": 4.2992,
      "step": 261120
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5716205778048864e-05,
      "loss": 4.3043,
      "step": 261632
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5707819830538344e-05,
      "loss": 4.3198,
      "step": 262144
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5699433883027824e-05,
      "loss": 4.3128,
      "step": 262656
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5691047935517304e-05,
      "loss": 4.3036,
      "step": 263168
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.568267836681051e-05,
      "loss": 4.3055,
      "step": 263680
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.567429241929999e-05,
      "loss": 4.3026,
      "step": 264192
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.566590647178948e-05,
      "loss": 4.3241,
      "step": 264704
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.565752052427896e-05,
      "loss": 4.3143,
      "step": 265216
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.564915095557217e-05,
      "loss": 4.3091,
      "step": 265728
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.564076500806165e-05,
      "loss": 4.3166,
      "step": 266240
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.563237906055113e-05,
      "loss": 4.3194,
      "step": 266752
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.562399311304061e-05,
      "loss": 4.3181,
      "step": 267264
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.561560716553009e-05,
      "loss": 4.2971,
      "step": 267776
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.56072375968233e-05,
      "loss": 4.3113,
      "step": 268288
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.559886802811651e-05,
      "loss": 4.3075,
      "step": 268800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.559048208060599e-05,
      "loss": 4.2986,
      "step": 269312
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.558209613309547e-05,
      "loss": 4.3192,
      "step": 269824
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.557371018558495e-05,
      "loss": 4.3041,
      "step": 270336
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5565324238074434e-05,
      "loss": 4.3127,
      "step": 270848
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5556938290563914e-05,
      "loss": 4.3135,
      "step": 271360
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5548552343053394e-05,
      "loss": 4.2968,
      "step": 271872
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5540166395542874e-05,
      "loss": 4.3043,
      "step": 272384
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.553179682683608e-05,
      "loss": 4.3007,
      "step": 272896
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.552341087932556e-05,
      "loss": 4.306,
      "step": 273408
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.551502493181504e-05,
      "loss": 4.307,
      "step": 273920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.550663898430452e-05,
      "loss": 4.3039,
      "step": 274432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5498253036794e-05,
      "loss": 4.3074,
      "step": 274944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.548986708928348e-05,
      "loss": 4.3083,
      "step": 275456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.548149752057669e-05,
      "loss": 4.3017,
      "step": 275968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.547311157306617e-05,
      "loss": 4.3012,
      "step": 276480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.546472562555565e-05,
      "loss": 4.2992,
      "step": 276992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.545633967804513e-05,
      "loss": 4.3078,
      "step": 277504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.544795373053462e-05,
      "loss": 4.2985,
      "step": 278016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.54395677830241e-05,
      "loss": 4.2972,
      "step": 278528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.543118183551358e-05,
      "loss": 4.3025,
      "step": 279040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.542279588800305e-05,
      "loss": 4.2975,
      "step": 279552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5414442698099996e-05,
      "loss": 4.3149,
      "step": 280064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5406056750589476e-05,
      "loss": 4.2963,
      "step": 280576
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5397670803078956e-05,
      "loss": 4.2964,
      "step": 281088
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5389284855568436e-05,
      "loss": 4.2931,
      "step": 281600
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5380915286861645e-05,
      "loss": 4.2962,
      "step": 282112
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5372529339351125e-05,
      "loss": 4.2754,
      "step": 282624
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5364143391840605e-05,
      "loss": 4.2877,
      "step": 283136
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5355757444330085e-05,
      "loss": 4.2944,
      "step": 283648
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.534737149681957e-05,
      "loss": 4.2899,
      "step": 284160
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.533900192811278e-05,
      "loss": 4.2875,
      "step": 284672
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.533061598060226e-05,
      "loss": 4.2821,
      "step": 285184
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.532223003309174e-05,
      "loss": 4.2831,
      "step": 285696
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.531384408558122e-05,
      "loss": 4.3043,
      "step": 286208
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5305458138070694e-05,
      "loss": 4.285,
      "step": 286720
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.529708856936391e-05,
      "loss": 4.2961,
      "step": 287232
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528870262185339e-05,
      "loss": 4.2903,
      "step": 287744
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528031667434287e-05,
      "loss": 4.3052,
      "step": 288256
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.527193072683234e-05,
      "loss": 4.2912,
      "step": 288768
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.526354477932182e-05,
      "loss": 4.2797,
      "step": 289280
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.525515883181131e-05,
      "loss": 4.2912,
      "step": 289792
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.524677288430079e-05,
      "loss": 4.2951,
      "step": 290304
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.523838693679027e-05,
      "loss": 4.2758,
      "step": 290816
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.523001736808348e-05,
      "loss": 4.2784,
      "step": 291328
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.522163142057296e-05,
      "loss": 4.2834,
      "step": 291840
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5213278230669904e-05,
      "loss": 4.2845,
      "step": 292352
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5204892283159384e-05,
      "loss": 4.2796,
      "step": 292864
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5196506335648863e-05,
      "loss": 4.2964,
      "step": 293376
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5188120388138343e-05,
      "loss": 4.2689,
      "step": 293888
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5179734440627817e-05,
      "loss": 4.2831,
      "step": 294400
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5171348493117297e-05,
      "loss": 4.2709,
      "step": 294912
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5162962545606777e-05,
      "loss": 4.2951,
      "step": 295424
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5154576598096256e-05,
      "loss": 4.2923,
      "step": 295936
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.514620702938947e-05,
      "loss": 4.282,
      "step": 296448
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.513782108187895e-05,
      "loss": 4.2733,
      "step": 296960
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.512943513436843e-05,
      "loss": 4.2861,
      "step": 297472
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.512104918685791e-05,
      "loss": 4.2863,
      "step": 297984
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.511267961815112e-05,
      "loss": 4.2851,
      "step": 298496
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.51042936706406e-05,
      "loss": 4.281,
      "step": 299008
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.509590772313008e-05,
      "loss": 4.2795,
      "step": 299520
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.508752177561956e-05,
      "loss": 4.2799,
      "step": 300032
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5079168585716506e-05,
      "loss": 4.2888,
      "step": 300544
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5070782638205986e-05,
      "loss": 4.2702,
      "step": 301056
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5062396690695466e-05,
      "loss": 4.2753,
      "step": 301568
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5054010743184946e-05,
      "loss": 4.2727,
      "step": 302080
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5045624795674426e-05,
      "loss": 4.2644,
      "step": 302592
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.503725522696764e-05,
      "loss": 4.2813,
      "step": 303104
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5028869279457115e-05,
      "loss": 4.2736,
      "step": 303616
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5020483331946595e-05,
      "loss": 4.276,
      "step": 304128
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5012097384436075e-05,
      "loss": 4.2774,
      "step": 304640
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5003711436925555e-05,
      "loss": 4.2716,
      "step": 305152
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.259642124176025,
      "eval_runtime": 329.0937,
      "eval_samples_per_second": 1159.521,
      "eval_steps_per_second": 36.236,
      "step": 305280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4995325489415035e-05,
      "loss": 4.2743,
      "step": 305664
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4986939541904515e-05,
      "loss": 4.2592,
      "step": 306176
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4978553594393995e-05,
      "loss": 4.2854,
      "step": 306688
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4970167646883475e-05,
      "loss": 4.2676,
      "step": 307200
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4961781699372955e-05,
      "loss": 4.2856,
      "step": 307712
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4953395751862435e-05,
      "loss": 4.2675,
      "step": 308224
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4945009804351915e-05,
      "loss": 4.2645,
      "step": 308736
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4936623856841395e-05,
      "loss": 4.2576,
      "step": 309248
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.492823790933088e-05,
      "loss": 4.2703,
      "step": 309760
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4919851961820354e-05,
      "loss": 4.2739,
      "step": 310272
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4911466014309834e-05,
      "loss": 4.2716,
      "step": 310784
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4903080066799314e-05,
      "loss": 4.2789,
      "step": 311296
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.489471049809253e-05,
      "loss": 4.261,
      "step": 311808
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4886324550582003e-05,
      "loss": 4.2647,
      "step": 312320
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4877938603071483e-05,
      "loss": 4.26,
      "step": 312832
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.486955265556096e-05,
      "loss": 4.254,
      "step": 313344
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.486116670805044e-05,
      "loss": 4.2663,
      "step": 313856
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.485278076053992e-05,
      "loss": 4.2651,
      "step": 314368
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.48443948130294e-05,
      "loss": 4.2581,
      "step": 314880
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.483600886551888e-05,
      "loss": 4.2715,
      "step": 315392
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.48276392968121e-05,
      "loss": 4.2654,
      "step": 315904
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.481925334930158e-05,
      "loss": 4.2642,
      "step": 316416
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.481086740179106e-05,
      "loss": 4.2624,
      "step": 316928
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.480248145428054e-05,
      "loss": 4.2702,
      "step": 317440
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.479409550677002e-05,
      "loss": 4.2553,
      "step": 317952
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.47857095592595e-05,
      "loss": 4.2649,
      "step": 318464
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.477732361174898e-05,
      "loss": 4.2518,
      "step": 318976
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.476893766423846e-05,
      "loss": 4.2563,
      "step": 319488
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.476056809553167e-05,
      "loss": 4.2545,
      "step": 320000
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.475219852682488e-05,
      "loss": 4.2552,
      "step": 320512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4743828958118086e-05,
      "loss": 4.2586,
      "step": 321024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.473544301060757e-05,
      "loss": 4.2581,
      "step": 321536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.472705706309705e-05,
      "loss": 4.2619,
      "step": 322048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.471867111558653e-05,
      "loss": 4.2554,
      "step": 322560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.471028516807601e-05,
      "loss": 4.2494,
      "step": 323072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.470189922056549e-05,
      "loss": 4.2566,
      "step": 323584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.469351327305497e-05,
      "loss": 4.2471,
      "step": 324096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.468512732554445e-05,
      "loss": 4.2418,
      "step": 324608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.467674137803393e-05,
      "loss": 4.2419,
      "step": 325120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.466835543052341e-05,
      "loss": 4.2428,
      "step": 325632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.465996948301289e-05,
      "loss": 4.2544,
      "step": 326144
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4651583535502366e-05,
      "loss": 4.2563,
      "step": 326656
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.464323034559931e-05,
      "loss": 4.2503,
      "step": 327168
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.463484439808879e-05,
      "loss": 4.2469,
      "step": 327680
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.462645845057827e-05,
      "loss": 4.2604,
      "step": 328192
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.461807250306775e-05,
      "loss": 4.2491,
      "step": 328704
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.460968655555724e-05,
      "loss": 4.2494,
      "step": 329216
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.460130060804672e-05,
      "loss": 4.2433,
      "step": 329728
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.459291466053619e-05,
      "loss": 4.2309,
      "step": 330240
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.458452871302567e-05,
      "loss": 4.2574,
      "step": 330752
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4576159144318886e-05,
      "loss": 4.2408,
      "step": 331264
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4567773196808366e-05,
      "loss": 4.249,
      "step": 331776
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.455938724929784e-05,
      "loss": 4.2422,
      "step": 332288
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.455100130178732e-05,
      "loss": 4.2414,
      "step": 332800
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.45426153542768e-05,
      "loss": 4.2338,
      "step": 333312
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.453422940676628e-05,
      "loss": 4.2464,
      "step": 333824
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.452584345925576e-05,
      "loss": 4.2381,
      "step": 334336
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4517473890548975e-05,
      "loss": 4.2506,
      "step": 334848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4509087943038455e-05,
      "loss": 4.2421,
      "step": 335360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4500701995527935e-05,
      "loss": 4.2348,
      "step": 335872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4492316048017415e-05,
      "loss": 4.2306,
      "step": 336384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4483946479310624e-05,
      "loss": 4.2501,
      "step": 336896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4475560531800104e-05,
      "loss": 4.2249,
      "step": 337408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4467174584289584e-05,
      "loss": 4.227,
      "step": 337920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4458788636779064e-05,
      "loss": 4.2386,
      "step": 338432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.445041906807227e-05,
      "loss": 4.24,
      "step": 338944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.444203312056175e-05,
      "loss": 4.228,
      "step": 339456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.443364717305123e-05,
      "loss": 4.2267,
      "step": 339968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.442526122554071e-05,
      "loss": 4.225,
      "step": 340480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.441689165683393e-05,
      "loss": 4.244,
      "step": 340992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.440850570932341e-05,
      "loss": 4.2408,
      "step": 341504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.440011976181289e-05,
      "loss": 4.2361,
      "step": 342016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.439173381430237e-05,
      "loss": 4.2391,
      "step": 342528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.438334786679185e-05,
      "loss": 4.2414,
      "step": 343040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.437496191928133e-05,
      "loss": 4.2461,
      "step": 343552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.436659235057454e-05,
      "loss": 4.2254,
      "step": 344064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.435820640306402e-05,
      "loss": 4.2319,
      "step": 344576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.43498204555535e-05,
      "loss": 4.2343,
      "step": 345088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.434143450804298e-05,
      "loss": 4.2213,
      "step": 345600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.433304856053246e-05,
      "loss": 4.2467,
      "step": 346112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.432466261302194e-05,
      "loss": 4.2288,
      "step": 346624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.431627666551142e-05,
      "loss": 4.2375,
      "step": 347136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.43078907180009e-05,
      "loss": 4.2457,
      "step": 347648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.429952114929411e-05,
      "loss": 4.2228,
      "step": 348160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.429113520178359e-05,
      "loss": 4.2289,
      "step": 348672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.42827656330768e-05,
      "loss": 4.2248,
      "step": 349184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.427437968556628e-05,
      "loss": 4.2305,
      "step": 349696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.426599373805576e-05,
      "loss": 4.2333,
      "step": 350208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.425762416934897e-05,
      "loss": 4.2319,
      "step": 350720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.424923822183845e-05,
      "loss": 4.2344,
      "step": 351232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.424085227432793e-05,
      "loss": 4.2344,
      "step": 351744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.423246632681741e-05,
      "loss": 4.225,
      "step": 352256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.422408037930689e-05,
      "loss": 4.2303,
      "step": 352768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.421569443179637e-05,
      "loss": 4.229,
      "step": 353280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.420730848428585e-05,
      "loss": 4.2361,
      "step": 353792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.419892253677533e-05,
      "loss": 4.2275,
      "step": 354304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.419053658926481e-05,
      "loss": 4.2232,
      "step": 354816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.418215064175429e-05,
      "loss": 4.2306,
      "step": 355328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.417376469424377e-05,
      "loss": 4.2252,
      "step": 355840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.416537874673325e-05,
      "loss": 4.2457,
      "step": 356352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.415700917802646e-05,
      "loss": 4.2259,
      "step": 356864
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.414862323051594e-05,
      "loss": 4.2248,
      "step": 357376
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.414023728300542e-05,
      "loss": 4.2211,
      "step": 357888
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.41318513354949e-05,
      "loss": 4.2282,
      "step": 358400
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.412348176678811e-05,
      "loss": 4.2034,
      "step": 358912
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.411509581927759e-05,
      "loss": 4.2148,
      "step": 359424
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.410670987176707e-05,
      "loss": 4.2236,
      "step": 359936
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.409832392425655e-05,
      "loss": 4.2239,
      "step": 360448
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.408993797674603e-05,
      "loss": 4.2195,
      "step": 360960
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4081552029235515e-05,
      "loss": 4.2114,
      "step": 361472
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4073166081724995e-05,
      "loss": 4.2148,
      "step": 361984
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4064780134214475e-05,
      "loss": 4.2312,
      "step": 362496
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4056394186703955e-05,
      "loss": 4.2217,
      "step": 363008
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.404804099680089e-05,
      "loss": 4.2239,
      "step": 363520
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.403965504929037e-05,
      "loss": 4.2232,
      "step": 364032
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.403126910177985e-05,
      "loss": 4.2332,
      "step": 364544
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.402288315426933e-05,
      "loss": 4.2247,
      "step": 365056
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.401449720675881e-05,
      "loss": 4.211,
      "step": 365568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.400611125924829e-05,
      "loss": 4.2195,
      "step": 366080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.399772531173777e-05,
      "loss": 4.225,
      "step": 366592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.398933936422725e-05,
      "loss": 4.209,
      "step": 367104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.39809861743242e-05,
      "loss": 4.2122,
      "step": 367616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.397260022681368e-05,
      "loss": 4.2125,
      "step": 368128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.396421427930316e-05,
      "loss": 4.2181,
      "step": 368640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.395582833179264e-05,
      "loss": 4.2138,
      "step": 369152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.394744238428212e-05,
      "loss": 4.2268,
      "step": 369664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.393907281557533e-05,
      "loss": 4.2003,
      "step": 370176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.393068686806481e-05,
      "loss": 4.2161,
      "step": 370688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.392230092055429e-05,
      "loss": 4.2026,
      "step": 371200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.391391497304377e-05,
      "loss": 4.2317,
      "step": 371712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.390552902553325e-05,
      "loss": 4.2166,
      "step": 372224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.389714307802273e-05,
      "loss": 4.2211,
      "step": 372736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3888773509315936e-05,
      "loss": 4.2044,
      "step": 373248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.388038756180542e-05,
      "loss": 4.2213,
      "step": 373760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.38720016142949e-05,
      "loss": 4.2264,
      "step": 374272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.386361566678438e-05,
      "loss": 4.2153,
      "step": 374784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.385522971927386e-05,
      "loss": 4.2118,
      "step": 375296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3846843771763336e-05,
      "loss": 4.2199,
      "step": 375808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.383847420305655e-05,
      "loss": 4.2114,
      "step": 376320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.383008825554603e-05,
      "loss": 4.2221,
      "step": 376832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3821702308035505e-05,
      "loss": 4.2019,
      "step": 377344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3813316360524985e-05,
      "loss": 4.2113,
      "step": 377856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3804930413014465e-05,
      "loss": 4.2061,
      "step": 378368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.379656084430768e-05,
      "loss": 4.2012,
      "step": 378880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.378817489679716e-05,
      "loss": 4.2141,
      "step": 379392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3779805328090376e-05,
      "loss": 4.209,
      "step": 379904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3771419380579856e-05,
      "loss": 4.2125,
      "step": 380416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3763033433069336e-05,
      "loss": 4.2124,
      "step": 380928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.375464748555881e-05,
      "loss": 4.2095,
      "step": 381440
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.202969551086426,
      "eval_runtime": 294.3955,
      "eval_samples_per_second": 1296.185,
      "eval_steps_per_second": 40.507,
      "step": 381600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.374626153804829e-05,
      "loss": 4.2062,
      "step": 381952
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.373787559053777e-05,
      "loss": 4.1931,
      "step": 382464
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.372948964302725e-05,
      "loss": 4.2216,
      "step": 382976
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.372112007432046e-05,
      "loss": 4.2062,
      "step": 383488
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.371273412680994e-05,
      "loss": 4.2176,
      "step": 384000
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.370434817929942e-05,
      "loss": 4.2059,
      "step": 384512
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.36959622317889e-05,
      "loss": 4.1969,
      "step": 385024
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.368757628427838e-05,
      "loss": 4.1945,
      "step": 385536
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.367919033676786e-05,
      "loss": 4.2054,
      "step": 386048
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3670804389257345e-05,
      "loss": 4.2114,
      "step": 386560
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3662418441746825e-05,
      "loss": 4.2082,
      "step": 387072
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3654048873040034e-05,
      "loss": 4.2182,
      "step": 387584
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.364567930433324e-05,
      "loss": 4.1985,
      "step": 388096
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.363729335682272e-05,
      "loss": 4.2045,
      "step": 388608
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.36289074093122e-05,
      "loss": 4.1994,
      "step": 389120
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.362052146180168e-05,
      "loss": 4.1904,
      "step": 389632
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.361213551429116e-05,
      "loss": 4.2034,
      "step": 390144
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.360374956678064e-05,
      "loss": 4.2008,
      "step": 390656
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.359536361927012e-05,
      "loss": 4.1987,
      "step": 391168
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.35869776717596e-05,
      "loss": 4.2089,
      "step": 391680
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.357859172424908e-05,
      "loss": 4.2085,
      "step": 392192
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.357020577673856e-05,
      "loss": 4.2034,
      "step": 392704
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.356181982922804e-05,
      "loss": 4.1991,
      "step": 393216
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.355343388171752e-05,
      "loss": 4.2068,
      "step": 393728
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.354506431301074e-05,
      "loss": 4.1962,
      "step": 394240
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.353667836550022e-05,
      "loss": 4.2014,
      "step": 394752
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.352829241798969e-05,
      "loss": 4.1924,
      "step": 395264
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.351990647047917e-05,
      "loss": 4.1951,
      "step": 395776
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.351153690177239e-05,
      "loss": 4.1974,
      "step": 396288
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.350315095426187e-05,
      "loss": 4.1907,
      "step": 396800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3494781385555076e-05,
      "loss": 4.1982,
      "step": 397312
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3486395438044556e-05,
      "loss": 4.199,
      "step": 397824
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3478009490534036e-05,
      "loss": 4.1996,
      "step": 398336
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3469623543023516e-05,
      "loss": 4.1946,
      "step": 398848
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3461237595512996e-05,
      "loss": 4.1869,
      "step": 399360
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.345286802680621e-05,
      "loss": 4.2016,
      "step": 399872
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.344448207929569e-05,
      "loss": 4.1898,
      "step": 400384
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3436096131785165e-05,
      "loss": 4.1754,
      "step": 400896
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3427710184274645e-05,
      "loss": 4.1871,
      "step": 401408
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.341934061556786e-05,
      "loss": 4.1821,
      "step": 401920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.341095466805734e-05,
      "loss": 4.1922,
      "step": 402432
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3402568720546814e-05,
      "loss": 4.1983,
      "step": 402944
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3394182773036294e-05,
      "loss": 4.1895,
      "step": 403456
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3385796825525774e-05,
      "loss": 4.1928,
      "step": 403968
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3377410878015254e-05,
      "loss": 4.1977,
      "step": 404480
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3369024930504734e-05,
      "loss": 4.1861,
      "step": 404992
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3360638982994214e-05,
      "loss": 4.1927,
      "step": 405504
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.335226941428743e-05,
      "loss": 4.1866,
      "step": 406016
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.334389984558064e-05,
      "loss": 4.1681,
      "step": 406528
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.333551389807012e-05,
      "loss": 4.2004,
      "step": 407040
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.33271279505596e-05,
      "loss": 4.1835,
      "step": 407552
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.331874200304908e-05,
      "loss": 4.1962,
      "step": 408064
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.331035605553856e-05,
      "loss": 4.1838,
      "step": 408576
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.330197010802804e-05,
      "loss": 4.1814,
      "step": 409088
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.329358416051752e-05,
      "loss": 4.1792,
      "step": 409600
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3285198213007e-05,
      "loss": 4.1836,
      "step": 410112
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.327682864430021e-05,
      "loss": 4.1843,
      "step": 410624
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.326844269678969e-05,
      "loss": 4.1929,
      "step": 411136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3260073128082904e-05,
      "loss": 4.1842,
      "step": 411648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3251687180572383e-05,
      "loss": 4.1821,
      "step": 412160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3243301233061863e-05,
      "loss": 4.1678,
      "step": 412672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.323493166435507e-05,
      "loss": 4.1957,
      "step": 413184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.322654571684455e-05,
      "loss": 4.1668,
      "step": 413696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.321815976933403e-05,
      "loss": 4.1681,
      "step": 414208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.320977382182351e-05,
      "loss": 4.1825,
      "step": 414720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.320138787431299e-05,
      "loss": 4.1843,
      "step": 415232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.319300192680247e-05,
      "loss": 4.1668,
      "step": 415744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.318461597929195e-05,
      "loss": 4.1757,
      "step": 416256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.317623003178143e-05,
      "loss": 4.1636,
      "step": 416768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.316784408427091e-05,
      "loss": 4.1912,
      "step": 417280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.315947451556412e-05,
      "loss": 4.1843,
      "step": 417792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.315108856805361e-05,
      "loss": 4.1816,
      "step": 418304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.314270262054309e-05,
      "loss": 4.1797,
      "step": 418816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.313431667303257e-05,
      "loss": 4.1847,
      "step": 419328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.312593072552205e-05,
      "loss": 4.1929,
      "step": 419840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.311754477801153e-05,
      "loss": 4.1678,
      "step": 420352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.310917520930474e-05,
      "loss": 4.1815,
      "step": 420864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.310078926179422e-05,
      "loss": 4.1787,
      "step": 421376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.30924033142837e-05,
      "loss": 4.1649,
      "step": 421888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.308401736677318e-05,
      "loss": 4.1854,
      "step": 422400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3075647798066386e-05,
      "loss": 4.1804,
      "step": 422912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3067261850555866e-05,
      "loss": 4.1815,
      "step": 423424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3058875903045346e-05,
      "loss": 4.1927,
      "step": 423936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3050489955534826e-05,
      "loss": 4.1693,
      "step": 424448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3042104008024306e-05,
      "loss": 4.1748,
      "step": 424960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3033718060513786e-05,
      "loss": 4.1693,
      "step": 425472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3025332113003266e-05,
      "loss": 4.1754,
      "step": 425984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3016962544296475e-05,
      "loss": 4.1793,
      "step": 426496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3008576596785955e-05,
      "loss": 4.178,
      "step": 427008
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.3000190649275435e-05,
      "loss": 4.1752,
      "step": 427520
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2991804701764915e-05,
      "loss": 4.183,
      "step": 428032
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2983435133058124e-05,
      "loss": 4.172,
      "step": 428544
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2975049185547604e-05,
      "loss": 4.1747,
      "step": 429056
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2966663238037084e-05,
      "loss": 4.1774,
      "step": 429568
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2958277290526564e-05,
      "loss": 4.1801,
      "step": 430080
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2949891343016043e-05,
      "loss": 4.1772,
      "step": 430592
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.294150539550553e-05,
      "loss": 4.1663,
      "step": 431104
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.293311944799501e-05,
      "loss": 4.1782,
      "step": 431616
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.292473350048449e-05,
      "loss": 4.1737,
      "step": 432128
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.291638031058143e-05,
      "loss": 4.1858,
      "step": 432640
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.290799436307091e-05,
      "loss": 4.1736,
      "step": 433152
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.289960841556039e-05,
      "loss": 4.1714,
      "step": 433664
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.289122246804987e-05,
      "loss": 4.1665,
      "step": 434176
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.288283652053935e-05,
      "loss": 4.1814,
      "step": 434688
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.287445057302883e-05,
      "loss": 4.1496,
      "step": 435200
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.286606462551831e-05,
      "loss": 4.1673,
      "step": 435712
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.285767867800779e-05,
      "loss": 4.1702,
      "step": 436224
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2849309109301e-05,
      "loss": 4.169,
      "step": 436736
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.284092316179048e-05,
      "loss": 4.1663,
      "step": 437248
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2832537214279964e-05,
      "loss": 4.1599,
      "step": 437760
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2824151266769444e-05,
      "loss": 4.1655,
      "step": 438272
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2815765319258924e-05,
      "loss": 4.1782,
      "step": 438784
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.280739575055213e-05,
      "loss": 4.1723,
      "step": 439296
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.279900980304161e-05,
      "loss": 4.1671,
      "step": 439808
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.279062385553109e-05,
      "loss": 4.1748,
      "step": 440320
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.278223790802057e-05,
      "loss": 4.1817,
      "step": 440832
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.277385196051005e-05,
      "loss": 4.1731,
      "step": 441344
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.276546601299953e-05,
      "loss": 4.1594,
      "step": 441856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.275708006548901e-05,
      "loss": 4.1685,
      "step": 442368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2748694117978486e-05,
      "loss": 4.1715,
      "step": 442880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2740308170467966e-05,
      "loss": 4.156,
      "step": 443392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.273193860176118e-05,
      "loss": 4.1597,
      "step": 443904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.272355265425066e-05,
      "loss": 4.1663,
      "step": 444416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.271516670674014e-05,
      "loss": 4.1639,
      "step": 444928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.270679713803336e-05,
      "loss": 4.1653,
      "step": 445440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.269841119052284e-05,
      "loss": 4.1735,
      "step": 445952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.269002524301231e-05,
      "loss": 4.1519,
      "step": 446464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2681655674305526e-05,
      "loss": 4.1642,
      "step": 446976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2673269726795006e-05,
      "loss": 4.156,
      "step": 447488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2664883779284486e-05,
      "loss": 4.177,
      "step": 448000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.265649783177396e-05,
      "loss": 4.1665,
      "step": 448512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.264811188426344e-05,
      "loss": 4.1694,
      "step": 449024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2639742315556655e-05,
      "loss": 4.1585,
      "step": 449536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2631356368046135e-05,
      "loss": 4.1701,
      "step": 450048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2622970420535615e-05,
      "loss": 4.1728,
      "step": 450560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2614584473025095e-05,
      "loss": 4.1647,
      "step": 451072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.260621490431831e-05,
      "loss": 4.1645,
      "step": 451584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2597828956807784e-05,
      "loss": 4.1689,
      "step": 452096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2589443009297264e-05,
      "loss": 4.1617,
      "step": 452608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2581057061786744e-05,
      "loss": 4.1737,
      "step": 453120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2572671114276224e-05,
      "loss": 4.1539,
      "step": 453632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.256430154556943e-05,
      "loss": 4.1643,
      "step": 454144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.255591559805891e-05,
      "loss": 4.1516,
      "step": 454656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.254752965054839e-05,
      "loss": 4.1538,
      "step": 455168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.253914370303787e-05,
      "loss": 4.1649,
      "step": 455680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.253077413433109e-05,
      "loss": 4.1551,
      "step": 456192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.252238818682057e-05,
      "loss": 4.1703,
      "step": 456704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.251400223931005e-05,
      "loss": 4.1538,
      "step": 457216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.250561629179953e-05,
      "loss": 4.1671,
      "step": 457728
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.1620635986328125,
      "eval_runtime": 293.9667,
      "eval_samples_per_second": 1298.075,
      "eval_steps_per_second": 40.566,
      "step": 457920
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.249723034428901e-05,
      "loss": 4.1695,
      "step": 458240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.248884439677849e-05,
      "loss": 4.1418,
      "step": 458752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.248045844926797e-05,
      "loss": 4.1682,
      "step": 459264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.247207250175745e-05,
      "loss": 4.1615,
      "step": 459776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.246368655424693e-05,
      "loss": 4.1688,
      "step": 460288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.245530060673641e-05,
      "loss": 4.1564,
      "step": 460800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.244691465922589e-05,
      "loss": 4.1472,
      "step": 461312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.243852871171537e-05,
      "loss": 4.1485,
      "step": 461824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.243014276420485e-05,
      "loss": 4.1577,
      "step": 462336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.242175681669432e-05,
      "loss": 4.1598,
      "step": 462848
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.241337086918381e-05,
      "loss": 4.1629,
      "step": 463360
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.240498492167329e-05,
      "loss": 4.167,
      "step": 463872
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.23966153529665e-05,
      "loss": 4.1559,
      "step": 464384
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.238822940545598e-05,
      "loss": 4.1568,
      "step": 464896
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.237984345794546e-05,
      "loss": 4.1494,
      "step": 465408
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.237145751043494e-05,
      "loss": 4.1398,
      "step": 465920
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2363087941728146e-05,
      "loss": 4.1587,
      "step": 466432
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2354701994217626e-05,
      "loss": 4.1511,
      "step": 466944
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2346316046707106e-05,
      "loss": 4.1509,
      "step": 467456
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2337930099196586e-05,
      "loss": 4.1594,
      "step": 467968
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2329544151686066e-05,
      "loss": 4.1613,
      "step": 468480
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2321158204175546e-05,
      "loss": 4.1559,
      "step": 468992
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2312772256665026e-05,
      "loss": 4.151,
      "step": 469504
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2304386309154506e-05,
      "loss": 4.1576,
      "step": 470016
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2296000361643986e-05,
      "loss": 4.1535,
      "step": 470528
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.228761441413347e-05,
      "loss": 4.1542,
      "step": 471040
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.227922846662295e-05,
      "loss": 4.1444,
      "step": 471552
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.227085889791616e-05,
      "loss": 4.1492,
      "step": 472064
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.226247295040564e-05,
      "loss": 4.1508,
      "step": 472576
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.225408700289512e-05,
      "loss": 4.1473,
      "step": 473088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.22457010553846e-05,
      "loss": 4.1478,
      "step": 473600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.223731510787408e-05,
      "loss": 4.1512,
      "step": 474112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.222892916036356e-05,
      "loss": 4.1536,
      "step": 474624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2220543212853035e-05,
      "loss": 4.1516,
      "step": 475136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2212157265342515e-05,
      "loss": 4.1418,
      "step": 475648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.220378769663573e-05,
      "loss": 4.1523,
      "step": 476160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.219540174912521e-05,
      "loss": 4.1469,
      "step": 476672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.218701580161469e-05,
      "loss": 4.1332,
      "step": 477184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.217862985410417e-05,
      "loss": 4.1379,
      "step": 477696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.217024390659365e-05,
      "loss": 4.1366,
      "step": 478208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.216185795908313e-05,
      "loss": 4.1422,
      "step": 478720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.215347201157261e-05,
      "loss": 4.1505,
      "step": 479232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.214508606406209e-05,
      "loss": 4.1443,
      "step": 479744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.21367164953553e-05,
      "loss": 4.1511,
      "step": 480256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.212833054784478e-05,
      "loss": 4.1523,
      "step": 480768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.211994460033426e-05,
      "loss": 4.1393,
      "step": 481280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.211157503162747e-05,
      "loss": 4.145,
      "step": 481792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.210318908411695e-05,
      "loss": 4.1441,
      "step": 482304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.209480313660643e-05,
      "loss": 4.1195,
      "step": 482816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.208641718909591e-05,
      "loss": 4.1595,
      "step": 483328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2078031241585395e-05,
      "loss": 4.1365,
      "step": 483840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2069645294074875e-05,
      "loss": 4.154,
      "step": 484352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2061275725368084e-05,
      "loss": 4.134,
      "step": 484864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2052889777857564e-05,
      "loss": 4.1406,
      "step": 485376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2044503830347044e-05,
      "loss": 4.1348,
      "step": 485888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2036117882836524e-05,
      "loss": 4.1366,
      "step": 486400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2027731935326004e-05,
      "loss": 4.1403,
      "step": 486912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2019345987815484e-05,
      "loss": 4.15,
      "step": 487424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.201097641910869e-05,
      "loss": 4.1456,
      "step": 487936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.200259047159817e-05,
      "loss": 4.1325,
      "step": 488448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.199420452408765e-05,
      "loss": 4.1247,
      "step": 488960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.198581857657713e-05,
      "loss": 4.1558,
      "step": 489472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.197744900787035e-05,
      "loss": 4.1223,
      "step": 489984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.196906306035983e-05,
      "loss": 4.1216,
      "step": 490496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.196067711284931e-05,
      "loss": 4.1405,
      "step": 491008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.195229116533879e-05,
      "loss": 4.1455,
      "step": 491520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.194390521782827e-05,
      "loss": 4.1211,
      "step": 492032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.193553564912148e-05,
      "loss": 4.1324,
      "step": 492544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.192714970161096e-05,
      "loss": 4.1235,
      "step": 493056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.191876375410044e-05,
      "loss": 4.1453,
      "step": 493568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1910394185393647e-05,
      "loss": 4.1419,
      "step": 494080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1902008237883126e-05,
      "loss": 4.1406,
      "step": 494592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1893622290372606e-05,
      "loss": 4.1322,
      "step": 495104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1885236342862086e-05,
      "loss": 4.1401,
      "step": 495616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1876850395351566e-05,
      "loss": 4.153,
      "step": 496128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1868464447841046e-05,
      "loss": 4.1203,
      "step": 496640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1860078500330526e-05,
      "loss": 4.1385,
      "step": 497152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1851692552820006e-05,
      "loss": 4.1382,
      "step": 497664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1843306605309486e-05,
      "loss": 4.123,
      "step": 498176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1834920657798966e-05,
      "loss": 4.14,
      "step": 498688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1826534710288446e-05,
      "loss": 4.139,
      "step": 499200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1818148762777926e-05,
      "loss": 4.1401,
      "step": 499712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1809779194071135e-05,
      "loss": 4.1501,
      "step": 500224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1801393246560615e-05,
      "loss": 4.1243,
      "step": 500736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1793007299050095e-05,
      "loss": 4.1315,
      "step": 501248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1784621351539575e-05,
      "loss": 4.1279,
      "step": 501760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1776235404029055e-05,
      "loss": 4.1343,
      "step": 502272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1767865835322264e-05,
      "loss": 4.1341,
      "step": 502784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.175947988781175e-05,
      "loss": 4.1417,
      "step": 503296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.175109394030123e-05,
      "loss": 4.125,
      "step": 503808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.174270799279071e-05,
      "loss": 4.1448,
      "step": 504320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.173432204528019e-05,
      "loss": 4.1322,
      "step": 504832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.17259524765734e-05,
      "loss": 4.1347,
      "step": 505344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.171756652906288e-05,
      "loss": 4.136,
      "step": 505856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.170918058155236e-05,
      "loss": 4.1342,
      "step": 506368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.170079463404184e-05,
      "loss": 4.1399,
      "step": 506880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.169240868653132e-05,
      "loss": 4.124,
      "step": 507392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.16840227390208e-05,
      "loss": 4.1342,
      "step": 507904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.167563679151028e-05,
      "loss": 4.1355,
      "step": 508416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.166726722280349e-05,
      "loss": 4.1414,
      "step": 508928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.165888127529297e-05,
      "loss": 4.1353,
      "step": 509440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1650511706586184e-05,
      "loss": 4.1281,
      "step": 509952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1642142137879393e-05,
      "loss": 4.1277,
      "step": 510464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1633756190368873e-05,
      "loss": 4.137,
      "step": 510976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1625370242858353e-05,
      "loss": 4.1079,
      "step": 511488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.161698429534783e-05,
      "loss": 4.1246,
      "step": 512000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.160859834783731e-05,
      "loss": 4.127,
      "step": 512512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.160021240032679e-05,
      "loss": 4.1289,
      "step": 513024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.159182645281627e-05,
      "loss": 4.1234,
      "step": 513536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.158344050530575e-05,
      "loss": 4.1236,
      "step": 514048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1575054557795226e-05,
      "loss": 4.1248,
      "step": 514560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1566668610284706e-05,
      "loss": 4.1357,
      "step": 515072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1558282662774186e-05,
      "loss": 4.1335,
      "step": 515584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.15499130940674e-05,
      "loss": 4.1229,
      "step": 516096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.154152714655688e-05,
      "loss": 4.134,
      "step": 516608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.153314119904636e-05,
      "loss": 4.1422,
      "step": 517120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.152475525153584e-05,
      "loss": 4.1352,
      "step": 517632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.151636930402532e-05,
      "loss": 4.1195,
      "step": 518144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.15079833565148e-05,
      "loss": 4.1271,
      "step": 518656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.149961378780801e-05,
      "loss": 4.131,
      "step": 519168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.149122784029749e-05,
      "loss": 4.1145,
      "step": 519680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.148284189278697e-05,
      "loss": 4.1207,
      "step": 520192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.147445594527645e-05,
      "loss": 4.125,
      "step": 520704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.146606999776593e-05,
      "loss": 4.1204,
      "step": 521216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.145768405025541e-05,
      "loss": 4.1271,
      "step": 521728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.144931448154863e-05,
      "loss": 4.1334,
      "step": 522240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.144092853403811e-05,
      "loss": 4.1167,
      "step": 522752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1432542586527587e-05,
      "loss": 4.1189,
      "step": 523264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1424173017820796e-05,
      "loss": 4.1231,
      "step": 523776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1415787070310276e-05,
      "loss": 4.1338,
      "step": 524288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1407401122799756e-05,
      "loss": 4.1279,
      "step": 524800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1399015175289236e-05,
      "loss": 4.1323,
      "step": 525312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1390629227778716e-05,
      "loss": 4.1166,
      "step": 525824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1382243280268195e-05,
      "loss": 4.1313,
      "step": 526336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1373857332757675e-05,
      "loss": 4.132,
      "step": 526848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1365471385247155e-05,
      "loss": 4.125,
      "step": 527360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1357085437736635e-05,
      "loss": 4.1274,
      "step": 527872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1348715869029844e-05,
      "loss": 4.1279,
      "step": 528384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1340329921519324e-05,
      "loss": 4.1229,
      "step": 528896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1331943974008804e-05,
      "loss": 4.1348,
      "step": 529408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.132355802649829e-05,
      "loss": 4.1158,
      "step": 529920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.131517207898777e-05,
      "loss": 4.1238,
      "step": 530432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1306786131477244e-05,
      "loss": 4.1146,
      "step": 530944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1298400183966724e-05,
      "loss": 4.1144,
      "step": 531456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1290014236456204e-05,
      "loss": 4.1225,
      "step": 531968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.128164466774942e-05,
      "loss": 4.1188,
      "step": 532480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.127325872023889e-05,
      "loss": 4.1289,
      "step": 532992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.126487277272837e-05,
      "loss": 4.1172,
      "step": 533504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.125648682521785e-05,
      "loss": 4.1296,
      "step": 534016
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.131078243255615,
      "eval_runtime": 352.6289,
      "eval_samples_per_second": 1082.132,
      "eval_steps_per_second": 33.817,
      "step": 534240
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.124810087770733e-05,
      "loss": 4.1146,
      "step": 534528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.123973130900054e-05,
      "loss": 4.1048,
      "step": 535040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.123134536149003e-05,
      "loss": 4.1285,
      "step": 535552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.122295941397951e-05,
      "loss": 4.1207,
      "step": 536064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.121457346646899e-05,
      "loss": 4.1286,
      "step": 536576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.120618751895847e-05,
      "loss": 4.1201,
      "step": 537088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.119780157144795e-05,
      "loss": 4.1126,
      "step": 537600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.118943200274116e-05,
      "loss": 4.1136,
      "step": 538112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.118104605523064e-05,
      "loss": 4.1189,
      "step": 538624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.117266010772012e-05,
      "loss": 4.1224,
      "step": 539136
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.11642741602096e-05,
      "loss": 4.1223,
      "step": 539648
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.115588821269908e-05,
      "loss": 4.1297,
      "step": 540160
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.114751864399229e-05,
      "loss": 4.1166,
      "step": 540672
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.113913269648177e-05,
      "loss": 4.1163,
      "step": 541184
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1130746748971247e-05,
      "loss": 4.113,
      "step": 541696
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1122360801460727e-05,
      "loss": 4.1024,
      "step": 542208
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.111399123275394e-05,
      "loss": 4.1205,
      "step": 542720
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.110560528524342e-05,
      "loss": 4.1151,
      "step": 543232
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.10972193377329e-05,
      "loss": 4.1133,
      "step": 543744
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.108883339022238e-05,
      "loss": 4.1145,
      "step": 544256
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.108044744271186e-05,
      "loss": 4.1301,
      "step": 544768
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.107206149520134e-05,
      "loss": 4.1179,
      "step": 545280
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.106367554769082e-05,
      "loss": 4.114,
      "step": 545792
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.10552896001803e-05,
      "loss": 4.1178,
      "step": 546304
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.104692003147351e-05,
      "loss": 4.1149,
      "step": 546816
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.103853408396299e-05,
      "loss": 4.1167,
      "step": 547328
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.10301645152562e-05,
      "loss": 4.1102,
      "step": 547840
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.102177856774568e-05,
      "loss": 4.1146,
      "step": 548352
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.101339262023517e-05,
      "loss": 4.1103,
      "step": 548864
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.100500667272465e-05,
      "loss": 4.1081,
      "step": 549376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.099662072521413e-05,
      "loss": 4.1092,
      "step": 549888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.098823477770361e-05,
      "loss": 4.1163,
      "step": 550400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.097984883019308e-05,
      "loss": 4.1166,
      "step": 550912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.097146288268256e-05,
      "loss": 4.1145,
      "step": 551424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0963093313975776e-05,
      "loss": 4.1067,
      "step": 551936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0954723745268985e-05,
      "loss": 4.1111,
      "step": 552448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0946337797758465e-05,
      "loss": 4.1112,
      "step": 552960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0937951850247945e-05,
      "loss": 4.1001,
      "step": 553472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0929565902737425e-05,
      "loss": 4.1003,
      "step": 553984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0921179955226905e-05,
      "loss": 4.0984,
      "step": 554496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0912794007716385e-05,
      "loss": 4.1109,
      "step": 555008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0904408060205865e-05,
      "loss": 4.1124,
      "step": 555520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0896022112695345e-05,
      "loss": 4.1053,
      "step": 556032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0887636165184825e-05,
      "loss": 4.1164,
      "step": 556544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0879266596478034e-05,
      "loss": 4.1129,
      "step": 557056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0870880648967514e-05,
      "loss": 4.1045,
      "step": 557568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0862494701456994e-05,
      "loss": 4.1115,
      "step": 558080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.08541251327502e-05,
      "loss": 4.1063,
      "step": 558592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.084573918523968e-05,
      "loss": 4.0857,
      "step": 559104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.083735323772916e-05,
      "loss": 4.1187,
      "step": 559616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.082896729021864e-05,
      "loss": 4.099,
      "step": 560128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.082058134270812e-05,
      "loss": 4.1189,
      "step": 560640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.08121953951976e-05,
      "loss": 4.1026,
      "step": 561152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.080380944768708e-05,
      "loss": 4.1031,
      "step": 561664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.079542350017657e-05,
      "loss": 4.0962,
      "step": 562176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.078707031027351e-05,
      "loss": 4.1005,
      "step": 562688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.077868436276299e-05,
      "loss": 4.1069,
      "step": 563200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.077029841525247e-05,
      "loss": 4.1105,
      "step": 563712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.076191246774195e-05,
      "loss": 4.1093,
      "step": 564224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.075352652023143e-05,
      "loss": 4.1006,
      "step": 564736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0745156951524636e-05,
      "loss": 4.0894,
      "step": 565248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0736771004014116e-05,
      "loss": 4.1195,
      "step": 565760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0728385056503596e-05,
      "loss": 4.09,
      "step": 566272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0719999108993076e-05,
      "loss": 4.0795,
      "step": 566784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0711613161482556e-05,
      "loss": 4.1043,
      "step": 567296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0703227213972036e-05,
      "loss": 4.1106,
      "step": 567808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.069484126646152e-05,
      "loss": 4.0898,
      "step": 568320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0686455318951e-05,
      "loss": 4.0984,
      "step": 568832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.067810212904794e-05,
      "loss": 4.0847,
      "step": 569344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.066971618153742e-05,
      "loss": 4.1079,
      "step": 569856
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.06613302340269e-05,
      "loss": 4.1086,
      "step": 570368
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.065294428651638e-05,
      "loss": 4.1028,
      "step": 570880
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.064455833900586e-05,
      "loss": 4.0998,
      "step": 571392
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.063617239149534e-05,
      "loss": 4.1046,
      "step": 571904
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.062778644398482e-05,
      "loss": 4.1168,
      "step": 572416
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.061941687527803e-05,
      "loss": 4.089,
      "step": 572928
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.061103092776751e-05,
      "loss": 4.1054,
      "step": 573440
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.060264498025699e-05,
      "loss": 4.0974,
      "step": 573952
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0594259032746476e-05,
      "loss": 4.0935,
      "step": 574464
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0585873085235956e-05,
      "loss": 4.1031,
      "step": 574976
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0577487137725436e-05,
      "loss": 4.1072,
      "step": 575488
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.056910119021491e-05,
      "loss": 4.1014,
      "step": 576000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0560731621508125e-05,
      "loss": 4.1173,
      "step": 576512
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0552345673997605e-05,
      "loss": 4.0891,
      "step": 577024
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0543959726487085e-05,
      "loss": 4.0927,
      "step": 577536
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.053557377897656e-05,
      "loss": 4.0953,
      "step": 578048
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0527204210269774e-05,
      "loss": 4.1024,
      "step": 578560
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0518818262759254e-05,
      "loss": 4.099,
      "step": 579072
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0510432315248734e-05,
      "loss": 4.1074,
      "step": 579584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0502046367738214e-05,
      "loss": 4.0901,
      "step": 580096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0493660420227694e-05,
      "loss": 4.1099,
      "step": 580608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0485274472717174e-05,
      "loss": 4.1016,
      "step": 581120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0476888525206654e-05,
      "loss": 4.0943,
      "step": 581632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0468502577696134e-05,
      "loss": 4.1005,
      "step": 582144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.046013300898934e-05,
      "loss": 4.1041,
      "step": 582656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.045174706147882e-05,
      "loss": 4.1066,
      "step": 583168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.04433611139683e-05,
      "loss": 4.0921,
      "step": 583680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.043497516645778e-05,
      "loss": 4.0962,
      "step": 584192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.042660559775099e-05,
      "loss": 4.0993,
      "step": 584704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.041823602904421e-05,
      "loss": 4.1105,
      "step": 585216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.040985008153368e-05,
      "loss": 4.1041,
      "step": 585728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.040146413402317e-05,
      "loss": 4.0916,
      "step": 586240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.039307818651265e-05,
      "loss": 4.0955,
      "step": 586752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.038469223900213e-05,
      "loss": 4.1037,
      "step": 587264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.037630629149161e-05,
      "loss": 4.0734,
      "step": 587776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.036792034398109e-05,
      "loss": 4.0921,
      "step": 588288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.035953439647057e-05,
      "loss": 4.0932,
      "step": 588800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.035114844896005e-05,
      "loss": 4.0981,
      "step": 589312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.034277888025326e-05,
      "loss": 4.0881,
      "step": 589824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.033439293274274e-05,
      "loss": 4.0922,
      "step": 590336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0326023364035946e-05,
      "loss": 4.094,
      "step": 590848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0317637416525426e-05,
      "loss": 4.0976,
      "step": 591360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0309251469014906e-05,
      "loss": 4.0966,
      "step": 591872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0300865521504386e-05,
      "loss": 4.0929,
      "step": 592384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0292479573993866e-05,
      "loss": 4.1011,
      "step": 592896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.028411000528708e-05,
      "loss": 4.1086,
      "step": 593408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.027572405777656e-05,
      "loss": 4.1013,
      "step": 593920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.026733811026604e-05,
      "loss": 4.0867,
      "step": 594432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.025895216275552e-05,
      "loss": 4.0941,
      "step": 594944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.025058259404873e-05,
      "loss": 4.1008,
      "step": 595456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.024219664653821e-05,
      "loss": 4.0808,
      "step": 595968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.023381069902769e-05,
      "loss": 4.0898,
      "step": 596480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.022542475151717e-05,
      "loss": 4.0921,
      "step": 596992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.021705518281038e-05,
      "loss": 4.0897,
      "step": 597504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.020866923529986e-05,
      "loss": 4.0898,
      "step": 598016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.020028328778934e-05,
      "loss": 4.1013,
      "step": 598528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.019189734027882e-05,
      "loss": 4.0861,
      "step": 599040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0183511392768306e-05,
      "loss": 4.0877,
      "step": 599552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0175141824061515e-05,
      "loss": 4.0863,
      "step": 600064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0166755876550995e-05,
      "loss": 4.1053,
      "step": 600576
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0158369929040475e-05,
      "loss": 4.0915,
      "step": 601088
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0149983981529955e-05,
      "loss": 4.0998,
      "step": 601600
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0141598034019435e-05,
      "loss": 4.0842,
      "step": 602112
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0133228465312644e-05,
      "loss": 4.0977,
      "step": 602624
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0124842517802124e-05,
      "loss": 4.1005,
      "step": 603136
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0116456570291604e-05,
      "loss": 4.0951,
      "step": 603648
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0108070622781084e-05,
      "loss": 4.0944,
      "step": 604160
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0099684675270564e-05,
      "loss": 4.0934,
      "step": 604672
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.009131510656377e-05,
      "loss": 4.091,
      "step": 605184
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.008292915905325e-05,
      "loss": 4.103,
      "step": 605696
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.007454321154274e-05,
      "loss": 4.0853,
      "step": 606208
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.006615726403222e-05,
      "loss": 4.0913,
      "step": 606720
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.005777131652169e-05,
      "loss": 4.0854,
      "step": 607232
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.004938536901117e-05,
      "loss": 4.0831,
      "step": 607744
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.004099942150065e-05,
      "loss": 4.0853,
      "step": 608256
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.003261347399013e-05,
      "loss": 4.0872,
      "step": 608768
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.002424390528334e-05,
      "loss": 4.099,
      "step": 609280
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.001585795777282e-05,
      "loss": 4.086,
      "step": 609792
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.00074720102623e-05,
      "loss": 4.0958,
      "step": 610304
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.106538772583008,
      "eval_runtime": 282.9465,
      "eval_samples_per_second": 1348.633,
      "eval_steps_per_second": 42.146,
      "step": 610560
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.999908606275178e-05,
      "loss": 4.0944,
      "step": 610816
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.999070011524126e-05,
      "loss": 4.0762,
      "step": 611328
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.998231416773074e-05,
      "loss": 4.0959,
      "step": 611840
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.997392822022022e-05,
      "loss": 4.0914,
      "step": 612352
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.996554227270971e-05,
      "loss": 4.0967,
      "step": 612864
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.995715632519919e-05,
      "loss": 4.0881,
      "step": 613376
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.994877037768867e-05,
      "loss": 4.0875,
      "step": 613888
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.994038443017815e-05,
      "loss": 4.0768,
      "step": 614400
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.993199848266763e-05,
      "loss": 4.0861,
      "step": 614912
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.992361253515711e-05,
      "loss": 4.0941,
      "step": 615424
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.991522658764658e-05,
      "loss": 4.0881,
      "step": 615936
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.990684064013606e-05,
      "loss": 4.0952,
      "step": 616448
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.989847107142928e-05,
      "loss": 4.0947,
      "step": 616960
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.989008512391876e-05,
      "loss": 4.0782,
      "step": 617472
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.988169917640823e-05,
      "loss": 4.0786,
      "step": 617984
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.987331322889771e-05,
      "loss": 4.0764,
      "step": 618496
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.986492728138719e-05,
      "loss": 4.0884,
      "step": 619008
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.985654133387668e-05,
      "loss": 4.0852,
      "step": 619520
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.984815538636616e-05,
      "loss": 4.0799,
      "step": 620032
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.983976943885564e-05,
      "loss": 4.0859,
      "step": 620544
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9831399870148846e-05,
      "loss": 4.0979,
      "step": 621056
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9823013922638326e-05,
      "loss": 4.0894,
      "step": 621568
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9814627975127806e-05,
      "loss": 4.0858,
      "step": 622080
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9806242027617286e-05,
      "loss": 4.0839,
      "step": 622592
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9797856080106766e-05,
      "loss": 4.085,
      "step": 623104
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9789470132596246e-05,
      "loss": 4.086,
      "step": 623616
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9781084185085726e-05,
      "loss": 4.0752,
      "step": 624128
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9772698237575205e-05,
      "loss": 4.0899,
      "step": 624640
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9764312290064685e-05,
      "loss": 4.079,
      "step": 625152
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9755942721357895e-05,
      "loss": 4.0745,
      "step": 625664
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9747556773847374e-05,
      "loss": 4.0824,
      "step": 626176
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.973917082633686e-05,
      "loss": 4.0805,
      "step": 626688
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.973080125763007e-05,
      "loss": 4.0856,
      "step": 627200
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.972241531011955e-05,
      "loss": 4.0872,
      "step": 627712
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.971402936260903e-05,
      "loss": 4.0765,
      "step": 628224
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.970564341509851e-05,
      "loss": 4.0805,
      "step": 628736
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.969725746758799e-05,
      "loss": 4.0858,
      "step": 629248
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.968887152007747e-05,
      "loss": 4.0652,
      "step": 629760
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.968048557256694e-05,
      "loss": 4.0734,
      "step": 630272
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.967209962505642e-05,
      "loss": 4.0709,
      "step": 630784
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.966374643515337e-05,
      "loss": 4.0743,
      "step": 631296
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.965536048764285e-05,
      "loss": 4.0803,
      "step": 631808
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.964697454013233e-05,
      "loss": 4.0774,
      "step": 632320
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.963858859262181e-05,
      "loss": 4.09,
      "step": 632832
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9630202645111295e-05,
      "loss": 4.0774,
      "step": 633344
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.962181669760077e-05,
      "loss": 4.079,
      "step": 633856
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9613447128893984e-05,
      "loss": 4.0786,
      "step": 634368
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9605061181383464e-05,
      "loss": 4.0783,
      "step": 634880
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9596675233872944e-05,
      "loss": 4.0537,
      "step": 635392
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.958828928636242e-05,
      "loss": 4.0883,
      "step": 635904
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.957991971765563e-05,
      "loss": 4.0716,
      "step": 636416
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.957153377014511e-05,
      "loss": 4.0865,
      "step": 636928
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.956314782263459e-05,
      "loss": 4.0738,
      "step": 637440
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9554761875124066e-05,
      "loss": 4.0757,
      "step": 637952
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9546375927613546e-05,
      "loss": 4.0641,
      "step": 638464
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.953798998010303e-05,
      "loss": 4.0726,
      "step": 638976
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.952960403259251e-05,
      "loss": 4.0819,
      "step": 639488
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.952121808508199e-05,
      "loss": 4.0736,
      "step": 640000
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.95128485163752e-05,
      "loss": 4.0823,
      "step": 640512
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.950446256886468e-05,
      "loss": 4.078,
      "step": 641024
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.949607662135416e-05,
      "loss": 4.0542,
      "step": 641536
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.948769067384364e-05,
      "loss": 4.0907,
      "step": 642048
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.947932110513685e-05,
      "loss": 4.0642,
      "step": 642560
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9470951536430066e-05,
      "loss": 4.0526,
      "step": 643072
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.946256558891954e-05,
      "loss": 4.0782,
      "step": 643584
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.945417964140902e-05,
      "loss": 4.0754,
      "step": 644096
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.94457936938985e-05,
      "loss": 4.0602,
      "step": 644608
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9437407746387986e-05,
      "loss": 4.0698,
      "step": 645120
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9429021798877466e-05,
      "loss": 4.0594,
      "step": 645632
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9420652230170675e-05,
      "loss": 4.076,
      "step": 646144
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9412266282660155e-05,
      "loss": 4.0774,
      "step": 646656
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9403880335149635e-05,
      "loss": 4.072,
      "step": 647168
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9395494387639115e-05,
      "loss": 4.0731,
      "step": 647680
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9387108440128595e-05,
      "loss": 4.0736,
      "step": 648192
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9378738871421804e-05,
      "loss": 4.0867,
      "step": 648704
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9370352923911284e-05,
      "loss": 4.0591,
      "step": 649216
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9361966976400764e-05,
      "loss": 4.0744,
      "step": 649728
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9353581028890244e-05,
      "loss": 4.0711,
      "step": 650240
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9345195081379724e-05,
      "loss": 4.0663,
      "step": 650752
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9336809133869204e-05,
      "loss": 4.0748,
      "step": 651264
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9328423186358684e-05,
      "loss": 4.0767,
      "step": 651776
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.932003723884817e-05,
      "loss": 4.0751,
      "step": 652288
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.931166767014138e-05,
      "loss": 4.0892,
      "step": 652800
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.930328172263086e-05,
      "loss": 4.0607,
      "step": 653312
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.929489577512034e-05,
      "loss": 4.0658,
      "step": 653824
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.928650982760982e-05,
      "loss": 4.0665,
      "step": 654336
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.92781238800993e-05,
      "loss": 4.0726,
      "step": 654848
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.926973793258878e-05,
      "loss": 4.0683,
      "step": 655360
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.926136836388199e-05,
      "loss": 4.0774,
      "step": 655872
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.925298241637147e-05,
      "loss": 4.0666,
      "step": 656384
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.924459646886095e-05,
      "loss": 4.0799,
      "step": 656896
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.923621052135042e-05,
      "loss": 4.0734,
      "step": 657408
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.922782457383991e-05,
      "loss": 4.0689,
      "step": 657920
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.921943862632939e-05,
      "loss": 4.0732,
      "step": 658432
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9211069057622604e-05,
      "loss": 4.0722,
      "step": 658944
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.920268311011208e-05,
      "loss": 4.0785,
      "step": 659456
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.919429716260156e-05,
      "loss": 4.0652,
      "step": 659968
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.918591121509104e-05,
      "loss": 4.0647,
      "step": 660480
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.917752526758052e-05,
      "loss": 4.0741,
      "step": 660992
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.916913932007e-05,
      "loss": 4.0824,
      "step": 661504
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.916075337255948e-05,
      "loss": 4.0726,
      "step": 662016
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9152383803852686e-05,
      "loss": 4.0695,
      "step": 662528
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9143997856342166e-05,
      "loss": 4.0666,
      "step": 663040
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9135611908831646e-05,
      "loss": 4.0751,
      "step": 663552
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.912724234012486e-05,
      "loss": 4.046,
      "step": 664064
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.911885639261434e-05,
      "loss": 4.0666,
      "step": 664576
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.911047044510382e-05,
      "loss": 4.0536,
      "step": 665088
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.91020844975933e-05,
      "loss": 4.0709,
      "step": 665600
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.909369855008278e-05,
      "loss": 4.0609,
      "step": 666112
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.908531260257226e-05,
      "loss": 4.07,
      "step": 666624
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.907692665506174e-05,
      "loss": 4.0636,
      "step": 667136
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.906854070755122e-05,
      "loss": 4.0711,
      "step": 667648
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.906017113884443e-05,
      "loss": 4.0685,
      "step": 668160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.905178519133391e-05,
      "loss": 4.0672,
      "step": 668672
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.904339924382339e-05,
      "loss": 4.0749,
      "step": 669184
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.903501329631287e-05,
      "loss": 4.078,
      "step": 669696
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.902662734880235e-05,
      "loss": 4.0735,
      "step": 670208
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.901824140129183e-05,
      "loss": 4.0642,
      "step": 670720
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.900985545378131e-05,
      "loss": 4.0618,
      "step": 671232
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9001485885074527e-05,
      "loss": 4.0715,
      "step": 671744
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8993099937564007e-05,
      "loss": 4.0581,
      "step": 672256
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8984713990053487e-05,
      "loss": 4.06,
      "step": 672768
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8976328042542966e-05,
      "loss": 4.0639,
      "step": 673280
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8967958473836176e-05,
      "loss": 4.0625,
      "step": 673792
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8959572526325656e-05,
      "loss": 4.0599,
      "step": 674304
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8951186578815135e-05,
      "loss": 4.0746,
      "step": 674816
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8942800631304615e-05,
      "loss": 4.0598,
      "step": 675328
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.893441468379409e-05,
      "loss": 4.0616,
      "step": 675840
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.892602873628357e-05,
      "loss": 4.0563,
      "step": 676352
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.891764278877305e-05,
      "loss": 4.0763,
      "step": 676864
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.890925684126253e-05,
      "loss": 4.0681,
      "step": 677376
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8900887272555744e-05,
      "loss": 4.0676,
      "step": 677888
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8892501325045224e-05,
      "loss": 4.0602,
      "step": 678400
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8884115377534704e-05,
      "loss": 4.0731,
      "step": 678912
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8875729430024184e-05,
      "loss": 4.072,
      "step": 679424
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8867343482513664e-05,
      "loss": 4.0694,
      "step": 679936
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8858957535003144e-05,
      "loss": 4.0692,
      "step": 680448
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8850571587492624e-05,
      "loss": 4.0603,
      "step": 680960
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8842185639982104e-05,
      "loss": 4.0658,
      "step": 681472
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.883381607127531e-05,
      "loss": 4.0765,
      "step": 681984
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.882543012376479e-05,
      "loss": 4.0584,
      "step": 682496
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.881704417625427e-05,
      "loss": 4.0645,
      "step": 683008
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.880865822874375e-05,
      "loss": 4.0603,
      "step": 683520
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.880028866003696e-05,
      "loss": 4.0578,
      "step": 684032
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.879190271252645e-05,
      "loss": 4.0562,
      "step": 684544
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.878351676501593e-05,
      "loss": 4.0623,
      "step": 685056
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.877514719630914e-05,
      "loss": 4.0702,
      "step": 685568
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.876676124879862e-05,
      "loss": 4.063,
      "step": 686080
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.87583753012881e-05,
      "loss": 4.0672,
      "step": 686592
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.087634563446045,
      "eval_runtime": 302.5332,
      "eval_samples_per_second": 1261.319,
      "eval_steps_per_second": 39.417,
      "step": 686880
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.874998935377758e-05,
      "loss": 4.0545,
      "step": 687104
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.874160340626706e-05,
      "loss": 4.0515,
      "step": 687616
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.873321745875654e-05,
      "loss": 4.0678,
      "step": 688128
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.872483151124602e-05,
      "loss": 4.0629,
      "step": 688640
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.87164455637355e-05,
      "loss": 4.0729,
      "step": 689152
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.870805961622497e-05,
      "loss": 4.0607,
      "step": 689664
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.869969004751819e-05,
      "loss": 4.0601,
      "step": 690176
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8691304100007667e-05,
      "loss": 4.0494,
      "step": 690688
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8682918152497147e-05,
      "loss": 4.0611,
      "step": 691200
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8674532204986626e-05,
      "loss": 4.0666,
      "step": 691712
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8666146257476106e-05,
      "loss": 4.062,
      "step": 692224
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.865777668876932e-05,
      "loss": 4.0693,
      "step": 692736
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.864940712006253e-05,
      "loss": 4.0738,
      "step": 693248
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.864102117255201e-05,
      "loss": 4.0497,
      "step": 693760
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.863263522504149e-05,
      "loss": 4.0532,
      "step": 694272
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.862424927753097e-05,
      "loss": 4.0493,
      "step": 694784
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8615863330020444e-05,
      "loss": 4.0649,
      "step": 695296
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8607477382509924e-05,
      "loss": 4.0535,
      "step": 695808
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8599091434999404e-05,
      "loss": 4.0551,
      "step": 696320
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8590705487488884e-05,
      "loss": 4.0597,
      "step": 696832
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.858231953997837e-05,
      "loss": 4.0689,
      "step": 697344
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.857393359246785e-05,
      "loss": 4.0645,
      "step": 697856
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.856554764495733e-05,
      "loss": 4.0603,
      "step": 698368
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.855716169744681e-05,
      "loss": 4.0603,
      "step": 698880
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.854877574993629e-05,
      "loss": 4.0585,
      "step": 699392
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.85404061812295e-05,
      "loss": 4.0636,
      "step": 699904
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.853203661252271e-05,
      "loss": 4.0457,
      "step": 700416
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.852365066501219e-05,
      "loss": 4.0643,
      "step": 700928
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.851526471750167e-05,
      "loss": 4.053,
      "step": 701440
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.850687876999115e-05,
      "loss": 4.0496,
      "step": 701952
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.849850920128436e-05,
      "loss": 4.0594,
      "step": 702464
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.849012325377384e-05,
      "loss": 4.0568,
      "step": 702976
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.848173730626332e-05,
      "loss": 4.055,
      "step": 703488
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8473351358752805e-05,
      "loss": 4.066,
      "step": 704000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8464965411242285e-05,
      "loss": 4.0503,
      "step": 704512
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8456579463731765e-05,
      "loss": 4.0557,
      "step": 705024
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8448193516221245e-05,
      "loss": 4.0586,
      "step": 705536
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8439807568710724e-05,
      "loss": 4.0362,
      "step": 706048
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8431421621200204e-05,
      "loss": 4.0502,
      "step": 706560
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8423035673689684e-05,
      "loss": 4.0461,
      "step": 707072
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8414649726179164e-05,
      "loss": 4.0499,
      "step": 707584
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8406280157472373e-05,
      "loss": 4.0573,
      "step": 708096
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8397894209961853e-05,
      "loss": 4.0486,
      "step": 708608
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.838950826245133e-05,
      "loss": 4.0622,
      "step": 709120
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8381122314940807e-05,
      "loss": 4.0538,
      "step": 709632
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8372736367430287e-05,
      "loss": 4.0567,
      "step": 710144
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.83643667987235e-05,
      "loss": 4.0534,
      "step": 710656
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.835598085121299e-05,
      "loss": 4.0522,
      "step": 711168
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.834759490370246e-05,
      "loss": 4.034,
      "step": 711680
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.833920895619194e-05,
      "loss": 4.06,
      "step": 712192
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.833082300868142e-05,
      "loss": 4.0478,
      "step": 712704
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.832245343997464e-05,
      "loss": 4.0637,
      "step": 713216
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.831406749246411e-05,
      "loss": 4.0467,
      "step": 713728
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.830568154495359e-05,
      "loss": 4.0502,
      "step": 714240
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.829729559744307e-05,
      "loss": 4.0399,
      "step": 714752
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.828890964993255e-05,
      "loss": 4.0476,
      "step": 715264
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.828054008122576e-05,
      "loss": 4.0562,
      "step": 715776
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.827215413371524e-05,
      "loss": 4.0472,
      "step": 716288
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.826376818620473e-05,
      "loss": 4.0607,
      "step": 716800
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.825538223869421e-05,
      "loss": 4.0508,
      "step": 717312
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.824699629118369e-05,
      "loss": 4.0264,
      "step": 717824
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.823861034367317e-05,
      "loss": 4.0638,
      "step": 718336
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.823022439616265e-05,
      "loss": 4.0399,
      "step": 718848
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.822183844865213e-05,
      "loss": 4.0309,
      "step": 719360
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8213468879945336e-05,
      "loss": 4.0508,
      "step": 719872
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8205099311238545e-05,
      "loss": 4.0527,
      "step": 720384
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8196713363728025e-05,
      "loss": 4.0368,
      "step": 720896
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8188327416217505e-05,
      "loss": 4.0437,
      "step": 721408
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8179941468706985e-05,
      "loss": 4.0346,
      "step": 721920
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8171555521196465e-05,
      "loss": 4.0516,
      "step": 722432
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.816318595248968e-05,
      "loss": 4.0543,
      "step": 722944
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.815480000497916e-05,
      "loss": 4.0447,
      "step": 723456
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.814641405746864e-05,
      "loss": 4.0513,
      "step": 723968
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.813802810995812e-05,
      "loss": 4.0484,
      "step": 724480
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.812965854125133e-05,
      "loss": 4.0613,
      "step": 724992
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.812127259374081e-05,
      "loss": 4.038,
      "step": 725504
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.811288664623029e-05,
      "loss": 4.0486,
      "step": 726016
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.810450069871977e-05,
      "loss": 4.0451,
      "step": 726528
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.809611475120925e-05,
      "loss": 4.0433,
      "step": 727040
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.808774518250246e-05,
      "loss": 4.0494,
      "step": 727552
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.807935923499194e-05,
      "loss": 4.0478,
      "step": 728064
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.807097328748142e-05,
      "loss": 4.054,
      "step": 728576
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.80625873399709e-05,
      "loss": 4.0602,
      "step": 729088
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.805420139246038e-05,
      "loss": 4.0428,
      "step": 729600
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.804581544494986e-05,
      "loss": 4.0398,
      "step": 730112
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8037429497439345e-05,
      "loss": 4.0409,
      "step": 730624
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8029043549928825e-05,
      "loss": 4.0487,
      "step": 731136
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.80206576024183e-05,
      "loss": 4.0428,
      "step": 731648
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8012288033711514e-05,
      "loss": 4.0473,
      "step": 732160
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.8003902086200994e-05,
      "loss": 4.0502,
      "step": 732672
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.799551613869047e-05,
      "loss": 4.0514,
      "step": 733184
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.798713019117995e-05,
      "loss": 4.0484,
      "step": 733696
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.797874424366943e-05,
      "loss": 4.0438,
      "step": 734208
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.797035829615891e-05,
      "loss": 4.0501,
      "step": 734720
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7961988727452116e-05,
      "loss": 4.0504,
      "step": 735232
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7953602779941596e-05,
      "loss": 4.0517,
      "step": 735744
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.794521683243108e-05,
      "loss": 4.0387,
      "step": 736256
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.793683088492056e-05,
      "loss": 4.0451,
      "step": 736768
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.792846131621377e-05,
      "loss": 4.0478,
      "step": 737280
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.792007536870325e-05,
      "loss": 4.0572,
      "step": 737792
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.791168942119273e-05,
      "loss": 4.0484,
      "step": 738304
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.790330347368221e-05,
      "loss": 4.047,
      "step": 738816
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.789491752617169e-05,
      "loss": 4.0434,
      "step": 739328
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.788653157866117e-05,
      "loss": 4.0481,
      "step": 739840
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.787814563115065e-05,
      "loss": 4.0262,
      "step": 740352
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.786975968364013e-05,
      "loss": 4.0387,
      "step": 740864
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.786140649373707e-05,
      "loss": 4.0308,
      "step": 741376
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.785302054622655e-05,
      "loss": 4.0469,
      "step": 741888
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7844634598716036e-05,
      "loss": 4.0399,
      "step": 742400
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7836248651205516e-05,
      "loss": 4.041,
      "step": 742912
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7827862703694996e-05,
      "loss": 4.0412,
      "step": 743424
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7819476756184476e-05,
      "loss": 4.0445,
      "step": 743936
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7811107187477685e-05,
      "loss": 4.0479,
      "step": 744448
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7802721239967165e-05,
      "loss": 4.0419,
      "step": 744960
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7794335292456645e-05,
      "loss": 4.0495,
      "step": 745472
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7785949344946125e-05,
      "loss": 4.0526,
      "step": 745984
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7777563397435605e-05,
      "loss": 4.0451,
      "step": 746496
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7769177449925085e-05,
      "loss": 4.0454,
      "step": 747008
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7760791502414565e-05,
      "loss": 4.0348,
      "step": 747520
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7752405554904045e-05,
      "loss": 4.0439,
      "step": 748032
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7744035986197254e-05,
      "loss": 4.0425,
      "step": 748544
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7735650038686734e-05,
      "loss": 4.0279,
      "step": 749056
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.772726409117622e-05,
      "loss": 4.0398,
      "step": 749568
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.77188781436657e-05,
      "loss": 4.0372,
      "step": 750080
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.771049219615518e-05,
      "loss": 4.042,
      "step": 750592
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.770212262744839e-05,
      "loss": 4.0484,
      "step": 751104
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.769373667993787e-05,
      "loss": 4.0389,
      "step": 751616
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.768536711123108e-05,
      "loss": 4.0385,
      "step": 752128
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.767698116372056e-05,
      "loss": 4.0352,
      "step": 752640
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.766859521621004e-05,
      "loss": 4.0518,
      "step": 753152
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.766020926869952e-05,
      "loss": 4.044,
      "step": 753664
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7651823321189e-05,
      "loss": 4.0455,
      "step": 754176
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.764343737367848e-05,
      "loss": 4.0313,
      "step": 754688
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.763505142616796e-05,
      "loss": 4.0543,
      "step": 755200
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.762666547865744e-05,
      "loss": 4.0448,
      "step": 755712
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.761827953114692e-05,
      "loss": 4.0459,
      "step": 756224
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.760990996244013e-05,
      "loss": 4.0442,
      "step": 756736
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7601540393733344e-05,
      "loss": 4.0401,
      "step": 757248
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7593154446222823e-05,
      "loss": 4.0402,
      "step": 757760
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7584768498712303e-05,
      "loss": 4.0587,
      "step": 758272
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.757638255120178e-05,
      "loss": 4.0306,
      "step": 758784
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7567996603691257e-05,
      "loss": 4.0426,
      "step": 759296
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7559610656180737e-05,
      "loss": 4.0381,
      "step": 759808
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7551224708670216e-05,
      "loss": 4.0298,
      "step": 760320
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7542838761159696e-05,
      "loss": 4.0355,
      "step": 760832
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.753446919245291e-05,
      "loss": 4.039,
      "step": 761344
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.752608324494239e-05,
      "loss": 4.0464,
      "step": 761856
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.75177136762356e-05,
      "loss": 4.0419,
      "step": 762368
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.750932772872508e-05,
      "loss": 4.0405,
      "step": 762880
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.071054458618164,
      "eval_runtime": 303.3292,
      "eval_samples_per_second": 1258.009,
      "eval_steps_per_second": 39.314,
      "step": 763200
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.750094178121456e-05,
      "loss": 4.0402,
      "step": 763392
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.749255583370404e-05,
      "loss": 4.0275,
      "step": 763904
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.748416988619352e-05,
      "loss": 4.045,
      "step": 764416
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7475783938683e-05,
      "loss": 4.0426,
      "step": 764928
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.746739799117248e-05,
      "loss": 4.0486,
      "step": 765440
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.745901204366196e-05,
      "loss": 4.0386,
      "step": 765952
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.745062609615144e-05,
      "loss": 4.0357,
      "step": 766464
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.744224014864092e-05,
      "loss": 4.0301,
      "step": 766976
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.74338542011304e-05,
      "loss": 4.0332,
      "step": 767488
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.742546825361988e-05,
      "loss": 4.0428,
      "step": 768000
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.741708230610936e-05,
      "loss": 4.0393,
      "step": 768512
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.740871273740258e-05,
      "loss": 4.0481,
      "step": 769024
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7400343168695786e-05,
      "loss": 4.0475,
      "step": 769536
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7391957221185266e-05,
      "loss": 4.0278,
      "step": 770048
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7383571273674746e-05,
      "loss": 4.0285,
      "step": 770560
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7375185326164226e-05,
      "loss": 4.0265,
      "step": 771072
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7366799378653706e-05,
      "loss": 4.0448,
      "step": 771584
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7358413431143186e-05,
      "loss": 4.0335,
      "step": 772096
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7350027483632666e-05,
      "loss": 4.0313,
      "step": 772608
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.734164153612214e-05,
      "loss": 4.034,
      "step": 773120
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.733325558861162e-05,
      "loss": 4.0461,
      "step": 773632
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.73248696411011e-05,
      "loss": 4.0384,
      "step": 774144
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.731648369359058e-05,
      "loss": 4.0409,
      "step": 774656
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.730809774608006e-05,
      "loss": 4.036,
      "step": 775168
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7299711798569545e-05,
      "loss": 4.039,
      "step": 775680
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7291325851059025e-05,
      "loss": 4.0383,
      "step": 776192
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7282939903548505e-05,
      "loss": 4.0222,
      "step": 776704
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7274570334841714e-05,
      "loss": 4.0418,
      "step": 777216
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7266184387331194e-05,
      "loss": 4.0278,
      "step": 777728
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7257798439820674e-05,
      "loss": 4.0241,
      "step": 778240
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7249412492310154e-05,
      "loss": 4.0422,
      "step": 778752
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7241026544799634e-05,
      "loss": 4.0334,
      "step": 779264
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7232640597289114e-05,
      "loss": 4.0325,
      "step": 779776
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7224254649778594e-05,
      "loss": 4.0421,
      "step": 780288
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7215868702268074e-05,
      "loss": 4.0287,
      "step": 780800
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7207482754757554e-05,
      "loss": 4.035,
      "step": 781312
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.719909680724703e-05,
      "loss": 4.0351,
      "step": 781824
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7190710859736514e-05,
      "loss": 4.0119,
      "step": 782336
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7182324912225994e-05,
      "loss": 4.0279,
      "step": 782848
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.717395534351921e-05,
      "loss": 4.0248,
      "step": 783360
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.716556939600868e-05,
      "loss": 4.0245,
      "step": 783872
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.715718344849816e-05,
      "loss": 4.0381,
      "step": 784384
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.714881387979138e-05,
      "loss": 4.0276,
      "step": 784896
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.714042793228085e-05,
      "loss": 4.0403,
      "step": 785408
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.713204198477033e-05,
      "loss": 4.0335,
      "step": 785920
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.712365603725981e-05,
      "loss": 4.0327,
      "step": 786432
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.711527008974929e-05,
      "loss": 4.0318,
      "step": 786944
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.710688414223877e-05,
      "loss": 4.0307,
      "step": 787456
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.709849819472825e-05,
      "loss": 4.0118,
      "step": 787968
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.709011224721773e-05,
      "loss": 4.0369,
      "step": 788480
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.708174267851095e-05,
      "loss": 4.0277,
      "step": 788992
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7073373109804157e-05,
      "loss": 4.0401,
      "step": 789504
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7064987162293637e-05,
      "loss": 4.0254,
      "step": 790016
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7056601214783116e-05,
      "loss": 4.0218,
      "step": 790528
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7048215267272596e-05,
      "loss": 4.0221,
      "step": 791040
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7039829319762076e-05,
      "loss": 4.024,
      "step": 791552
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7031443372251556e-05,
      "loss": 4.0344,
      "step": 792064
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7023057424741036e-05,
      "loss": 4.0216,
      "step": 792576
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7014671477230516e-05,
      "loss": 4.0415,
      "step": 793088
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7006285529719996e-05,
      "loss": 4.032,
      "step": 793600
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6997899582209476e-05,
      "loss": 4.0017,
      "step": 794112
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6989513634698956e-05,
      "loss": 4.0426,
      "step": 794624
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6981127687188436e-05,
      "loss": 4.018,
      "step": 795136
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6972758118481645e-05,
      "loss": 4.0096,
      "step": 795648
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.696437217097113e-05,
      "loss": 4.0247,
      "step": 796160
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.695598622346061e-05,
      "loss": 4.0327,
      "step": 796672
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.694760027595009e-05,
      "loss": 4.0131,
      "step": 797184
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.69392307072433e-05,
      "loss": 4.0231,
      "step": 797696
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.693084475973278e-05,
      "loss": 4.0123,
      "step": 798208
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.692245881222226e-05,
      "loss": 4.0241,
      "step": 798720
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.691407286471174e-05,
      "loss": 4.0338,
      "step": 799232
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.690570329600495e-05,
      "loss": 4.0229,
      "step": 799744
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.689731734849443e-05,
      "loss": 4.0303,
      "step": 800256
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.688893140098391e-05,
      "loss": 4.0247,
      "step": 800768
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.688054545347339e-05,
      "loss": 4.0364,
      "step": 801280
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.687215950596287e-05,
      "loss": 4.0179,
      "step": 801792
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6863789937256086e-05,
      "loss": 4.0324,
      "step": 802304
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6855403989745566e-05,
      "loss": 4.0209,
      "step": 802816
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6847034421038775e-05,
      "loss": 4.0211,
      "step": 803328
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6838648473528255e-05,
      "loss": 4.0277,
      "step": 803840
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6830262526017735e-05,
      "loss": 4.0237,
      "step": 804352
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6821876578507214e-05,
      "loss": 4.03,
      "step": 804864
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.681349063099669e-05,
      "loss": 4.0419,
      "step": 805376
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.680510468348617e-05,
      "loss": 4.0257,
      "step": 805888
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6796735114779383e-05,
      "loss": 4.0156,
      "step": 806400
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6788349167268863e-05,
      "loss": 4.0181,
      "step": 806912
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.677996321975834e-05,
      "loss": 4.0296,
      "step": 807424
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.677157727224782e-05,
      "loss": 4.0183,
      "step": 807936
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.676320770354104e-05,
      "loss": 4.0316,
      "step": 808448
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.675482175603051e-05,
      "loss": 4.0289,
      "step": 808960
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.674643580851999e-05,
      "loss": 4.0281,
      "step": 809472
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.673804986100947e-05,
      "loss": 4.0275,
      "step": 809984
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.672966391349895e-05,
      "loss": 4.0204,
      "step": 810496
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.672127796598843e-05,
      "loss": 4.0322,
      "step": 811008
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.671290839728164e-05,
      "loss": 4.0262,
      "step": 811520
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.670452244977112e-05,
      "loss": 4.0317,
      "step": 812032
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.669615288106434e-05,
      "loss": 4.0218,
      "step": 812544
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.668776693355381e-05,
      "loss": 4.0197,
      "step": 813056
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.667938098604329e-05,
      "loss": 4.0261,
      "step": 813568
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.667099503853278e-05,
      "loss": 4.0354,
      "step": 814080
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.666260909102226e-05,
      "loss": 4.0285,
      "step": 814592
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.665422314351174e-05,
      "loss": 4.0226,
      "step": 815104
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.664583719600122e-05,
      "loss": 4.0268,
      "step": 815616
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.66374512484907e-05,
      "loss": 4.0233,
      "step": 816128
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6629081679783906e-05,
      "loss": 4.0112,
      "step": 816640
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6620695732273386e-05,
      "loss": 4.0155,
      "step": 817152
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6612309784762866e-05,
      "loss": 4.0108,
      "step": 817664
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6603923837252346e-05,
      "loss": 4.0292,
      "step": 818176
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6595537889741826e-05,
      "loss": 4.0194,
      "step": 818688
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6587151942231306e-05,
      "loss": 4.0215,
      "step": 819200
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6578765994720786e-05,
      "loss": 4.0165,
      "step": 819712
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6570396426013995e-05,
      "loss": 4.0285,
      "step": 820224
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.656202685730721e-05,
      "loss": 4.0243,
      "step": 820736
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.655364090979669e-05,
      "loss": 4.0221,
      "step": 821248
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.654525496228617e-05,
      "loss": 4.0298,
      "step": 821760
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.653686901477565e-05,
      "loss": 4.0326,
      "step": 822272
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.652848306726513e-05,
      "loss": 4.029,
      "step": 822784
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.652009711975461e-05,
      "loss": 4.0225,
      "step": 823296
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.651171117224409e-05,
      "loss": 4.0137,
      "step": 823808
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.65033416035373e-05,
      "loss": 4.0245,
      "step": 824320
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.649495565602678e-05,
      "loss": 4.0255,
      "step": 824832
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.648656970851626e-05,
      "loss": 4.0026,
      "step": 825344
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.647818376100574e-05,
      "loss": 4.0233,
      "step": 825856
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.646979781349522e-05,
      "loss": 4.0202,
      "step": 826368
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.64614118659847e-05,
      "loss": 4.0198,
      "step": 826880
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.645302591847418e-05,
      "loss": 4.0303,
      "step": 827392
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.644463997096366e-05,
      "loss": 4.0177,
      "step": 827904
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.643625402345314e-05,
      "loss": 4.0191,
      "step": 828416
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.642788445474635e-05,
      "loss": 4.0132,
      "step": 828928
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.641949850723583e-05,
      "loss": 4.0346,
      "step": 829440
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.641111255972531e-05,
      "loss": 4.022,
      "step": 829952
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.640272661221479e-05,
      "loss": 4.0272,
      "step": 830464
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6394357043508e-05,
      "loss": 4.0112,
      "step": 830976
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.638597109599748e-05,
      "loss": 4.0289,
      "step": 831488
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.637758514848696e-05,
      "loss": 4.0264,
      "step": 832000
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.636919920097644e-05,
      "loss": 4.0267,
      "step": 832512
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.636081325346592e-05,
      "loss": 4.0254,
      "step": 833024
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.63524273059554e-05,
      "loss": 4.0167,
      "step": 833536
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.634404135844488e-05,
      "loss": 4.0231,
      "step": 834048
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6335655410934364e-05,
      "loss": 4.036,
      "step": 834560
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.63273022210313e-05,
      "loss": 4.013,
      "step": 835072
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.631891627352078e-05,
      "loss": 4.0205,
      "step": 835584
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6310546704814e-05,
      "loss": 4.0175,
      "step": 836096
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.630216075730347e-05,
      "loss": 4.0162,
      "step": 836608
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.629377480979295e-05,
      "loss": 4.0128,
      "step": 837120
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.628538886228243e-05,
      "loss": 4.0183,
      "step": 837632
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.627700291477191e-05,
      "loss": 4.023,
      "step": 838144
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.626861696726139e-05,
      "loss": 4.029,
      "step": 838656
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.626023101975087e-05,
      "loss": 4.0207,
      "step": 839168
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.0583062171936035,
      "eval_runtime": 306.0123,
      "eval_samples_per_second": 1246.979,
      "eval_steps_per_second": 38.969,
      "step": 839520
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.625184507224035e-05,
      "loss": 4.0097,
      "step": 839680
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.624345912472983e-05,
      "loss": 4.01,
      "step": 840192
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.623507317721932e-05,
      "loss": 4.0237,
      "step": 840704
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.62266872297088e-05,
      "loss": 4.024,
      "step": 841216
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6218317661002006e-05,
      "loss": 4.0296,
      "step": 841728
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6209931713491486e-05,
      "loss": 4.0218,
      "step": 842240
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6201545765980966e-05,
      "loss": 4.0187,
      "step": 842752
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6193159818470446e-05,
      "loss": 4.0104,
      "step": 843264
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6184773870959926e-05,
      "loss": 4.0151,
      "step": 843776
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6176387923449406e-05,
      "loss": 4.0162,
      "step": 844288
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6168001975938886e-05,
      "loss": 4.0259,
      "step": 844800
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.615961602842836e-05,
      "loss": 4.028,
      "step": 845312
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.615127921732904e-05,
      "loss": 4.033,
      "step": 845824
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.614289326981852e-05,
      "loss": 4.004,
      "step": 846336
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6134507322308e-05,
      "loss": 4.017,
      "step": 846848
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.612612137479748e-05,
      "loss": 3.9999,
      "step": 847360
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.611773542728696e-05,
      "loss": 4.0252,
      "step": 847872
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.610934947977644e-05,
      "loss": 4.0152,
      "step": 848384
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.610096353226592e-05,
      "loss": 4.0083,
      "step": 848896
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.60925775847554e-05,
      "loss": 4.0151,
      "step": 849408
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.608420801604861e-05,
      "loss": 4.0322,
      "step": 849920
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.607582206853809e-05,
      "loss": 4.0152,
      "step": 850432
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.606743612102757e-05,
      "loss": 4.0227,
      "step": 850944
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.605905017351705e-05,
      "loss": 4.0179,
      "step": 851456
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.605066422600653e-05,
      "loss": 4.0196,
      "step": 851968
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.604227827849601e-05,
      "loss": 4.0173,
      "step": 852480
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.603389233098549e-05,
      "loss": 4.0068,
      "step": 852992
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.602550638347497e-05,
      "loss": 4.0193,
      "step": 853504
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.601712043596445e-05,
      "loss": 4.0143,
      "step": 854016
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.600873448845393e-05,
      "loss": 4.0073,
      "step": 854528
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.600036491974714e-05,
      "loss": 4.0178,
      "step": 855040
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.599197897223662e-05,
      "loss": 4.0138,
      "step": 855552
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.59835930247261e-05,
      "loss": 4.0151,
      "step": 856064
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.597520707721558e-05,
      "loss": 4.0228,
      "step": 856576
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.596683750850879e-05,
      "loss": 4.0093,
      "step": 857088
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.595845156099827e-05,
      "loss": 4.015,
      "step": 857600
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5950065613487747e-05,
      "loss": 4.0154,
      "step": 858112
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5941679665977227e-05,
      "loss": 3.9944,
      "step": 858624
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5933293718466706e-05,
      "loss": 4.0092,
      "step": 859136
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.592490777095619e-05,
      "loss": 4.0036,
      "step": 859648
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.591652182344567e-05,
      "loss": 4.0037,
      "step": 860160
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.590813587593515e-05,
      "loss": 4.0174,
      "step": 860672
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.589976630722836e-05,
      "loss": 4.0184,
      "step": 861184
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.589139673852157e-05,
      "loss": 4.0128,
      "step": 861696
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.588301079101105e-05,
      "loss": 4.0186,
      "step": 862208
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.587462484350053e-05,
      "loss": 4.011,
      "step": 862720
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.586623889599001e-05,
      "loss": 4.0134,
      "step": 863232
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.585785294847949e-05,
      "loss": 4.0087,
      "step": 863744
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.584946700096897e-05,
      "loss": 3.9971,
      "step": 864256
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.584108105345845e-05,
      "loss": 4.0133,
      "step": 864768
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.583269510594793e-05,
      "loss": 4.013,
      "step": 865280
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.582430915843741e-05,
      "loss": 4.0205,
      "step": 865792
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.581593958973063e-05,
      "loss": 4.0097,
      "step": 866304
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.580755364222011e-05,
      "loss": 4.0064,
      "step": 866816
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.579916769470959e-05,
      "loss": 4.0028,
      "step": 867328
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5790798126002796e-05,
      "loss": 4.0024,
      "step": 867840
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5782412178492276e-05,
      "loss": 4.0189,
      "step": 868352
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5774026230981756e-05,
      "loss": 3.9998,
      "step": 868864
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5765640283471236e-05,
      "loss": 4.0255,
      "step": 869376
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5757254335960716e-05,
      "loss": 4.014,
      "step": 869888
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.574886838845019e-05,
      "loss": 3.988,
      "step": 870400
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.574048244093967e-05,
      "loss": 4.0178,
      "step": 870912
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.573209649342915e-05,
      "loss": 4.0013,
      "step": 871424
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5723726924722365e-05,
      "loss": 3.9926,
      "step": 871936
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5715340977211845e-05,
      "loss": 4.008,
      "step": 872448
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5706955029701325e-05,
      "loss": 4.0126,
      "step": 872960
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5698569082190804e-05,
      "loss": 3.9959,
      "step": 873472
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.569019951348402e-05,
      "loss": 4.0085,
      "step": 873984
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.568182994477723e-05,
      "loss": 3.991,
      "step": 874496
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.567344399726671e-05,
      "loss": 4.0069,
      "step": 875008
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.566505804975619e-05,
      "loss": 4.016,
      "step": 875520
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.565667210224566e-05,
      "loss": 4.0068,
      "step": 876032
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.564828615473514e-05,
      "loss": 4.0124,
      "step": 876544
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.563990020722462e-05,
      "loss": 4.0059,
      "step": 877056
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.56315142597141e-05,
      "loss": 4.0202,
      "step": 877568
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.562312831220358e-05,
      "loss": 4.0028,
      "step": 878080
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.561474236469306e-05,
      "loss": 4.0122,
      "step": 878592
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.560635641718255e-05,
      "loss": 4.0017,
      "step": 879104
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5598003227279494e-05,
      "loss": 4.0047,
      "step": 879616
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.558961727976897e-05,
      "loss": 4.0142,
      "step": 880128
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.558123133225845e-05,
      "loss": 4.0023,
      "step": 880640
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.557284538474793e-05,
      "loss": 4.0134,
      "step": 881152
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.556445943723741e-05,
      "loss": 4.0259,
      "step": 881664
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.555607348972689e-05,
      "loss": 4.0092,
      "step": 882176
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5547703921020096e-05,
      "loss": 3.9928,
      "step": 882688
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.553933435231331e-05,
      "loss": 4.0021,
      "step": 883200
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5530948404802785e-05,
      "loss": 4.0102,
      "step": 883712
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.552256245729227e-05,
      "loss": 4.0062,
      "step": 884224
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.551417650978175e-05,
      "loss": 4.0092,
      "step": 884736
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.550579056227123e-05,
      "loss": 4.0121,
      "step": 885248
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.549740461476071e-05,
      "loss": 4.0078,
      "step": 885760
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.548901866725019e-05,
      "loss": 4.0093,
      "step": 886272
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.548063271973967e-05,
      "loss": 4.0044,
      "step": 886784
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.547224677222915e-05,
      "loss": 4.0132,
      "step": 887296
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.546386082471863e-05,
      "loss": 4.0095,
      "step": 887808
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.545547487720811e-05,
      "loss": 4.019,
      "step": 888320
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.544710530850132e-05,
      "loss": 3.9975,
      "step": 888832
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.543873573979453e-05,
      "loss": 4.0093,
      "step": 889344
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.543034979228401e-05,
      "loss": 4.0094,
      "step": 889856
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.542196384477349e-05,
      "loss": 4.0127,
      "step": 890368
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.541357789726297e-05,
      "loss": 4.0137,
      "step": 890880
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5405191949752456e-05,
      "loss": 4.002,
      "step": 891392
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5396806002241936e-05,
      "loss": 4.008,
      "step": 891904
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5388420054731416e-05,
      "loss": 4.0049,
      "step": 892416
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5380034107220896e-05,
      "loss": 4.0014,
      "step": 892928
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5371648159710376e-05,
      "loss": 3.9937,
      "step": 893440
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5363262212199856e-05,
      "loss": 3.9951,
      "step": 893952
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.535487626468933e-05,
      "loss": 4.0098,
      "step": 894464
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5346506695982545e-05,
      "loss": 4.0059,
      "step": 894976
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5338120748472025e-05,
      "loss": 3.9987,
      "step": 895488
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.53297348009615e-05,
      "loss": 3.9964,
      "step": 896000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.532134885345098e-05,
      "loss": 4.0128,
      "step": 896512
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.531296290594046e-05,
      "loss": 4.0086,
      "step": 897024
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.530457695842994e-05,
      "loss": 4.0031,
      "step": 897536
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5296191010919425e-05,
      "loss": 4.0138,
      "step": 898048
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5287805063408905e-05,
      "loss": 4.0168,
      "step": 898560
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5279419115898385e-05,
      "loss": 4.0076,
      "step": 899072
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5271033168387865e-05,
      "loss": 4.0024,
      "step": 899584
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5262663599681074e-05,
      "loss": 3.9999,
      "step": 900096
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.525429403097428e-05,
      "loss": 4.0022,
      "step": 900608
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.524590808346376e-05,
      "loss": 4.014,
      "step": 901120
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.523752213595324e-05,
      "loss": 3.9818,
      "step": 901632
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.522913618844272e-05,
      "loss": 4.0066,
      "step": 902144
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.52207502409322e-05,
      "loss": 4.0031,
      "step": 902656
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.521236429342168e-05,
      "loss": 4.0046,
      "step": 903168
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.520399472471489e-05,
      "loss": 4.0103,
      "step": 903680
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.519560877720437e-05,
      "loss": 4.0013,
      "step": 904192
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.518722282969386e-05,
      "loss": 4.0012,
      "step": 904704
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.517883688218334e-05,
      "loss": 3.9926,
      "step": 905216
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.517046731347655e-05,
      "loss": 4.0183,
      "step": 905728
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.516208136596603e-05,
      "loss": 4.0028,
      "step": 906240
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.515369541845551e-05,
      "loss": 4.0136,
      "step": 906752
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.514530947094499e-05,
      "loss": 3.9938,
      "step": 907264
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.513692352343447e-05,
      "loss": 4.0105,
      "step": 907776
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.512853757592395e-05,
      "loss": 4.014,
      "step": 908288
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.512015162841343e-05,
      "loss": 4.0045,
      "step": 908800
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.511176568090291e-05,
      "loss": 4.01,
      "step": 909312
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5103396112196116e-05,
      "loss": 4.0001,
      "step": 909824
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5095010164685596e-05,
      "loss": 4.0036,
      "step": 910336
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5086624217175076e-05,
      "loss": 4.017,
      "step": 910848
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5078238269664556e-05,
      "loss": 4.0004,
      "step": 911360
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.506986870095777e-05,
      "loss": 4.002,
      "step": 911872
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.506148275344725e-05,
      "loss": 4.0019,
      "step": 912384
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.505309680593673e-05,
      "loss": 3.9957,
      "step": 912896
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.504471085842621e-05,
      "loss": 3.9969,
      "step": 913408
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5036324910915685e-05,
      "loss": 4.0056,
      "step": 913920
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.50279553422089e-05,
      "loss": 4.0008,
      "step": 914432
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.501956939469838e-05,
      "loss": 4.0089,
      "step": 914944
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.501118344718786e-05,
      "loss": 4.0077,
      "step": 915456
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.047134876251221,
      "eval_runtime": 296.5129,
      "eval_samples_per_second": 1286.929,
      "eval_steps_per_second": 40.217,
      "step": 915840
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.5002797499677334e-05,
      "loss": 4.0215,
      "step": 915968
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4994411552166814e-05,
      "loss": 3.9893,
      "step": 916480
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4986025604656294e-05,
      "loss": 4.0026,
      "step": 916992
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.497763965714578e-05,
      "loss": 4.0104,
      "step": 917504
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.496925370963526e-05,
      "loss": 4.007,
      "step": 918016
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.496086776212474e-05,
      "loss": 4.0056,
      "step": 918528
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.495248181461422e-05,
      "loss": 4.0017,
      "step": 919040
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.49440958671037e-05,
      "loss": 3.9934,
      "step": 919552
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.493570991959318e-05,
      "loss": 3.9958,
      "step": 920064
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.492732397208266e-05,
      "loss": 4.0008,
      "step": 920576
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.491893802457214e-05,
      "loss": 4.0092,
      "step": 921088
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.491055207706162e-05,
      "loss": 4.009,
      "step": 921600
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.490218250835483e-05,
      "loss": 4.0221,
      "step": 922112
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.489379656084431e-05,
      "loss": 3.9845,
      "step": 922624
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.488541061333379e-05,
      "loss": 3.9964,
      "step": 923136
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.487702466582327e-05,
      "loss": 3.9838,
      "step": 923648
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.486863871831275e-05,
      "loss": 4.0104,
      "step": 924160
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.486025277080223e-05,
      "loss": 4.0006,
      "step": 924672
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.485186682329171e-05,
      "loss": 3.9905,
      "step": 925184
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.484348087578119e-05,
      "loss": 3.9965,
      "step": 925696
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.483509492827067e-05,
      "loss": 4.0131,
      "step": 926208
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.482672535956388e-05,
      "loss": 3.9987,
      "step": 926720
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.481833941205336e-05,
      "loss": 4.0047,
      "step": 927232
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.480995346454284e-05,
      "loss": 4.0018,
      "step": 927744
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.480156751703232e-05,
      "loss": 4.005,
      "step": 928256
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.47931815695218e-05,
      "loss": 3.9994,
      "step": 928768
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.478479562201128e-05,
      "loss": 3.9895,
      "step": 929280
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.477640967450076e-05,
      "loss": 4.0041,
      "step": 929792
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.476802372699024e-05,
      "loss": 3.9905,
      "step": 930304
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.475963777947972e-05,
      "loss": 3.9913,
      "step": 930816
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.475126821077293e-05,
      "loss": 4.0023,
      "step": 931328
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4742882263262414e-05,
      "loss": 3.9967,
      "step": 931840
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.473451269455562e-05,
      "loss": 3.9972,
      "step": 932352
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.47261267470451e-05,
      "loss": 4.0077,
      "step": 932864
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.471774079953458e-05,
      "loss": 3.9941,
      "step": 933376
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.470935485202406e-05,
      "loss": 3.9975,
      "step": 933888
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.470096890451354e-05,
      "loss": 4.001,
      "step": 934400
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.469258295700302e-05,
      "loss": 3.9764,
      "step": 934912
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.468421338829623e-05,
      "loss": 3.9953,
      "step": 935424
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.467582744078571e-05,
      "loss": 3.9852,
      "step": 935936
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.466744149327519e-05,
      "loss": 3.99,
      "step": 936448
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.465905554576467e-05,
      "loss": 3.998,
      "step": 936960
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.465066959825415e-05,
      "loss": 4.0022,
      "step": 937472
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.464228365074363e-05,
      "loss": 3.9964,
      "step": 937984
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.463391408203685e-05,
      "loss": 4.0011,
      "step": 938496
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.462552813452633e-05,
      "loss": 3.9973,
      "step": 939008
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.461714218701581e-05,
      "loss": 3.994,
      "step": 939520
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.460875623950529e-05,
      "loss": 3.9959,
      "step": 940032
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.460037029199476e-05,
      "loss": 3.9782,
      "step": 940544
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.459198434448424e-05,
      "loss": 3.9954,
      "step": 941056
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.458359839697372e-05,
      "loss": 3.9993,
      "step": 941568
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.45752124494632e-05,
      "loss": 4.0024,
      "step": 942080
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.456684288075641e-05,
      "loss": 3.9941,
      "step": 942592
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.455845693324589e-05,
      "loss": 3.9866,
      "step": 943104
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.455007098573537e-05,
      "loss": 3.9845,
      "step": 943616
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4541701417028585e-05,
      "loss": 3.9847,
      "step": 944128
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4533315469518065e-05,
      "loss": 4.0037,
      "step": 944640
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4524929522007545e-05,
      "loss": 3.9861,
      "step": 945152
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4516543574497025e-05,
      "loss": 4.0065,
      "step": 945664
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4508174005790234e-05,
      "loss": 3.9943,
      "step": 946176
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4499788058279714e-05,
      "loss": 3.9786,
      "step": 946688
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4491402110769194e-05,
      "loss": 3.9986,
      "step": 947200
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4483016163258674e-05,
      "loss": 3.9855,
      "step": 947712
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.447464659455188e-05,
      "loss": 3.978,
      "step": 948224
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.446626064704136e-05,
      "loss": 3.9867,
      "step": 948736
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.445787469953084e-05,
      "loss": 3.9993,
      "step": 949248
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.444948875202032e-05,
      "loss": 3.9787,
      "step": 949760
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.44411028045098e-05,
      "loss": 3.9952,
      "step": 950272
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.443271685699929e-05,
      "loss": 3.9729,
      "step": 950784
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.44243472882925e-05,
      "loss": 3.9895,
      "step": 951296
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.441596134078198e-05,
      "loss": 4.0031,
      "step": 951808
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.440757539327146e-05,
      "loss": 3.9928,
      "step": 952320
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.439918944576094e-05,
      "loss": 3.9945,
      "step": 952832
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.439080349825042e-05,
      "loss": 3.9897,
      "step": 953344
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.43824175507399e-05,
      "loss": 4.0029,
      "step": 953856
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.437406436083684e-05,
      "loss": 3.9877,
      "step": 954368
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.436567841332632e-05,
      "loss": 3.997,
      "step": 954880
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.43572924658158e-05,
      "loss": 3.9884,
      "step": 955392
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.434890651830528e-05,
      "loss": 3.987,
      "step": 955904
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4340520570794757e-05,
      "loss": 3.9949,
      "step": 956416
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.433213462328424e-05,
      "loss": 3.9853,
      "step": 956928
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.432374867577372e-05,
      "loss": 4.0004,
      "step": 957440
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.43153627282632e-05,
      "loss": 4.0056,
      "step": 957952
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.430699315955641e-05,
      "loss": 3.9967,
      "step": 958464
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.429860721204589e-05,
      "loss": 3.9791,
      "step": 958976
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.429022126453537e-05,
      "loss": 3.9855,
      "step": 959488
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.428183531702485e-05,
      "loss": 3.9918,
      "step": 960000
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.427346574831806e-05,
      "loss": 3.9921,
      "step": 960512
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.426507980080754e-05,
      "loss": 3.9925,
      "step": 961024
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.425669385329702e-05,
      "loss": 3.9963,
      "step": 961536
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.42483079057865e-05,
      "loss": 3.9934,
      "step": 962048
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.423992195827598e-05,
      "loss": 3.9947,
      "step": 962560
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.423153601076546e-05,
      "loss": 3.9901,
      "step": 963072
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.422315006325494e-05,
      "loss": 3.9975,
      "step": 963584
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.421476411574442e-05,
      "loss": 3.9942,
      "step": 964096
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.420639454703764e-05,
      "loss": 4.003,
      "step": 964608
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.419800859952712e-05,
      "loss": 3.9825,
      "step": 965120
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.41896226520166e-05,
      "loss": 3.9925,
      "step": 965632
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.418123670450607e-05,
      "loss": 3.9949,
      "step": 966144
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4172867135799286e-05,
      "loss": 3.9972,
      "step": 966656
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4164481188288766e-05,
      "loss": 4.0007,
      "step": 967168
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4156111619581975e-05,
      "loss": 3.9875,
      "step": 967680
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4147725672071455e-05,
      "loss": 3.9927,
      "step": 968192
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4139339724560935e-05,
      "loss": 3.9895,
      "step": 968704
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4130953777050415e-05,
      "loss": 3.987,
      "step": 969216
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4122567829539895e-05,
      "loss": 3.9783,
      "step": 969728
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4114181882029375e-05,
      "loss": 3.9791,
      "step": 970240
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4105795934518855e-05,
      "loss": 3.9931,
      "step": 970752
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.409742636581207e-05,
      "loss": 3.9919,
      "step": 971264
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4089040418301544e-05,
      "loss": 3.9839,
      "step": 971776
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4080654470791024e-05,
      "loss": 3.9774,
      "step": 972288
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4072268523280504e-05,
      "loss": 3.9979,
      "step": 972800
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4063882575769984e-05,
      "loss": 3.9912,
      "step": 973312
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4055496628259463e-05,
      "loss": 3.9899,
      "step": 973824
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4047110680748943e-05,
      "loss": 3.9981,
      "step": 974336
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4038724733238423e-05,
      "loss": 3.9999,
      "step": 974848
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.40303387857279e-05,
      "loss": 3.991,
      "step": 975360
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.402198559582485e-05,
      "loss": 3.9894,
      "step": 975872
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.401359964831433e-05,
      "loss": 3.9867,
      "step": 976384
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.400521370080381e-05,
      "loss": 3.9903,
      "step": 976896
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.399682775329329e-05,
      "loss": 3.999,
      "step": 977408
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.398844180578277e-05,
      "loss": 3.9667,
      "step": 977920
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.398005585827225e-05,
      "loss": 3.9927,
      "step": 978432
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.397166991076173e-05,
      "loss": 3.9846,
      "step": 978944
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.396328396325121e-05,
      "loss": 3.9867,
      "step": 979456
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.395491439454442e-05,
      "loss": 3.9946,
      "step": 979968
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.39465284470339e-05,
      "loss": 3.9869,
      "step": 980480
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.393814249952338e-05,
      "loss": 3.9841,
      "step": 980992
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3929772930816586e-05,
      "loss": 3.9794,
      "step": 981504
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3921386983306066e-05,
      "loss": 4.0018,
      "step": 982016
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.391300103579555e-05,
      "loss": 3.9873,
      "step": 982528
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.390461508828503e-05,
      "loss": 3.9951,
      "step": 983040
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.389622914077451e-05,
      "loss": 3.982,
      "step": 983552
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.388784319326399e-05,
      "loss": 3.9966,
      "step": 984064
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.387945724575347e-05,
      "loss": 3.9954,
      "step": 984576
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.387107129824295e-05,
      "loss": 3.9895,
      "step": 985088
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.386270172953616e-05,
      "loss": 3.9929,
      "step": 985600
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.385431578202564e-05,
      "loss": 3.9849,
      "step": 986112
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.384594621331885e-05,
      "loss": 3.9924,
      "step": 986624
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.383756026580833e-05,
      "loss": 4.0053,
      "step": 987136
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.382917431829781e-05,
      "loss": 3.9853,
      "step": 987648
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.382078837078729e-05,
      "loss": 3.9839,
      "step": 988160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.381240242327677e-05,
      "loss": 3.9901,
      "step": 988672
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.380401647576625e-05,
      "loss": 3.9832,
      "step": 989184
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.379563052825573e-05,
      "loss": 3.9782,
      "step": 989696
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3787260959548946e-05,
      "loss": 3.9974,
      "step": 990208
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3778891390842155e-05,
      "loss": 3.9815,
      "step": 990720
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3770505443331635e-05,
      "loss": 3.9928,
      "step": 991232
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3762119495821115e-05,
      "loss": 3.9936,
      "step": 991744
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.038577079772949,
      "eval_runtime": 293.4714,
      "eval_samples_per_second": 1300.267,
      "eval_steps_per_second": 40.634,
      "step": 992160
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3753733548310595e-05,
      "loss": 3.9803,
      "step": 992256
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3745347600800075e-05,
      "loss": 3.9769,
      "step": 992768
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3736961653289555e-05,
      "loss": 3.9841,
      "step": 993280
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.372857570577903e-05,
      "loss": 4.0005,
      "step": 993792
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.372018975826851e-05,
      "loss": 3.9918,
      "step": 994304
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.371180381075799e-05,
      "loss": 3.9935,
      "step": 994816
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3703434242051204e-05,
      "loss": 3.9829,
      "step": 995328
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.369506467334442e-05,
      "loss": 3.9816,
      "step": 995840
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.36866787258339e-05,
      "loss": 3.9799,
      "step": 996352
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.367829277832338e-05,
      "loss": 3.9928,
      "step": 996864
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.366990683081285e-05,
      "loss": 3.9879,
      "step": 997376
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.366152088330233e-05,
      "loss": 4.0008,
      "step": 997888
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.365315131459555e-05,
      "loss": 4.0012,
      "step": 998400
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.364476536708503e-05,
      "loss": 3.9721,
      "step": 998912
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.36363794195745e-05,
      "loss": 3.9826,
      "step": 999424
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.362799347206398e-05,
      "loss": 3.9708,
      "step": 999936
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.361960752455346e-05,
      "loss": 3.994,
      "step": 1000448
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.361122157704294e-05,
      "loss": 3.9836,
      "step": 1000960
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.360283562953243e-05,
      "loss": 3.9798,
      "step": 1001472
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.359444968202191e-05,
      "loss": 3.9836,
      "step": 1001984
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.358606373451139e-05,
      "loss": 3.9946,
      "step": 1002496
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.35776941658046e-05,
      "loss": 3.988,
      "step": 1003008
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.356930821829408e-05,
      "loss": 3.988,
      "step": 1003520
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.356092227078356e-05,
      "loss": 3.9882,
      "step": 1004032
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.355253632327304e-05,
      "loss": 3.9882,
      "step": 1004544
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.354415037576252e-05,
      "loss": 3.9884,
      "step": 1005056
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3535764428252e-05,
      "loss": 3.9729,
      "step": 1005568
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.352739485954521e-05,
      "loss": 3.9914,
      "step": 1006080
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3519008912034687e-05,
      "loss": 3.9755,
      "step": 1006592
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3510622964524167e-05,
      "loss": 3.9789,
      "step": 1007104
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3502237017013646e-05,
      "loss": 3.987,
      "step": 1007616
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3493851069503126e-05,
      "loss": 3.9798,
      "step": 1008128
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3485465121992606e-05,
      "loss": 3.9841,
      "step": 1008640
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.347707917448209e-05,
      "loss": 3.9891,
      "step": 1009152
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.34687096057753e-05,
      "loss": 3.9835,
      "step": 1009664
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.346034003706851e-05,
      "loss": 3.9875,
      "step": 1010176
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.345195408955799e-05,
      "loss": 3.9834,
      "step": 1010688
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.344356814204747e-05,
      "loss": 3.9641,
      "step": 1011200
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.343518219453695e-05,
      "loss": 3.9823,
      "step": 1011712
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.342679624702643e-05,
      "loss": 3.9688,
      "step": 1012224
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.341841029951591e-05,
      "loss": 3.98,
      "step": 1012736
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3410024352005384e-05,
      "loss": 3.9793,
      "step": 1013248
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3401638404494864e-05,
      "loss": 3.9887,
      "step": 1013760
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.339326883578808e-05,
      "loss": 3.9807,
      "step": 1014272
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.338488288827756e-05,
      "loss": 3.9857,
      "step": 1014784
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.337649694076704e-05,
      "loss": 3.9813,
      "step": 1015296
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3368127372060256e-05,
      "loss": 3.983,
      "step": 1015808
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3359741424549736e-05,
      "loss": 3.9796,
      "step": 1016320
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3351355477039216e-05,
      "loss": 3.9655,
      "step": 1016832
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.334296952952869e-05,
      "loss": 3.9791,
      "step": 1017344
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.333458358201817e-05,
      "loss": 3.9871,
      "step": 1017856
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3326214013311385e-05,
      "loss": 3.9891,
      "step": 1018368
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.331782806580086e-05,
      "loss": 3.9782,
      "step": 1018880
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.330944211829034e-05,
      "loss": 3.9748,
      "step": 1019392
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.330105617077982e-05,
      "loss": 3.9728,
      "step": 1019904
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.32926702232693e-05,
      "loss": 3.9682,
      "step": 1020416
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3284300654562514e-05,
      "loss": 3.9974,
      "step": 1020928
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3275914707051994e-05,
      "loss": 3.9684,
      "step": 1021440
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3267528759541474e-05,
      "loss": 3.9916,
      "step": 1021952
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3259142812030954e-05,
      "loss": 3.9798,
      "step": 1022464
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3250756864520434e-05,
      "loss": 3.9646,
      "step": 1022976
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3242370917009914e-05,
      "loss": 3.9828,
      "step": 1023488
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3233984969499393e-05,
      "loss": 3.9713,
      "step": 1024000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3225599021988873e-05,
      "loss": 3.9656,
      "step": 1024512
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.321722945328208e-05,
      "loss": 3.9713,
      "step": 1025024
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.320885988457529e-05,
      "loss": 3.9837,
      "step": 1025536
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.320047393706477e-05,
      "loss": 3.9649,
      "step": 1026048
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.319208798955425e-05,
      "loss": 3.9801,
      "step": 1026560
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.318370204204374e-05,
      "loss": 3.9622,
      "step": 1027072
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.317531609453322e-05,
      "loss": 3.9729,
      "step": 1027584
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.31669301470227e-05,
      "loss": 3.987,
      "step": 1028096
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.315854419951218e-05,
      "loss": 3.9817,
      "step": 1028608
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.315015825200166e-05,
      "loss": 3.9813,
      "step": 1029120
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.314177230449114e-05,
      "loss": 3.9741,
      "step": 1029632
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.313338635698062e-05,
      "loss": 3.9874,
      "step": 1030144
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.312501678827383e-05,
      "loss": 3.973,
      "step": 1030656
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.311663084076331e-05,
      "loss": 3.9803,
      "step": 1031168
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.310824489325279e-05,
      "loss": 3.9761,
      "step": 1031680
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.309985894574227e-05,
      "loss": 3.9747,
      "step": 1032192
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.309147299823175e-05,
      "loss": 3.9808,
      "step": 1032704
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.308308705072122e-05,
      "loss": 3.9738,
      "step": 1033216
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.307470110321071e-05,
      "loss": 3.9831,
      "step": 1033728
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.306631515570019e-05,
      "loss": 3.9922,
      "step": 1034240
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.30579455869934e-05,
      "loss": 3.9793,
      "step": 1034752
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3049559639482876e-05,
      "loss": 3.9655,
      "step": 1035264
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3041173691972356e-05,
      "loss": 3.9717,
      "step": 1035776
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3032787744461836e-05,
      "loss": 3.9795,
      "step": 1036288
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.302441817575505e-05,
      "loss": 3.977,
      "step": 1036800
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3016032228244525e-05,
      "loss": 3.9762,
      "step": 1037312
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3007646280734005e-05,
      "loss": 3.9815,
      "step": 1037824
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2999260333223485e-05,
      "loss": 3.979,
      "step": 1038336
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2990890764516694e-05,
      "loss": 3.9791,
      "step": 1038848
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.298252119580991e-05,
      "loss": 3.9779,
      "step": 1039360
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.297413524829939e-05,
      "loss": 3.9838,
      "step": 1039872
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.296574930078887e-05,
      "loss": 3.9752,
      "step": 1040384
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.295736335327835e-05,
      "loss": 3.9903,
      "step": 1040896
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.294897740576783e-05,
      "loss": 3.9655,
      "step": 1041408
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.294059145825731e-05,
      "loss": 3.9843,
      "step": 1041920
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.293220551074679e-05,
      "loss": 3.9818,
      "step": 1042432
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.292381956323627e-05,
      "loss": 3.9816,
      "step": 1042944
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.291543361572575e-05,
      "loss": 3.9856,
      "step": 1043456
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.290704766821523e-05,
      "loss": 3.9747,
      "step": 1043968
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.289866172070471e-05,
      "loss": 3.9772,
      "step": 1044480
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.289029215199792e-05,
      "loss": 3.9768,
      "step": 1044992
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.28819062044874e-05,
      "loss": 3.9741,
      "step": 1045504
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.287352025697688e-05,
      "loss": 3.9624,
      "step": 1046016
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.286513430946636e-05,
      "loss": 3.9706,
      "step": 1046528
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.285674836195584e-05,
      "loss": 3.9747,
      "step": 1047040
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2848362414445325e-05,
      "loss": 3.9764,
      "step": 1047552
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2839976466934805e-05,
      "loss": 3.97,
      "step": 1048064
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2831590519424285e-05,
      "loss": 3.9658,
      "step": 1048576
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2823204571913765e-05,
      "loss": 3.9827,
      "step": 1049088
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.28148513820107e-05,
      "loss": 3.9783,
      "step": 1049600
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.280646543450018e-05,
      "loss": 3.9758,
      "step": 1050112
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.279807948698966e-05,
      "loss": 3.9865,
      "step": 1050624
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.278969353947914e-05,
      "loss": 3.9829,
      "step": 1051136
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.278130759196862e-05,
      "loss": 3.9846,
      "step": 1051648
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.27729216444581e-05,
      "loss": 3.9739,
      "step": 1052160
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.276455207575131e-05,
      "loss": 3.9722,
      "step": 1052672
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.275616612824079e-05,
      "loss": 3.9742,
      "step": 1053184
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.274778018073028e-05,
      "loss": 3.9866,
      "step": 1053696
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.273939423321976e-05,
      "loss": 3.9573,
      "step": 1054208
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.273100828570924e-05,
      "loss": 3.9767,
      "step": 1054720
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.272262233819871e-05,
      "loss": 3.9682,
      "step": 1055232
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.271423639068819e-05,
      "loss": 3.9767,
      "step": 1055744
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.270585044317767e-05,
      "loss": 3.9821,
      "step": 1056256
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.269748087447088e-05,
      "loss": 3.9745,
      "step": 1056768
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2689111305764097e-05,
      "loss": 3.9675,
      "step": 1057280
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2680725358253576e-05,
      "loss": 3.9672,
      "step": 1057792
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2672339410743056e-05,
      "loss": 3.9876,
      "step": 1058304
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.266395346323253e-05,
      "loss": 3.9743,
      "step": 1058816
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2655583894525745e-05,
      "loss": 3.9849,
      "step": 1059328
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.264719794701523e-05,
      "loss": 3.967,
      "step": 1059840
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.263881199950471e-05,
      "loss": 3.9852,
      "step": 1060352
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2630426051994185e-05,
      "loss": 3.9824,
      "step": 1060864
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2622040104483665e-05,
      "loss": 3.9783,
      "step": 1061376
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2613654156973145e-05,
      "loss": 3.9742,
      "step": 1061888
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2605268209462625e-05,
      "loss": 3.9757,
      "step": 1062400
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2596882261952105e-05,
      "loss": 3.9802,
      "step": 1062912
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2588512693245314e-05,
      "loss": 3.9854,
      "step": 1063424
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2580126745734794e-05,
      "loss": 3.9774,
      "step": 1063936
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2571740798224274e-05,
      "loss": 3.9691,
      "step": 1064448
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2563354850713754e-05,
      "loss": 3.9786,
      "step": 1064960
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2554968903203234e-05,
      "loss": 3.9672,
      "step": 1065472
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.254659933449645e-05,
      "loss": 3.966,
      "step": 1065984
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.253821338698593e-05,
      "loss": 3.9831,
      "step": 1066496
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.252982743947541e-05,
      "loss": 3.9677,
      "step": 1067008
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.252144149196489e-05,
      "loss": 3.9835,
      "step": 1067520
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.25130719232581e-05,
      "loss": 3.9748,
      "step": 1068032
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.031233787536621,
      "eval_runtime": 293.9518,
      "eval_samples_per_second": 1298.142,
      "eval_steps_per_second": 40.568,
      "step": 1068480
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.250468597574758e-05,
      "loss": 3.9808,
      "step": 1068544
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.249630002823706e-05,
      "loss": 3.9686,
      "step": 1069056
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.248791408072654e-05,
      "loss": 3.9704,
      "step": 1069568
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.247952813321602e-05,
      "loss": 3.9866,
      "step": 1070080
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.24711421857055e-05,
      "loss": 3.9813,
      "step": 1070592
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.246275623819498e-05,
      "loss": 3.9776,
      "step": 1071104
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.245437029068446e-05,
      "loss": 3.9691,
      "step": 1071616
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.244598434317394e-05,
      "loss": 3.9689,
      "step": 1072128
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.243759839566342e-05,
      "loss": 3.966,
      "step": 1072640
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.24292124481529e-05,
      "loss": 3.9756,
      "step": 1073152
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.242082650064238e-05,
      "loss": 3.9776,
      "step": 1073664
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.241244055313186e-05,
      "loss": 3.986,
      "step": 1074176
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2404070984425074e-05,
      "loss": 3.9905,
      "step": 1074688
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.239568503691455e-05,
      "loss": 3.9601,
      "step": 1075200
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.238729908940403e-05,
      "loss": 3.9701,
      "step": 1075712
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.237891314189351e-05,
      "loss": 3.9617,
      "step": 1076224
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.237052719438299e-05,
      "loss": 3.9781,
      "step": 1076736
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2362157625676196e-05,
      "loss": 3.9705,
      "step": 1077248
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2353771678165676e-05,
      "loss": 3.9678,
      "step": 1077760
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2345385730655156e-05,
      "loss": 3.9682,
      "step": 1078272
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2336999783144636e-05,
      "loss": 3.9815,
      "step": 1078784
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2328613835634116e-05,
      "loss": 3.9728,
      "step": 1079296
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.23202278881236e-05,
      "loss": 3.9771,
      "step": 1079808
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.231184194061308e-05,
      "loss": 3.975,
      "step": 1080320
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.230345599310256e-05,
      "loss": 3.9789,
      "step": 1080832
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.229507004559204e-05,
      "loss": 3.9726,
      "step": 1081344
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.228668409808152e-05,
      "loss": 3.9617,
      "step": 1081856
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2278298150571e-05,
      "loss": 3.9744,
      "step": 1082368
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.226992858186421e-05,
      "loss": 3.9628,
      "step": 1082880
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.226154263435369e-05,
      "loss": 3.9668,
      "step": 1083392
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.225315668684317e-05,
      "loss": 3.9709,
      "step": 1083904
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.224478711813638e-05,
      "loss": 3.9736,
      "step": 1084416
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.223640117062586e-05,
      "loss": 3.9694,
      "step": 1084928
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.222801522311534e-05,
      "loss": 3.9762,
      "step": 1085440
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.221962927560482e-05,
      "loss": 3.9715,
      "step": 1085952
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2211259706898037e-05,
      "loss": 3.9713,
      "step": 1086464
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2202873759387517e-05,
      "loss": 3.9724,
      "step": 1086976
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2194487811876996e-05,
      "loss": 3.9522,
      "step": 1087488
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2186101864366476e-05,
      "loss": 3.9668,
      "step": 1088000
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2177715916855956e-05,
      "loss": 3.9594,
      "step": 1088512
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.216932996934543e-05,
      "loss": 3.9639,
      "step": 1089024
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.216094402183491e-05,
      "loss": 3.9657,
      "step": 1089536
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.215255807432439e-05,
      "loss": 3.9761,
      "step": 1090048
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.214417212681387e-05,
      "loss": 3.9666,
      "step": 1090560
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.213578617930335e-05,
      "loss": 3.973,
      "step": 1091072
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.212740023179283e-05,
      "loss": 3.9709,
      "step": 1091584
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.211901428428231e-05,
      "loss": 3.9711,
      "step": 1092096
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.211062833677179e-05,
      "loss": 3.9664,
      "step": 1092608
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2102258768065005e-05,
      "loss": 3.9545,
      "step": 1093120
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2093889199358214e-05,
      "loss": 3.9642,
      "step": 1093632
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2085503251847694e-05,
      "loss": 3.9746,
      "step": 1094144
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2077117304337174e-05,
      "loss": 3.9758,
      "step": 1094656
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2068731356826654e-05,
      "loss": 3.9668,
      "step": 1095168
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.206036178811986e-05,
      "loss": 3.9574,
      "step": 1095680
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.205197584060934e-05,
      "loss": 3.9644,
      "step": 1096192
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.204358989309882e-05,
      "loss": 3.9518,
      "step": 1096704
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.20352039455883e-05,
      "loss": 3.9833,
      "step": 1097216
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.202681799807778e-05,
      "loss": 3.9536,
      "step": 1097728
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.201843205056726e-05,
      "loss": 3.9778,
      "step": 1098240
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.201006248186048e-05,
      "loss": 3.9652,
      "step": 1098752
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.200167653434996e-05,
      "loss": 3.9542,
      "step": 1099264
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.199329058683944e-05,
      "loss": 3.9649,
      "step": 1099776
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.198490463932892e-05,
      "loss": 3.9646,
      "step": 1100288
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.19765186918184e-05,
      "loss": 3.9527,
      "step": 1100800
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.196814912311161e-05,
      "loss": 3.9525,
      "step": 1101312
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.195976317560109e-05,
      "loss": 3.9766,
      "step": 1101824
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.195137722809057e-05,
      "loss": 3.9566,
      "step": 1102336
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.194299128058005e-05,
      "loss": 3.9611,
      "step": 1102848
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.193460533306953e-05,
      "loss": 3.9497,
      "step": 1103360
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.192621938555901e-05,
      "loss": 3.9613,
      "step": 1103872
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.191784981685222e-05,
      "loss": 3.9734,
      "step": 1104384
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1909480248145426e-05,
      "loss": 3.9675,
      "step": 1104896
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.190111067943864e-05,
      "loss": 3.9648,
      "step": 1105408
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.189272473192812e-05,
      "loss": 3.9662,
      "step": 1105920
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.18843387844176e-05,
      "loss": 3.9734,
      "step": 1106432
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.187595283690708e-05,
      "loss": 3.9636,
      "step": 1106944
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.186756688939656e-05,
      "loss": 3.9636,
      "step": 1107456
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.185918094188604e-05,
      "loss": 3.9648,
      "step": 1107968
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.185079499437552e-05,
      "loss": 3.965,
      "step": 1108480
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1842409046865e-05,
      "loss": 3.965,
      "step": 1108992
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.183402309935448e-05,
      "loss": 3.9637,
      "step": 1109504
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.182563715184396e-05,
      "loss": 3.9683,
      "step": 1110016
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.181725120433344e-05,
      "loss": 3.9767,
      "step": 1110528
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1808865256822914e-05,
      "loss": 3.9686,
      "step": 1111040
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1800479309312394e-05,
      "loss": 3.9536,
      "step": 1111552
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.179210974060561e-05,
      "loss": 3.9602,
      "step": 1112064
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.178372379309509e-05,
      "loss": 3.9699,
      "step": 1112576
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1775354224388306e-05,
      "loss": 3.9635,
      "step": 1113088
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1766968276877786e-05,
      "loss": 3.9657,
      "step": 1113600
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1758582329367266e-05,
      "loss": 3.9699,
      "step": 1114112
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.175019638185674e-05,
      "loss": 3.968,
      "step": 1114624
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.174181043434622e-05,
      "loss": 3.9653,
      "step": 1115136
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.17334244868357e-05,
      "loss": 3.9625,
      "step": 1115648
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.172503853932518e-05,
      "loss": 3.972,
      "step": 1116160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.171665259181466e-05,
      "loss": 3.963,
      "step": 1116672
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.170826664430414e-05,
      "loss": 3.977,
      "step": 1117184
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.169988069679362e-05,
      "loss": 3.9516,
      "step": 1117696
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.16914947492831e-05,
      "loss": 3.97,
      "step": 1118208
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.168310880177258e-05,
      "loss": 3.9704,
      "step": 1118720
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1674739233065795e-05,
      "loss": 3.9695,
      "step": 1119232
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1666353285555275e-05,
      "loss": 3.9718,
      "step": 1119744
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1657983716848484e-05,
      "loss": 3.9652,
      "step": 1120256
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1649597769337964e-05,
      "loss": 3.9639,
      "step": 1120768
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1641211821827444e-05,
      "loss": 3.9671,
      "step": 1121280
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1632825874316924e-05,
      "loss": 3.9576,
      "step": 1121792
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.162445630561013e-05,
      "loss": 3.9575,
      "step": 1122304
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.161607035809961e-05,
      "loss": 3.9537,
      "step": 1122816
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.160768441058909e-05,
      "loss": 3.9633,
      "step": 1123328
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.159929846307857e-05,
      "loss": 3.9654,
      "step": 1123840
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.159091251556805e-05,
      "loss": 3.9547,
      "step": 1124352
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.158252656805753e-05,
      "loss": 3.9497,
      "step": 1124864
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.157414062054702e-05,
      "loss": 3.9673,
      "step": 1125376
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.15657546730365e-05,
      "loss": 3.9717,
      "step": 1125888
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.155738510432971e-05,
      "loss": 3.9684,
      "step": 1126400
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.154899915681919e-05,
      "loss": 3.9682,
      "step": 1126912
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.154061320930867e-05,
      "loss": 3.9705,
      "step": 1127424
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.153222726179815e-05,
      "loss": 3.9695,
      "step": 1127936
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.152385769309136e-05,
      "loss": 3.9626,
      "step": 1128448
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.151547174558084e-05,
      "loss": 3.9604,
      "step": 1128960
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.150708579807032e-05,
      "loss": 3.9629,
      "step": 1129472
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.14986998505598e-05,
      "loss": 3.9729,
      "step": 1129984
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1490330281853006e-05,
      "loss": 3.9429,
      "step": 1130496
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1481944334342486e-05,
      "loss": 3.9664,
      "step": 1131008
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.147355838683197e-05,
      "loss": 3.9551,
      "step": 1131520
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.146517243932145e-05,
      "loss": 3.9636,
      "step": 1132032
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1456786491810926e-05,
      "loss": 3.9651,
      "step": 1132544
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1448400544300406e-05,
      "loss": 3.9647,
      "step": 1133056
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1440014596789886e-05,
      "loss": 3.9521,
      "step": 1133568
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1431628649279366e-05,
      "loss": 3.9566,
      "step": 1134080
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1423242701768846e-05,
      "loss": 3.9694,
      "step": 1134592
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1414856754258326e-05,
      "loss": 3.9648,
      "step": 1135104
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1406487185551535e-05,
      "loss": 3.9693,
      "step": 1135616
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1398101238041015e-05,
      "loss": 3.9591,
      "step": 1136128
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1389731669334224e-05,
      "loss": 3.9703,
      "step": 1136640
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.138134572182371e-05,
      "loss": 3.9705,
      "step": 1137152
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.137295977431319e-05,
      "loss": 3.965,
      "step": 1137664
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.13645902056064e-05,
      "loss": 3.9633,
      "step": 1138176
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.135620425809588e-05,
      "loss": 3.9626,
      "step": 1138688
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.134781831058536e-05,
      "loss": 3.971,
      "step": 1139200
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.133943236307484e-05,
      "loss": 3.9727,
      "step": 1139712
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.133104641556432e-05,
      "loss": 3.9615,
      "step": 1140224
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.13226604680538e-05,
      "loss": 3.9543,
      "step": 1140736
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.131427452054328e-05,
      "loss": 3.9667,
      "step": 1141248
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.130588857303276e-05,
      "loss": 3.9553,
      "step": 1141760
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.129751900432597e-05,
      "loss": 3.9493,
      "step": 1142272
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.128913305681545e-05,
      "loss": 3.9733,
      "step": 1142784
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.128076348810866e-05,
      "loss": 3.9576,
      "step": 1143296
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1272377540598144e-05,
      "loss": 3.9708,
      "step": 1143808
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1263991593087624e-05,
      "loss": 3.961,
      "step": 1144320
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.024241924285889,
      "eval_runtime": 292.5692,
      "eval_samples_per_second": 1304.276,
      "eval_steps_per_second": 40.76,
      "step": 1144800
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1255605645577104e-05,
      "loss": 3.9515,
      "step": 1144832
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1247219698066584e-05,
      "loss": 3.9588,
      "step": 1145344
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1238833750556064e-05,
      "loss": 3.958,
      "step": 1145856
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1230447803045544e-05,
      "loss": 3.9695,
      "step": 1146368
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1222061855535024e-05,
      "loss": 3.969,
      "step": 1146880
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.121369228682823e-05,
      "loss": 3.9654,
      "step": 1147392
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.120530633931771e-05,
      "loss": 3.9582,
      "step": 1147904
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.119692039180719e-05,
      "loss": 3.9535,
      "step": 1148416
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.11885508231004e-05,
      "loss": 3.9543,
      "step": 1148928
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.118016487558988e-05,
      "loss": 3.9644,
      "step": 1149440
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.117177892807936e-05,
      "loss": 3.9599,
      "step": 1149952
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.116339298056884e-05,
      "loss": 3.9784,
      "step": 1150464
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.115502341186206e-05,
      "loss": 3.9757,
      "step": 1150976
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.114663746435154e-05,
      "loss": 3.9493,
      "step": 1151488
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.113825151684102e-05,
      "loss": 3.9537,
      "step": 1152000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.11298655693305e-05,
      "loss": 3.9486,
      "step": 1152512
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.112147962181998e-05,
      "loss": 3.9631,
      "step": 1153024
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.111309367430946e-05,
      "loss": 3.96,
      "step": 1153536
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.110470772679894e-05,
      "loss": 3.9545,
      "step": 1154048
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.109632177928841e-05,
      "loss": 3.9507,
      "step": 1154560
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.108793583177789e-05,
      "loss": 3.9736,
      "step": 1155072
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.107954988426737e-05,
      "loss": 3.9561,
      "step": 1155584
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.107118031556058e-05,
      "loss": 3.9648,
      "step": 1156096
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1062794368050066e-05,
      "loss": 3.9643,
      "step": 1156608
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1054408420539546e-05,
      "loss": 3.9708,
      "step": 1157120
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1046022473029026e-05,
      "loss": 3.9538,
      "step": 1157632
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1037636525518506e-05,
      "loss": 3.9524,
      "step": 1158144
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1029250578007986e-05,
      "loss": 3.9566,
      "step": 1158656
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1020864630497466e-05,
      "loss": 3.9606,
      "step": 1159168
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1012478682986946e-05,
      "loss": 3.9526,
      "step": 1159680
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.1004109114280155e-05,
      "loss": 3.9552,
      "step": 1160192
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0995723166769635e-05,
      "loss": 3.9604,
      "step": 1160704
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0987337219259115e-05,
      "loss": 3.9582,
      "step": 1161216
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0978951271748595e-05,
      "loss": 3.9629,
      "step": 1161728
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0970565324238075e-05,
      "loss": 3.9583,
      "step": 1162240
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.096221213433502e-05,
      "loss": 3.96,
      "step": 1162752
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.09538261868245e-05,
      "loss": 3.9557,
      "step": 1163264
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.094544023931398e-05,
      "loss": 3.9423,
      "step": 1163776
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.093705429180346e-05,
      "loss": 3.955,
      "step": 1164288
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.092868472309667e-05,
      "loss": 3.9441,
      "step": 1164800
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.092029877558615e-05,
      "loss": 3.9546,
      "step": 1165312
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.091191282807563e-05,
      "loss": 3.9531,
      "step": 1165824
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.090352688056511e-05,
      "loss": 3.9659,
      "step": 1166336
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.089514093305459e-05,
      "loss": 3.9547,
      "step": 1166848
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.088675498554407e-05,
      "loss": 3.9589,
      "step": 1167360
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.087836903803355e-05,
      "loss": 3.9623,
      "step": 1167872
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.086998309052303e-05,
      "loss": 3.9568,
      "step": 1168384
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.086159714301251e-05,
      "loss": 3.9544,
      "step": 1168896
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.085322757430572e-05,
      "loss": 3.9414,
      "step": 1169408
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.08448416267952e-05,
      "loss": 3.9539,
      "step": 1169920
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0836455679284685e-05,
      "loss": 3.9589,
      "step": 1170432
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0828086110577894e-05,
      "loss": 3.9599,
      "step": 1170944
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0819700163067374e-05,
      "loss": 3.9583,
      "step": 1171456
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.081133059436058e-05,
      "loss": 3.9439,
      "step": 1171968
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.080294464685006e-05,
      "loss": 3.9545,
      "step": 1172480
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.079455869933954e-05,
      "loss": 3.9399,
      "step": 1172992
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.078617275182902e-05,
      "loss": 3.9701,
      "step": 1173504
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.07777868043185e-05,
      "loss": 3.9425,
      "step": 1174016
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.076940085680798e-05,
      "loss": 3.9709,
      "step": 1174528
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.076101490929746e-05,
      "loss": 3.9526,
      "step": 1175040
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.075262896178694e-05,
      "loss": 3.9389,
      "step": 1175552
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.074424301427642e-05,
      "loss": 3.9539,
      "step": 1176064
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.07358570667659e-05,
      "loss": 3.9545,
      "step": 1176576
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.072747111925538e-05,
      "loss": 3.939,
      "step": 1177088
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.071908517174486e-05,
      "loss": 3.9402,
      "step": 1177600
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.071069922423434e-05,
      "loss": 3.9637,
      "step": 1178112
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.070231327672382e-05,
      "loss": 3.9464,
      "step": 1178624
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.069394370801703e-05,
      "loss": 3.9459,
      "step": 1179136
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.068555776050651e-05,
      "loss": 3.9385,
      "step": 1179648
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.067717181299599e-05,
      "loss": 3.9488,
      "step": 1180160
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.06688022442892e-05,
      "loss": 3.9618,
      "step": 1180672
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0660432675582416e-05,
      "loss": 3.9576,
      "step": 1181184
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.065204672807189e-05,
      "loss": 3.9501,
      "step": 1181696
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0643660780561376e-05,
      "loss": 3.9554,
      "step": 1182208
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0635274833050856e-05,
      "loss": 3.9589,
      "step": 1182720
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0626888885540336e-05,
      "loss": 3.95,
      "step": 1183232
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0618502938029816e-05,
      "loss": 3.9534,
      "step": 1183744
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0610116990519296e-05,
      "loss": 3.9555,
      "step": 1184256
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0601731043008776e-05,
      "loss": 3.9516,
      "step": 1184768
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0593345095498256e-05,
      "loss": 3.9534,
      "step": 1185280
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0584975526791465e-05,
      "loss": 3.9526,
      "step": 1185792
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0576605958084674e-05,
      "loss": 3.9552,
      "step": 1186304
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0568220010574154e-05,
      "loss": 3.9675,
      "step": 1186816
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0559834063063634e-05,
      "loss": 3.9549,
      "step": 1187328
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0551448115553114e-05,
      "loss": 3.9418,
      "step": 1187840
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0543062168042594e-05,
      "loss": 3.9465,
      "step": 1188352
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0534676220532074e-05,
      "loss": 3.9572,
      "step": 1188864
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.052630665182529e-05,
      "loss": 3.955,
      "step": 1189376
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.051792070431477e-05,
      "loss": 3.9485,
      "step": 1189888
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.050953475680425e-05,
      "loss": 3.958,
      "step": 1190400
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.050114880929373e-05,
      "loss": 3.9584,
      "step": 1190912
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.049277924058694e-05,
      "loss": 3.951,
      "step": 1191424
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.048439329307642e-05,
      "loss": 3.9523,
      "step": 1191936
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.04760073455659e-05,
      "loss": 3.9633,
      "step": 1192448
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.046762139805538e-05,
      "loss": 3.9459,
      "step": 1192960
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.045923545054486e-05,
      "loss": 3.9646,
      "step": 1193472
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.045086588183807e-05,
      "loss": 3.9465,
      "step": 1193984
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.044247993432755e-05,
      "loss": 3.958,
      "step": 1194496
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.043409398681703e-05,
      "loss": 3.9606,
      "step": 1195008
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.042570803930651e-05,
      "loss": 3.9538,
      "step": 1195520
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.041733847059972e-05,
      "loss": 3.9646,
      "step": 1196032
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.04089525230892e-05,
      "loss": 3.9507,
      "step": 1196544
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0400566575578683e-05,
      "loss": 3.9507,
      "step": 1197056
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0392180628068163e-05,
      "loss": 3.9506,
      "step": 1197568
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0383794680557643e-05,
      "loss": 3.9507,
      "step": 1198080
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0375408733047123e-05,
      "loss": 3.9443,
      "step": 1198592
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0367022785536603e-05,
      "loss": 3.9439,
      "step": 1199104
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0358636838026076e-05,
      "loss": 3.9562,
      "step": 1199616
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0350267269319292e-05,
      "loss": 3.9519,
      "step": 1200128
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0341881321808775e-05,
      "loss": 3.9437,
      "step": 1200640
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0333511753101984e-05,
      "loss": 3.9377,
      "step": 1201152
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0325125805591464e-05,
      "loss": 3.9519,
      "step": 1201664
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0316739858080944e-05,
      "loss": 3.9609,
      "step": 1202176
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0308353910570424e-05,
      "loss": 3.9524,
      "step": 1202688
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0299967963059904e-05,
      "loss": 3.9569,
      "step": 1203200
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.029158201554938e-05,
      "loss": 3.9553,
      "step": 1203712
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.028319606803886e-05,
      "loss": 3.9666,
      "step": 1204224
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.027481012052834e-05,
      "loss": 3.9489,
      "step": 1204736
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0266456930625286e-05,
      "loss": 3.95,
      "step": 1205248
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0258070983114766e-05,
      "loss": 3.9541,
      "step": 1205760
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0249685035604246e-05,
      "loss": 3.9576,
      "step": 1206272
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.024129908809373e-05,
      "loss": 3.9359,
      "step": 1206784
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0232913140583202e-05,
      "loss": 3.9489,
      "step": 1207296
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0224527193072682e-05,
      "loss": 3.9458,
      "step": 1207808
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0216157624365898e-05,
      "loss": 3.9478,
      "step": 1208320
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0207771676855378e-05,
      "loss": 3.951,
      "step": 1208832
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0199385729344855e-05,
      "loss": 3.9603,
      "step": 1209344
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0190999781834334e-05,
      "loss": 3.942,
      "step": 1209856
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0182613834323814e-05,
      "loss": 3.9475,
      "step": 1210368
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0174227886813294e-05,
      "loss": 3.9505,
      "step": 1210880
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0165841939302774e-05,
      "loss": 3.9594,
      "step": 1211392
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0157455991792254e-05,
      "loss": 3.9559,
      "step": 1211904
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0149086423085467e-05,
      "loss": 3.9485,
      "step": 1212416
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0140700475574947e-05,
      "loss": 3.957,
      "step": 1212928
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0132314528064427e-05,
      "loss": 3.9609,
      "step": 1213440
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0123928580553907e-05,
      "loss": 3.9569,
      "step": 1213952
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0115559011847116e-05,
      "loss": 3.9497,
      "step": 1214464
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0107173064336596e-05,
      "loss": 3.9517,
      "step": 1214976
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0098787116826076e-05,
      "loss": 3.9611,
      "step": 1215488
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.009040116931556e-05,
      "loss": 3.957,
      "step": 1216000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.008201522180504e-05,
      "loss": 3.9553,
      "step": 1216512
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0073645653098248e-05,
      "loss": 3.944,
      "step": 1217024
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0065276084391457e-05,
      "loss": 3.9533,
      "step": 1217536
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0056890136880937e-05,
      "loss": 3.9409,
      "step": 1218048
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.004850418937042e-05,
      "loss": 3.9396,
      "step": 1218560
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.00401182418599e-05,
      "loss": 3.9597,
      "step": 1219072
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.003173229434938e-05,
      "loss": 3.9479,
      "step": 1219584
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.002334634683886e-05,
      "loss": 3.9555,
      "step": 1220096
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.001496039932834e-05,
      "loss": 3.9533,
      "step": 1220608
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.000657445181782e-05,
      "loss": 3.9432,
      "step": 1221120
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.018515586853027,
      "eval_runtime": 291.5041,
      "eval_samples_per_second": 1309.042,
      "eval_steps_per_second": 40.909,
      "step": 1221120
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.99981885043073e-05,
      "loss": 3.9488,
      "step": 1221632
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.998980255679678e-05,
      "loss": 3.9457,
      "step": 1222144
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.998141660928626e-05,
      "loss": 3.9615,
      "step": 1222656
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9973030661775737e-05,
      "loss": 3.9558,
      "step": 1223168
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9964644714265217e-05,
      "loss": 3.9564,
      "step": 1223680
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9956258766754697e-05,
      "loss": 3.9486,
      "step": 1224192
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9947872819244177e-05,
      "loss": 3.945,
      "step": 1224704
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9939486871733656e-05,
      "loss": 3.9392,
      "step": 1225216
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9931100924223136e-05,
      "loss": 3.9568,
      "step": 1225728
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9922714976712616e-05,
      "loss": 3.9505,
      "step": 1226240
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.99143290292021e-05,
      "loss": 3.9652,
      "step": 1226752
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.990594308169158e-05,
      "loss": 3.9606,
      "step": 1227264
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.989757351298479e-05,
      "loss": 3.9441,
      "step": 1227776
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.988918756547427e-05,
      "loss": 3.9434,
      "step": 1228288
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.988080161796375e-05,
      "loss": 3.9424,
      "step": 1228800
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.987241567045323e-05,
      "loss": 3.9451,
      "step": 1229312
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.986402972294271e-05,
      "loss": 3.9501,
      "step": 1229824
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9855643775432192e-05,
      "loss": 3.9416,
      "step": 1230336
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9847257827921672e-05,
      "loss": 3.9425,
      "step": 1230848
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9838871880411152e-05,
      "loss": 3.9662,
      "step": 1231360
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9830485932900625e-05,
      "loss": 3.942,
      "step": 1231872
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9822099985390105e-05,
      "loss": 3.959,
      "step": 1232384
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9813714037879585e-05,
      "loss": 3.9532,
      "step": 1232896
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.98053444691728e-05,
      "loss": 3.956,
      "step": 1233408
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9796974900466013e-05,
      "loss": 3.942,
      "step": 1233920
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9788588952955493e-05,
      "loss": 3.9424,
      "step": 1234432
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9780203005444973e-05,
      "loss": 3.9424,
      "step": 1234944
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9771817057934453e-05,
      "loss": 3.9491,
      "step": 1235456
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.976343111042393e-05,
      "loss": 3.941,
      "step": 1235968
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9755061541717146e-05,
      "loss": 3.9467,
      "step": 1236480
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9746675594206626e-05,
      "loss": 3.951,
      "step": 1236992
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.97382896466961e-05,
      "loss": 3.9485,
      "step": 1237504
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.972990369918558e-05,
      "loss": 3.9513,
      "step": 1238016
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.972151775167506e-05,
      "loss": 3.9504,
      "step": 1238528
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9713164561772007e-05,
      "loss": 3.9488,
      "step": 1239040
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9704778614261487e-05,
      "loss": 3.9443,
      "step": 1239552
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9696392666750967e-05,
      "loss": 3.9307,
      "step": 1240064
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9688006719240447e-05,
      "loss": 3.9481,
      "step": 1240576
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9679620771729927e-05,
      "loss": 3.9326,
      "step": 1241088
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.96712348242194e-05,
      "loss": 3.9432,
      "step": 1241600
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9662848876708883e-05,
      "loss": 3.9425,
      "step": 1242112
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9654462929198363e-05,
      "loss": 3.9536,
      "step": 1242624
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9646076981687843e-05,
      "loss": 3.9478,
      "step": 1243136
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9637707412981052e-05,
      "loss": 3.9433,
      "step": 1243648
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9629321465470532e-05,
      "loss": 3.953,
      "step": 1244160
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9620935517960012e-05,
      "loss": 3.9479,
      "step": 1244672
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9612549570449492e-05,
      "loss": 3.9449,
      "step": 1245184
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9604163622938976e-05,
      "loss": 3.935,
      "step": 1245696
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9595777675428456e-05,
      "loss": 3.9387,
      "step": 1246208
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9587408106721665e-05,
      "loss": 3.9524,
      "step": 1246720
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9579022159211145e-05,
      "loss": 3.9464,
      "step": 1247232
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9570636211700625e-05,
      "loss": 3.9457,
      "step": 1247744
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9562250264190105e-05,
      "loss": 3.9348,
      "step": 1248256
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9553864316679584e-05,
      "loss": 3.9421,
      "step": 1248768
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9545494747972797e-05,
      "loss": 3.9289,
      "step": 1249280
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9537108800462277e-05,
      "loss": 3.9567,
      "step": 1249792
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9528722852951757e-05,
      "loss": 3.9353,
      "step": 1250304
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9520336905441237e-05,
      "loss": 3.955,
      "step": 1250816
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9511950957930717e-05,
      "loss": 3.944,
      "step": 1251328
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9503565010420197e-05,
      "loss": 3.9318,
      "step": 1251840
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9495179062909677e-05,
      "loss": 3.9414,
      "step": 1252352
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9486793115399157e-05,
      "loss": 3.9452,
      "step": 1252864
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.947840716788864e-05,
      "loss": 3.9268,
      "step": 1253376
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9470021220378113e-05,
      "loss": 3.9284,
      "step": 1253888
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9461668030475058e-05,
      "loss": 3.9531,
      "step": 1254400
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9453282082964538e-05,
      "loss": 3.9407,
      "step": 1254912
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.944489613545402e-05,
      "loss": 3.9341,
      "step": 1255424
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.94365101879435e-05,
      "loss": 3.9262,
      "step": 1255936
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.942814061923671e-05,
      "loss": 3.9398,
      "step": 1256448
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.941975467172619e-05,
      "loss": 3.9519,
      "step": 1256960
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.941136872421567e-05,
      "loss": 3.9473,
      "step": 1257472
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.940298277670515e-05,
      "loss": 3.9383,
      "step": 1257984
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.939459682919463e-05,
      "loss": 3.9447,
      "step": 1258496
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.938621088168411e-05,
      "loss": 3.9487,
      "step": 1259008
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9377824934173587e-05,
      "loss": 3.9406,
      "step": 1259520
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9369455365466803e-05,
      "loss": 3.9391,
      "step": 1260032
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9361085796760012e-05,
      "loss": 3.9454,
      "step": 1260544
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9352699849249492e-05,
      "loss": 3.939,
      "step": 1261056
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9344313901738972e-05,
      "loss": 3.9448,
      "step": 1261568
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9335927954228455e-05,
      "loss": 3.9408,
      "step": 1262080
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9327542006717935e-05,
      "loss": 3.9449,
      "step": 1262592
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9319156059207408e-05,
      "loss": 3.955,
      "step": 1263104
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9310770111696888e-05,
      "loss": 3.9462,
      "step": 1263616
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9302384164186368e-05,
      "loss": 3.9323,
      "step": 1264128
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9293998216675848e-05,
      "loss": 3.9378,
      "step": 1264640
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.928561226916533e-05,
      "loss": 3.944,
      "step": 1265152
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.927722632165481e-05,
      "loss": 3.9477,
      "step": 1265664
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.926884037414429e-05,
      "loss": 3.9435,
      "step": 1266176
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.92604708054375e-05,
      "loss": 3.9444,
      "step": 1266688
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.925208485792698e-05,
      "loss": 3.9465,
      "step": 1267200
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.924369891041646e-05,
      "loss": 3.9439,
      "step": 1267712
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.923531296290594e-05,
      "loss": 3.9419,
      "step": 1268224
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9226927015395424e-05,
      "loss": 3.9454,
      "step": 1268736
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9218541067884904e-05,
      "loss": 3.9396,
      "step": 1269248
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9210155120374384e-05,
      "loss": 3.9523,
      "step": 1269760
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9201769172863864e-05,
      "loss": 3.9418,
      "step": 1270272
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9193399604157073e-05,
      "loss": 3.9435,
      "step": 1270784
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9185030035450285e-05,
      "loss": 3.9488,
      "step": 1271296
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9176644087939765e-05,
      "loss": 3.9457,
      "step": 1271808
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9168258140429245e-05,
      "loss": 3.951,
      "step": 1272320
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9159872192918725e-05,
      "loss": 3.9431,
      "step": 1272832
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9151486245408205e-05,
      "loss": 3.9386,
      "step": 1273344
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9143100297897685e-05,
      "loss": 3.9409,
      "step": 1273856
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9134714350387165e-05,
      "loss": 3.9425,
      "step": 1274368
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9126328402876645e-05,
      "loss": 3.9304,
      "step": 1274880
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9117958834169857e-05,
      "loss": 3.9285,
      "step": 1275392
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9109572886659337e-05,
      "loss": 3.9484,
      "step": 1275904
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9101186939148817e-05,
      "loss": 3.9424,
      "step": 1276416
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9092800991638297e-05,
      "loss": 3.9312,
      "step": 1276928
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9084431422931506e-05,
      "loss": 3.9313,
      "step": 1277440
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9076078233028448e-05,
      "loss": 3.9407,
      "step": 1277952
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9067692285517928e-05,
      "loss": 3.9501,
      "step": 1278464
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9059306338007408e-05,
      "loss": 3.9445,
      "step": 1278976
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9050920390496888e-05,
      "loss": 3.9419,
      "step": 1279488
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9042534442986368e-05,
      "loss": 3.9477,
      "step": 1280000
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9034148495475848e-05,
      "loss": 3.953,
      "step": 1280512
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.902576254796533e-05,
      "loss": 3.9391,
      "step": 1281024
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.901737660045481e-05,
      "loss": 3.9368,
      "step": 1281536
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.900899065294429e-05,
      "loss": 3.9457,
      "step": 1282048
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.90006210842375e-05,
      "loss": 3.9438,
      "step": 1282560
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.899223513672698e-05,
      "loss": 3.9299,
      "step": 1283072
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.898384918921646e-05,
      "loss": 3.9344,
      "step": 1283584
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.897546324170594e-05,
      "loss": 3.9382,
      "step": 1284096
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8967077294195423e-05,
      "loss": 3.9357,
      "step": 1284608
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8958691346684896e-05,
      "loss": 3.9413,
      "step": 1285120
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8950305399174376e-05,
      "loss": 3.9494,
      "step": 1285632
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8941919451663856e-05,
      "loss": 3.9344,
      "step": 1286144
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.89335662617608e-05,
      "loss": 3.9391,
      "step": 1286656
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8925180314250285e-05,
      "loss": 3.9336,
      "step": 1287168
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8916794366739765e-05,
      "loss": 3.9527,
      "step": 1287680
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8908408419229245e-05,
      "loss": 3.9462,
      "step": 1288192
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8900022471718718e-05,
      "loss": 3.9383,
      "step": 1288704
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8891636524208198e-05,
      "loss": 3.9446,
      "step": 1289216
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8883250576697678e-05,
      "loss": 3.9477,
      "step": 1289728
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.887486462918716e-05,
      "loss": 3.9488,
      "step": 1290240
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.886649506048037e-05,
      "loss": 3.9423,
      "step": 1290752
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.885810911296985e-05,
      "loss": 3.9417,
      "step": 1291264
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.884972316545933e-05,
      "loss": 3.9499,
      "step": 1291776
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.884133721794881e-05,
      "loss": 3.9447,
      "step": 1292288
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8832967649242022e-05,
      "loss": 3.9472,
      "step": 1292800
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8824581701731502e-05,
      "loss": 3.9351,
      "step": 1293312
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8816195754220982e-05,
      "loss": 3.9429,
      "step": 1293824
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8807809806710462e-05,
      "loss": 3.9339,
      "step": 1294336
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.879944023800367e-05,
      "loss": 3.926,
      "step": 1294848
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.879105429049315e-05,
      "loss": 3.9526,
      "step": 1295360
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8782684721786364e-05,
      "loss": 3.9392,
      "step": 1295872
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8774298774275844e-05,
      "loss": 3.9447,
      "step": 1296384
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8765912826765324e-05,
      "loss": 3.9436,
      "step": 1296896
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8757526879254804e-05,
      "loss": 3.9272,
      "step": 1297408
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.013129234313965,
      "eval_runtime": 287.7456,
      "eval_samples_per_second": 1326.14,
      "eval_steps_per_second": 41.443,
      "step": 1297440
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8749140931744284e-05,
      "loss": 3.9382,
      "step": 1297920
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8740754984233764e-05,
      "loss": 3.9365,
      "step": 1298432
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8732369036723244e-05,
      "loss": 3.953,
      "step": 1298944
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8723983089212724e-05,
      "loss": 3.9394,
      "step": 1299456
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8715597141702204e-05,
      "loss": 3.9482,
      "step": 1299968
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8707211194191687e-05,
      "loss": 3.9328,
      "step": 1300480
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8698825246681167e-05,
      "loss": 3.9367,
      "step": 1300992
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8690439299170647e-05,
      "loss": 3.9338,
      "step": 1301504
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8682069730463856e-05,
      "loss": 3.9422,
      "step": 1302016
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8673683782953336e-05,
      "loss": 3.9413,
      "step": 1302528
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.866531421424655e-05,
      "loss": 3.9516,
      "step": 1303040
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8656928266736028e-05,
      "loss": 3.9506,
      "step": 1303552
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8648558698029237e-05,
      "loss": 3.936,
      "step": 1304064
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8640172750518717e-05,
      "loss": 3.9372,
      "step": 1304576
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8631786803008197e-05,
      "loss": 3.9332,
      "step": 1305088
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8623400855497677e-05,
      "loss": 3.9313,
      "step": 1305600
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8615014907987157e-05,
      "loss": 3.94,
      "step": 1306112
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.860662896047664e-05,
      "loss": 3.9305,
      "step": 1306624
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.859824301296612e-05,
      "loss": 3.9335,
      "step": 1307136
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.85898570654556e-05,
      "loss": 3.9599,
      "step": 1307648
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.858147111794508e-05,
      "loss": 3.9362,
      "step": 1308160
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.857310154923829e-05,
      "loss": 3.9423,
      "step": 1308672
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.856471560172777e-05,
      "loss": 3.9432,
      "step": 1309184
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.855632965421725e-05,
      "loss": 3.9454,
      "step": 1309696
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8547943706706726e-05,
      "loss": 3.937,
      "step": 1310208
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8539557759196206e-05,
      "loss": 3.9363,
      "step": 1310720
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8531188190489422e-05,
      "loss": 3.9321,
      "step": 1311232
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8522802242978902e-05,
      "loss": 3.9422,
      "step": 1311744
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.851441629546838e-05,
      "loss": 3.9245,
      "step": 1312256
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8506030347957858e-05,
      "loss": 3.9389,
      "step": 1312768
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8497660779251074e-05,
      "loss": 3.9399,
      "step": 1313280
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8489274831740554e-05,
      "loss": 3.9376,
      "step": 1313792
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8480888884230027e-05,
      "loss": 3.9438,
      "step": 1314304
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8472502936719507e-05,
      "loss": 3.9355,
      "step": 1314816
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8464133368012723e-05,
      "loss": 3.9366,
      "step": 1315328
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.84557474205022e-05,
      "loss": 3.9362,
      "step": 1315840
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.844736147299168e-05,
      "loss": 3.9205,
      "step": 1316352
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.843897552548116e-05,
      "loss": 3.9346,
      "step": 1316864
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.843058957797064e-05,
      "loss": 3.9246,
      "step": 1317376
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.842220363046012e-05,
      "loss": 3.9308,
      "step": 1317888
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.84138176829496e-05,
      "loss": 3.9364,
      "step": 1318400
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.840543173543908e-05,
      "loss": 3.9439,
      "step": 1318912
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8397045787928563e-05,
      "loss": 3.9353,
      "step": 1319424
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8388659840418043e-05,
      "loss": 3.9335,
      "step": 1319936
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8380273892907523e-05,
      "loss": 3.945,
      "step": 1320448
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8371887945397003e-05,
      "loss": 3.9373,
      "step": 1320960
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8363518376690212e-05,
      "loss": 3.9315,
      "step": 1321472
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.835513242917969e-05,
      "loss": 3.9293,
      "step": 1321984
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8346779239276633e-05,
      "loss": 3.9257,
      "step": 1322496
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8338393291766113e-05,
      "loss": 3.9407,
      "step": 1323008
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8330007344255593e-05,
      "loss": 3.9369,
      "step": 1323520
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8321621396745073e-05,
      "loss": 3.9322,
      "step": 1324032
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8313235449234553e-05,
      "loss": 3.9296,
      "step": 1324544
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8304849501724033e-05,
      "loss": 3.9345,
      "step": 1325056
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8296463554213516e-05,
      "loss": 3.9192,
      "step": 1325568
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8288077606702996e-05,
      "loss": 3.9429,
      "step": 1326080
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8279691659192476e-05,
      "loss": 3.9273,
      "step": 1326592
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8271305711681956e-05,
      "loss": 3.9439,
      "step": 1327104
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8262936142975165e-05,
      "loss": 3.9332,
      "step": 1327616
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8254550195464645e-05,
      "loss": 3.9242,
      "step": 1328128
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8246164247954125e-05,
      "loss": 3.9285,
      "step": 1328640
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8237778300443605e-05,
      "loss": 3.9374,
      "step": 1329152
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8229408731736818e-05,
      "loss": 3.9173,
      "step": 1329664
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8221022784226298e-05,
      "loss": 3.921,
      "step": 1330176
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8212636836715778e-05,
      "loss": 3.9388,
      "step": 1330688
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8204250889205258e-05,
      "loss": 3.9316,
      "step": 1331200
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8195864941694738e-05,
      "loss": 3.9229,
      "step": 1331712
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8187478994184214e-05,
      "loss": 3.9193,
      "step": 1332224
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8179093046673694e-05,
      "loss": 3.9248,
      "step": 1332736
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8170707099163174e-05,
      "loss": 3.9404,
      "step": 1333248
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8162337530456383e-05,
      "loss": 3.9372,
      "step": 1333760
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.81539679617496e-05,
      "loss": 3.9311,
      "step": 1334272
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.814558201423908e-05,
      "loss": 3.9354,
      "step": 1334784
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.813719606672856e-05,
      "loss": 3.9344,
      "step": 1335296
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8128810119218036e-05,
      "loss": 3.9365,
      "step": 1335808
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.812044055051125e-05,
      "loss": 3.9249,
      "step": 1336320
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.811207098180446e-05,
      "loss": 3.9368,
      "step": 1336832
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.810368503429394e-05,
      "loss": 3.932,
      "step": 1337344
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.809529908678342e-05,
      "loss": 3.9319,
      "step": 1337856
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8086913139272904e-05,
      "loss": 3.9361,
      "step": 1338368
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8078527191762384e-05,
      "loss": 3.9336,
      "step": 1338880
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8070141244251857e-05,
      "loss": 3.944,
      "step": 1339392
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8061755296741337e-05,
      "loss": 3.9386,
      "step": 1339904
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8053369349230817e-05,
      "loss": 3.9177,
      "step": 1340416
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8044983401720297e-05,
      "loss": 3.9316,
      "step": 1340928
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.803661383301351e-05,
      "loss": 3.9301,
      "step": 1341440
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.802822788550299e-05,
      "loss": 3.9385,
      "step": 1341952
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.801984193799247e-05,
      "loss": 3.9323,
      "step": 1342464
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.801145599048195e-05,
      "loss": 3.9347,
      "step": 1342976
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.8003086421775158e-05,
      "loss": 3.9352,
      "step": 1343488
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.799470047426464e-05,
      "loss": 3.9333,
      "step": 1344000
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.798631452675412e-05,
      "loss": 3.9346,
      "step": 1344512
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.79779285792436e-05,
      "loss": 3.9344,
      "step": 1345024
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.796954263173308e-05,
      "loss": 3.9252,
      "step": 1345536
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.796115668422256e-05,
      "loss": 3.941,
      "step": 1346048
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.795278711551577e-05,
      "loss": 3.9392,
      "step": 1346560
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.794440116800525e-05,
      "loss": 3.9299,
      "step": 1347072
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7936015220494734e-05,
      "loss": 3.9419,
      "step": 1347584
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7927629272984214e-05,
      "loss": 3.9377,
      "step": 1348096
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7919243325473694e-05,
      "loss": 3.9458,
      "step": 1348608
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7910873756766903e-05,
      "loss": 3.9323,
      "step": 1349120
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7902487809256383e-05,
      "loss": 3.9266,
      "step": 1349632
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7894101861745863e-05,
      "loss": 3.9323,
      "step": 1350144
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7885715914235343e-05,
      "loss": 3.9332,
      "step": 1350656
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7877329966724826e-05,
      "loss": 3.9154,
      "step": 1351168
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7868944019214306e-05,
      "loss": 3.9278,
      "step": 1351680
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7860558071703786e-05,
      "loss": 3.9357,
      "step": 1352192
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7852172124193266e-05,
      "loss": 3.9326,
      "step": 1352704
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7843802555486475e-05,
      "loss": 3.9251,
      "step": 1353216
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7835416607975955e-05,
      "loss": 3.918,
      "step": 1353728
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7827030660465435e-05,
      "loss": 3.9303,
      "step": 1354240
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7818661091758647e-05,
      "loss": 3.9404,
      "step": 1354752
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7810275144248127e-05,
      "loss": 3.9342,
      "step": 1355264
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7801889196737607e-05,
      "loss": 3.9345,
      "step": 1355776
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7793503249227087e-05,
      "loss": 3.9341,
      "step": 1356288
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7785117301716567e-05,
      "loss": 3.95,
      "step": 1356800
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.777674773300978e-05,
      "loss": 3.927,
      "step": 1357312
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.776836178549926e-05,
      "loss": 3.9273,
      "step": 1357824
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.775997583798874e-05,
      "loss": 3.9334,
      "step": 1358336
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.775160626928195e-05,
      "loss": 3.9379,
      "step": 1358848
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.774322032177143e-05,
      "loss": 3.9194,
      "step": 1359360
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.773483437426091e-05,
      "loss": 3.9224,
      "step": 1359872
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.772644842675039e-05,
      "loss": 3.9304,
      "step": 1360384
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7718062479239872e-05,
      "loss": 3.9285,
      "step": 1360896
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7709676531729345e-05,
      "loss": 3.9333,
      "step": 1361408
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7701290584218825e-05,
      "loss": 3.9378,
      "step": 1361920
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.769292101551204e-05,
      "loss": 3.9217,
      "step": 1362432
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.768453506800152e-05,
      "loss": 3.9322,
      "step": 1362944
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7676149120490997e-05,
      "loss": 3.9205,
      "step": 1363456
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7667763172980477e-05,
      "loss": 3.9445,
      "step": 1363968
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7659393604273693e-05,
      "loss": 3.9402,
      "step": 1364480
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7651007656763166e-05,
      "loss": 3.928,
      "step": 1364992
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7642621709252646e-05,
      "loss": 3.9344,
      "step": 1365504
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7634235761742126e-05,
      "loss": 3.9334,
      "step": 1366016
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.762584981423161e-05,
      "loss": 3.9405,
      "step": 1366528
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.761746386672109e-05,
      "loss": 3.9367,
      "step": 1367040
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.760907791921057e-05,
      "loss": 3.9335,
      "step": 1367552
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.760069197170005e-05,
      "loss": 3.9359,
      "step": 1368064
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.759230602418953e-05,
      "loss": 3.9328,
      "step": 1368576
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.758392007667901e-05,
      "loss": 3.9339,
      "step": 1369088
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.757553412916849e-05,
      "loss": 3.9264,
      "step": 1369600
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.756714818165797e-05,
      "loss": 3.9342,
      "step": 1370112
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7558778612951182e-05,
      "loss": 3.9272,
      "step": 1370624
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7550392665440662e-05,
      "loss": 3.9174,
      "step": 1371136
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.754202309673387e-05,
      "loss": 3.9419,
      "step": 1371648
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.753363714922335e-05,
      "loss": 3.93,
      "step": 1372160
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.752525120171283e-05,
      "loss": 3.9349,
      "step": 1372672
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.751686525420231e-05,
      "loss": 3.9341,
      "step": 1373184
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7508495685495523e-05,
      "loss": 3.9219,
      "step": 1373696
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.008572578430176,
      "eval_runtime": 289.1556,
      "eval_samples_per_second": 1319.674,
      "eval_steps_per_second": 41.241,
      "step": 1373760
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7500109737985003e-05,
      "loss": 3.9303,
      "step": 1374208
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7491723790474483e-05,
      "loss": 3.9225,
      "step": 1374720
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7483337842963963e-05,
      "loss": 3.9432,
      "step": 1375232
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7474951895453443e-05,
      "loss": 3.9269,
      "step": 1375744
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7466565947942923e-05,
      "loss": 3.9418,
      "step": 1376256
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7458180000432403e-05,
      "loss": 3.9257,
      "step": 1376768
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.744979405292188e-05,
      "loss": 3.9249,
      "step": 1377280
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.744140810541136e-05,
      "loss": 3.9213,
      "step": 1377792
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.743302215790084e-05,
      "loss": 3.9354,
      "step": 1378304
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.742463621039032e-05,
      "loss": 3.9308,
      "step": 1378816
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.74162502628798e-05,
      "loss": 3.9393,
      "step": 1379328
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.740786431536928e-05,
      "loss": 3.9402,
      "step": 1379840
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7399494746662492e-05,
      "loss": 3.9267,
      "step": 1380352
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7391108799151972e-05,
      "loss": 3.9275,
      "step": 1380864
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.738272285164145e-05,
      "loss": 3.9241,
      "step": 1381376
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.737433690413093e-05,
      "loss": 3.9229,
      "step": 1381888
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.736595095662041e-05,
      "loss": 3.9289,
      "step": 1382400
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.735756500910989e-05,
      "loss": 3.9228,
      "step": 1382912
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.734917906159937e-05,
      "loss": 3.9234,
      "step": 1383424
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.734079311408885e-05,
      "loss": 3.9427,
      "step": 1383936
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7332407166578335e-05,
      "loss": 3.9286,
      "step": 1384448
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7324021219067815e-05,
      "loss": 3.9315,
      "step": 1384960
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7315635271557295e-05,
      "loss": 3.9335,
      "step": 1385472
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7307265702850504e-05,
      "loss": 3.9361,
      "step": 1385984
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7298896134143713e-05,
      "loss": 3.9266,
      "step": 1386496
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7290510186633196e-05,
      "loss": 3.9273,
      "step": 1387008
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7282124239122676e-05,
      "loss": 3.9253,
      "step": 1387520
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7273738291612156e-05,
      "loss": 3.9285,
      "step": 1388032
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7265352344101636e-05,
      "loss": 3.9177,
      "step": 1388544
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7256966396591116e-05,
      "loss": 3.9294,
      "step": 1389056
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7248580449080596e-05,
      "loss": 3.9278,
      "step": 1389568
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7240210880373805e-05,
      "loss": 3.9283,
      "step": 1390080
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.723182493286329e-05,
      "loss": 3.9348,
      "step": 1390592
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7223455364156498e-05,
      "loss": 3.9267,
      "step": 1391104
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7215085795449707e-05,
      "loss": 3.927,
      "step": 1391616
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7206699847939187e-05,
      "loss": 3.926,
      "step": 1392128
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7198313900428667e-05,
      "loss": 3.916,
      "step": 1392640
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.718992795291815e-05,
      "loss": 3.9191,
      "step": 1393152
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.718154200540763e-05,
      "loss": 3.9126,
      "step": 1393664
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.717315605789711e-05,
      "loss": 3.9191,
      "step": 1394176
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.716477011038659e-05,
      "loss": 3.9281,
      "step": 1394688
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.715638416287607e-05,
      "loss": 3.9322,
      "step": 1395200
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7147998215365543e-05,
      "loss": 3.9248,
      "step": 1395712
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7139612267855026e-05,
      "loss": 3.9234,
      "step": 1396224
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7131226320344506e-05,
      "loss": 3.9372,
      "step": 1396736
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7122840372833986e-05,
      "loss": 3.9235,
      "step": 1397248
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7114454425323466e-05,
      "loss": 3.925,
      "step": 1397760
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7106084856616675e-05,
      "loss": 3.9226,
      "step": 1398272
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7097698909106155e-05,
      "loss": 3.9112,
      "step": 1398784
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7089312961595635e-05,
      "loss": 3.9326,
      "step": 1399296
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7080927014085115e-05,
      "loss": 3.9251,
      "step": 1399808
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.70725410665746e-05,
      "loss": 3.923,
      "step": 1400320
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7064171497867808e-05,
      "loss": 3.9195,
      "step": 1400832
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7055785550357287e-05,
      "loss": 3.9254,
      "step": 1401344
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7047399602846767e-05,
      "loss": 3.91,
      "step": 1401856
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7039013655336247e-05,
      "loss": 3.9344,
      "step": 1402368
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7030627707825727e-05,
      "loss": 3.9157,
      "step": 1402880
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7022241760315207e-05,
      "loss": 3.9353,
      "step": 1403392
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.701385581280469e-05,
      "loss": 3.9214,
      "step": 1403904
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.700546986529417e-05,
      "loss": 3.921,
      "step": 1404416
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.699710029658738e-05,
      "loss": 3.9153,
      "step": 1404928
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.698873072788059e-05,
      "loss": 3.9285,
      "step": 1405440
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.698034478037007e-05,
      "loss": 3.9068,
      "step": 1405952
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6971958832859552e-05,
      "loss": 3.9153,
      "step": 1406464
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6963572885349032e-05,
      "loss": 3.9271,
      "step": 1406976
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6955186937838512e-05,
      "loss": 3.9226,
      "step": 1407488
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6946800990327992e-05,
      "loss": 3.9141,
      "step": 1408000
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6938415042817472e-05,
      "loss": 3.9088,
      "step": 1408512
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6930029095306952e-05,
      "loss": 3.912,
      "step": 1409024
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.692164314779643e-05,
      "loss": 3.9341,
      "step": 1409536
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6913289957893373e-05,
      "loss": 3.9257,
      "step": 1410048
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6904904010382853e-05,
      "loss": 3.92,
      "step": 1410560
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6896534441676063e-05,
      "loss": 3.9287,
      "step": 1411072
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6888148494165542e-05,
      "loss": 3.9262,
      "step": 1411584
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6879762546655022e-05,
      "loss": 3.9275,
      "step": 1412096
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6871376599144506e-05,
      "loss": 3.9118,
      "step": 1412608
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6863007030437715e-05,
      "loss": 3.9302,
      "step": 1413120
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6854621082927195e-05,
      "loss": 3.927,
      "step": 1413632
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6846235135416675e-05,
      "loss": 3.9175,
      "step": 1414144
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6837849187906155e-05,
      "loss": 3.9306,
      "step": 1414656
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6829463240395635e-05,
      "loss": 3.9222,
      "step": 1415168
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6821077292885115e-05,
      "loss": 3.9308,
      "step": 1415680
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6812691345374598e-05,
      "loss": 3.9304,
      "step": 1416192
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6804305397864078e-05,
      "loss": 3.9122,
      "step": 1416704
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.679591945035355e-05,
      "loss": 3.9214,
      "step": 1417216
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.678753350284303e-05,
      "loss": 3.9199,
      "step": 1417728
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.677914755533251e-05,
      "loss": 3.9273,
      "step": 1418240
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6770777986625727e-05,
      "loss": 3.9266,
      "step": 1418752
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6762392039115203e-05,
      "loss": 3.923,
      "step": 1419264
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6754006091604683e-05,
      "loss": 3.9263,
      "step": 1419776
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6745620144094163e-05,
      "loss": 3.9251,
      "step": 1420288
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6737234196583643e-05,
      "loss": 3.9227,
      "step": 1420800
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6728864627876852e-05,
      "loss": 3.9241,
      "step": 1421312
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6720478680366336e-05,
      "loss": 3.9236,
      "step": 1421824
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6712092732855816e-05,
      "loss": 3.9296,
      "step": 1422336
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6703706785345296e-05,
      "loss": 3.925,
      "step": 1422848
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6695337216638505e-05,
      "loss": 3.9205,
      "step": 1423360
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6686951269127985e-05,
      "loss": 3.9327,
      "step": 1423872
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6678565321617465e-05,
      "loss": 3.9272,
      "step": 1424384
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6670179374106945e-05,
      "loss": 3.9385,
      "step": 1424896
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6661809805400157e-05,
      "loss": 3.9206,
      "step": 1425408
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6653423857889637e-05,
      "loss": 3.9208,
      "step": 1425920
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6645037910379117e-05,
      "loss": 3.9235,
      "step": 1426432
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6636651962868597e-05,
      "loss": 3.9253,
      "step": 1426944
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6628282394161806e-05,
      "loss": 3.9038,
      "step": 1427456
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.661989644665129e-05,
      "loss": 3.9175,
      "step": 1427968
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.661151049914077e-05,
      "loss": 3.9261,
      "step": 1428480
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.660312455163025e-05,
      "loss": 3.9192,
      "step": 1428992
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.659473860411973e-05,
      "loss": 3.9204,
      "step": 1429504
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.658635265660921e-05,
      "loss": 3.9113,
      "step": 1430016
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.657796670909869e-05,
      "loss": 3.9184,
      "step": 1430528
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.65695971403919e-05,
      "loss": 3.9335,
      "step": 1431040
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.656121119288138e-05,
      "loss": 3.9202,
      "step": 1431552
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.655282524537086e-05,
      "loss": 3.9313,
      "step": 1432064
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.654443929786034e-05,
      "loss": 3.9239,
      "step": 1432576
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.653606972915355e-05,
      "loss": 3.9392,
      "step": 1433088
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.652768378164303e-05,
      "loss": 3.922,
      "step": 1433600
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6519314212936243e-05,
      "loss": 3.9206,
      "step": 1434112
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6510928265425723e-05,
      "loss": 3.9229,
      "step": 1434624
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6502542317915203e-05,
      "loss": 3.9271,
      "step": 1435136
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6494172749208412e-05,
      "loss": 3.912,
      "step": 1435648
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6485786801697892e-05,
      "loss": 3.9128,
      "step": 1436160
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6477400854187372e-05,
      "loss": 3.9185,
      "step": 1436672
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6469014906676852e-05,
      "loss": 3.9234,
      "step": 1437184
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6460628959166335e-05,
      "loss": 3.9238,
      "step": 1437696
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6452243011655815e-05,
      "loss": 3.9313,
      "step": 1438208
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6443857064145295e-05,
      "loss": 3.9116,
      "step": 1438720
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6435471116634775e-05,
      "loss": 3.9197,
      "step": 1439232
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6427085169124255e-05,
      "loss": 3.9128,
      "step": 1439744
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6418699221613735e-05,
      "loss": 3.9361,
      "step": 1440256
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6410313274103208e-05,
      "loss": 3.9341,
      "step": 1440768
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.640192732659269e-05,
      "loss": 3.9201,
      "step": 1441280
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6393557757885908e-05,
      "loss": 3.9219,
      "step": 1441792
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6385171810375387e-05,
      "loss": 3.9244,
      "step": 1442304
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.637678586286486e-05,
      "loss": 3.9284,
      "step": 1442816
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.636839991535434e-05,
      "loss": 3.9321,
      "step": 1443328
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.636001396784382e-05,
      "loss": 3.9206,
      "step": 1443840
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6351644399137033e-05,
      "loss": 3.9261,
      "step": 1444352
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6343258451626513e-05,
      "loss": 3.9227,
      "step": 1444864
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6334872504115993e-05,
      "loss": 3.9303,
      "step": 1445376
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6326486556605473e-05,
      "loss": 3.9193,
      "step": 1445888
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6318100609094953e-05,
      "loss": 3.9238,
      "step": 1446400
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6309714661584433e-05,
      "loss": 3.9209,
      "step": 1446912
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6301345092877645e-05,
      "loss": 3.9057,
      "step": 1447424
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6292959145367125e-05,
      "loss": 3.9332,
      "step": 1447936
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6284573197856605e-05,
      "loss": 3.9206,
      "step": 1448448
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6276203629149814e-05,
      "loss": 3.9241,
      "step": 1448960
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6267817681639294e-05,
      "loss": 3.9268,
      "step": 1449472
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.6259431734128774e-05,
      "loss": 3.9129,
      "step": 1449984
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.004698276519775,
      "eval_runtime": 295.3733,
      "eval_samples_per_second": 1291.894,
      "eval_steps_per_second": 40.373,
      "step": 1450080
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6251045786618254e-05,
      "loss": 3.9214,
      "step": 1450496
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6242659839107738e-05,
      "loss": 3.9128,
      "step": 1451008
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6234273891597217e-05,
      "loss": 3.9342,
      "step": 1451520
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6225887944086697e-05,
      "loss": 3.9171,
      "step": 1452032
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6217501996576177e-05,
      "loss": 3.9407,
      "step": 1452544
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6209132427869386e-05,
      "loss": 3.9125,
      "step": 1453056
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6200746480358866e-05,
      "loss": 3.92,
      "step": 1453568
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6192360532848346e-05,
      "loss": 3.9085,
      "step": 1454080
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.618397458533783e-05,
      "loss": 3.9279,
      "step": 1454592
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.617558863782731e-05,
      "loss": 3.9239,
      "step": 1455104
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.616720269031679e-05,
      "loss": 3.9258,
      "step": 1455616
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.615881674280627e-05,
      "loss": 3.9383,
      "step": 1456128
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.615044717409948e-05,
      "loss": 3.9148,
      "step": 1456640
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.614206122658896e-05,
      "loss": 3.9209,
      "step": 1457152
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.613367527907844e-05,
      "loss": 3.9139,
      "step": 1457664
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6125289331567922e-05,
      "loss": 3.9163,
      "step": 1458176
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6116903384057395e-05,
      "loss": 3.9189,
      "step": 1458688
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6108517436546875e-05,
      "loss": 3.9149,
      "step": 1459200
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6100131489036355e-05,
      "loss": 3.9159,
      "step": 1459712
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.609176192032957e-05,
      "loss": 3.931,
      "step": 1460224
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6083375972819047e-05,
      "loss": 3.9227,
      "step": 1460736
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6074990025308527e-05,
      "loss": 3.923,
      "step": 1461248
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6066620456601743e-05,
      "loss": 3.9276,
      "step": 1461760
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6058234509091223e-05,
      "loss": 3.9271,
      "step": 1462272
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6049848561580696e-05,
      "loss": 3.9173,
      "step": 1462784
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.6041462614070176e-05,
      "loss": 3.919,
      "step": 1463296
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.603307666655966e-05,
      "loss": 3.9179,
      "step": 1463808
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.602469071904914e-05,
      "loss": 3.9153,
      "step": 1464320
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.601630477153862e-05,
      "loss": 3.913,
      "step": 1464832
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.60079188240281e-05,
      "loss": 3.9148,
      "step": 1465344
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.599953287651758e-05,
      "loss": 3.9244,
      "step": 1465856
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.599116330781079e-05,
      "loss": 3.919,
      "step": 1466368
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.598277736030027e-05,
      "loss": 3.927,
      "step": 1466880
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5974391412789752e-05,
      "loss": 3.9197,
      "step": 1467392
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.596602184408296e-05,
      "loss": 3.919,
      "step": 1467904
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.595763589657244e-05,
      "loss": 3.916,
      "step": 1468416
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.594924994906192e-05,
      "loss": 3.9107,
      "step": 1468928
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.59408640015514e-05,
      "loss": 3.9086,
      "step": 1469440
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.593247805404088e-05,
      "loss": 3.9052,
      "step": 1469952
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.592409210653036e-05,
      "loss": 3.9109,
      "step": 1470464
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.591570615901984e-05,
      "loss": 3.9212,
      "step": 1470976
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5907320211509324e-05,
      "loss": 3.9214,
      "step": 1471488
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5898950642802533e-05,
      "loss": 3.9188,
      "step": 1472000
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5890564695292013e-05,
      "loss": 3.918,
      "step": 1472512
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5882178747781493e-05,
      "loss": 3.9312,
      "step": 1473024
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5873792800270973e-05,
      "loss": 3.9143,
      "step": 1473536
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5865423231564186e-05,
      "loss": 3.9161,
      "step": 1474048
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5857037284053666e-05,
      "loss": 3.9124,
      "step": 1474560
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5848651336543146e-05,
      "loss": 3.9048,
      "step": 1475072
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5840265389032625e-05,
      "loss": 3.928,
      "step": 1475584
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5831879441522105e-05,
      "loss": 3.9149,
      "step": 1476096
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.582349349401158e-05,
      "loss": 3.9155,
      "step": 1476608
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5815107546501062e-05,
      "loss": 3.9112,
      "step": 1477120
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5806721598990542e-05,
      "loss": 3.9226,
      "step": 1477632
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5798352030283758e-05,
      "loss": 3.8996,
      "step": 1478144
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.578996608277323e-05,
      "loss": 3.9228,
      "step": 1478656
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.578158013526271e-05,
      "loss": 3.91,
      "step": 1479168
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5773210566555927e-05,
      "loss": 3.9234,
      "step": 1479680
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5764824619045407e-05,
      "loss": 3.9184,
      "step": 1480192
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5756438671534883e-05,
      "loss": 3.909,
      "step": 1480704
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5748052724024363e-05,
      "loss": 3.9046,
      "step": 1481216
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5739666776513843e-05,
      "loss": 3.9255,
      "step": 1481728
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5731280829003323e-05,
      "loss": 3.8996,
      "step": 1482240
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5722894881492803e-05,
      "loss": 3.9017,
      "step": 1482752
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5714508933982283e-05,
      "loss": 3.9229,
      "step": 1483264
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5706122986471763e-05,
      "loss": 3.9163,
      "step": 1483776
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5697753417764976e-05,
      "loss": 3.9037,
      "step": 1484288
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5689367470254455e-05,
      "loss": 3.9049,
      "step": 1484800
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5680981522743935e-05,
      "loss": 3.9035,
      "step": 1485312
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5672611954037145e-05,
      "loss": 3.9244,
      "step": 1485824
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5664226006526624e-05,
      "loss": 3.9175,
      "step": 1486336
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5655840059016108e-05,
      "loss": 3.9126,
      "step": 1486848
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5647454111505588e-05,
      "loss": 3.9192,
      "step": 1487360
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5639068163995068e-05,
      "loss": 3.9207,
      "step": 1487872
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5630682216484548e-05,
      "loss": 3.9222,
      "step": 1488384
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5622296268974028e-05,
      "loss": 3.9033,
      "step": 1488896
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5613910321463508e-05,
      "loss": 3.9194,
      "step": 1489408
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5605524373952988e-05,
      "loss": 3.9183,
      "step": 1489920
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.55971548052462e-05,
      "loss": 3.9079,
      "step": 1490432
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.558876885773568e-05,
      "loss": 3.9253,
      "step": 1490944
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.558038291022516e-05,
      "loss": 3.9129,
      "step": 1491456
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.557199696271464e-05,
      "loss": 3.9229,
      "step": 1491968
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.556361101520412e-05,
      "loss": 3.9231,
      "step": 1492480
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.555524144649733e-05,
      "loss": 3.907,
      "step": 1492992
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.554685549898681e-05,
      "loss": 3.9145,
      "step": 1493504
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5538469551476292e-05,
      "loss": 3.9104,
      "step": 1494016
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5530083603965772e-05,
      "loss": 3.9151,
      "step": 1494528
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.5521697656455245e-05,
      "loss": 3.9186,
      "step": 1495040
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.551332808774846e-05,
      "loss": 3.9117,
      "step": 1495552
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.550494214023794e-05,
      "loss": 3.9207,
      "step": 1496064
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5496556192727418e-05,
      "loss": 3.9213,
      "step": 1496576
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5488186624020634e-05,
      "loss": 3.9123,
      "step": 1497088
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5479800676510114e-05,
      "loss": 3.9158,
      "step": 1497600
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5471414728999594e-05,
      "loss": 3.9149,
      "step": 1498112
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5463028781489067e-05,
      "loss": 3.9256,
      "step": 1498624
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5454642833978547e-05,
      "loss": 3.9171,
      "step": 1499136
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.544625688646803e-05,
      "loss": 3.9131,
      "step": 1499648
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.543787093895751e-05,
      "loss": 3.9209,
      "step": 1500160
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.542948499144699e-05,
      "loss": 3.9148,
      "step": 1500672
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.54211154227402e-05,
      "loss": 3.9313,
      "step": 1501184
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.541272947522968e-05,
      "loss": 3.9147,
      "step": 1501696
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.540434352771916e-05,
      "loss": 3.9155,
      "step": 1502208
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.539595758020864e-05,
      "loss": 3.9127,
      "step": 1502720
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.538757163269812e-05,
      "loss": 3.9189,
      "step": 1503232
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.537920206399133e-05,
      "loss": 3.8979,
      "step": 1503744
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.537081611648081e-05,
      "loss": 3.9077,
      "step": 1504256
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.536243016897029e-05,
      "loss": 3.9194,
      "step": 1504768
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.535404422145977e-05,
      "loss": 3.9134,
      "step": 1505280
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5345674652752984e-05,
      "loss": 3.9075,
      "step": 1505792
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5337288705242464e-05,
      "loss": 3.906,
      "step": 1506304
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5328902757731944e-05,
      "loss": 3.9087,
      "step": 1506816
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5320516810221424e-05,
      "loss": 3.928,
      "step": 1507328
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5312147241514633e-05,
      "loss": 3.9086,
      "step": 1507840
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5303761294004113e-05,
      "loss": 3.9225,
      "step": 1508352
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5295375346493593e-05,
      "loss": 3.9158,
      "step": 1508864
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5286989398983073e-05,
      "loss": 3.9326,
      "step": 1509376
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5278603451472556e-05,
      "loss": 3.917,
      "step": 1509888
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5270217503962036e-05,
      "loss": 3.9049,
      "step": 1510400
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5261831556451516e-05,
      "loss": 3.9166,
      "step": 1510912
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5253445608940996e-05,
      "loss": 3.9201,
      "step": 1511424
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5245059661430476e-05,
      "loss": 3.9023,
      "step": 1511936
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5236690092723685e-05,
      "loss": 3.9043,
      "step": 1512448
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5228304145213165e-05,
      "loss": 3.9129,
      "step": 1512960
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5219918197702648e-05,
      "loss": 3.9162,
      "step": 1513472
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5211532250192128e-05,
      "loss": 3.9125,
      "step": 1513984
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.52031463026816e-05,
      "loss": 3.9278,
      "step": 1514496
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5194776733974817e-05,
      "loss": 3.9006,
      "step": 1515008
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5186407165268026e-05,
      "loss": 3.9125,
      "step": 1515520
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.517802121775751e-05,
      "loss": 3.9048,
      "step": 1516032
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.516963527024699e-05,
      "loss": 3.9273,
      "step": 1516544
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.516124932273647e-05,
      "loss": 3.9261,
      "step": 1517056
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.515286337522595e-05,
      "loss": 3.9146,
      "step": 1517568
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.514447742771543e-05,
      "loss": 3.9108,
      "step": 1518080
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5136091480204903e-05,
      "loss": 3.9189,
      "step": 1518592
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5127705532694386e-05,
      "loss": 3.9181,
      "step": 1519104
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5119319585183866e-05,
      "loss": 3.9185,
      "step": 1519616
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5110933637673346e-05,
      "loss": 3.9157,
      "step": 1520128
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5102547690162826e-05,
      "loss": 3.9157,
      "step": 1520640
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5094161742652306e-05,
      "loss": 3.9172,
      "step": 1521152
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5085792173945515e-05,
      "loss": 3.9239,
      "step": 1521664
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5077406226434995e-05,
      "loss": 3.9091,
      "step": 1522176
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5069020278924478e-05,
      "loss": 3.9119,
      "step": 1522688
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5060634331413958e-05,
      "loss": 3.9116,
      "step": 1523200
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5052281141510903e-05,
      "loss": 3.902,
      "step": 1523712
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5043895194000376e-05,
      "loss": 3.9216,
      "step": 1524224
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.5035509246489856e-05,
      "loss": 3.9141,
      "step": 1524736
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.502712329897934e-05,
      "loss": 3.9173,
      "step": 1525248
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.501873735146882e-05,
      "loss": 3.9147,
      "step": 1525760
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.50103514039583e-05,
      "loss": 3.9121,
      "step": 1526272
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.001376152038574,
      "eval_runtime": 292.7259,
      "eval_samples_per_second": 1303.578,
      "eval_steps_per_second": 40.738,
      "step": 1526400
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.500196545644778e-05,
      "loss": 3.9164,
      "step": 1526784
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.499357950893726e-05,
      "loss": 3.9046,
      "step": 1527296
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.498519356142674e-05,
      "loss": 3.9244,
      "step": 1527808
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.497680761391622e-05,
      "loss": 3.9103,
      "step": 1528320
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.49684216664057e-05,
      "loss": 3.9294,
      "step": 1528832
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.496003571889518e-05,
      "loss": 3.9126,
      "step": 1529344
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.495164977138466e-05,
      "loss": 3.9095,
      "step": 1529856
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.494326382387414e-05,
      "loss": 3.9017,
      "step": 1530368
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.493487787636362e-05,
      "loss": 3.9136,
      "step": 1530880
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.49264919288531e-05,
      "loss": 3.9194,
      "step": 1531392
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.491810598134258e-05,
      "loss": 3.9191,
      "step": 1531904
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4909720033832056e-05,
      "loss": 3.9252,
      "step": 1532416
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.490135046512527e-05,
      "loss": 3.9105,
      "step": 1532928
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.489296451761475e-05,
      "loss": 3.9114,
      "step": 1533440
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.488457857010423e-05,
      "loss": 3.9072,
      "step": 1533952
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4876192622593708e-05,
      "loss": 3.9052,
      "step": 1534464
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4867806675083188e-05,
      "loss": 3.9139,
      "step": 1534976
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4859420727572668e-05,
      "loss": 3.91,
      "step": 1535488
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4851034780062148e-05,
      "loss": 3.9084,
      "step": 1536000
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4842648832551628e-05,
      "loss": 3.923,
      "step": 1536512
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.483426288504111e-05,
      "loss": 3.9173,
      "step": 1537024
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.482587693753059e-05,
      "loss": 3.916,
      "step": 1537536
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4817490990020068e-05,
      "loss": 3.9154,
      "step": 1538048
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4809105042509548e-05,
      "loss": 3.9206,
      "step": 1538560
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4800719094999028e-05,
      "loss": 3.9053,
      "step": 1539072
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.479234952629224e-05,
      "loss": 3.9183,
      "step": 1539584
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4783979957585452e-05,
      "loss": 3.9052,
      "step": 1540096
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4775594010074932e-05,
      "loss": 3.9093,
      "step": 1540608
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4767208062564412e-05,
      "loss": 3.9065,
      "step": 1541120
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.475882211505389e-05,
      "loss": 3.9117,
      "step": 1541632
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.47504525463471e-05,
      "loss": 3.9145,
      "step": 1542144
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.474206659883658e-05,
      "loss": 3.9155,
      "step": 1542656
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4733680651326065e-05,
      "loss": 3.9176,
      "step": 1543168
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.472529470381554e-05,
      "loss": 3.9127,
      "step": 1543680
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4716925135108754e-05,
      "loss": 3.9082,
      "step": 1544192
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4708539187598234e-05,
      "loss": 3.9111,
      "step": 1544704
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.470015324008771e-05,
      "loss": 3.9054,
      "step": 1545216
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4691767292577194e-05,
      "loss": 3.8994,
      "step": 1545728
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4683381345066674e-05,
      "loss": 3.9012,
      "step": 1546240
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4674995397556154e-05,
      "loss": 3.9002,
      "step": 1546752
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4666609450045634e-05,
      "loss": 3.9126,
      "step": 1547264
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4658223502535113e-05,
      "loss": 3.9162,
      "step": 1547776
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4649837555024593e-05,
      "loss": 3.9106,
      "step": 1548288
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.464145160751407e-05,
      "loss": 3.9066,
      "step": 1548800
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4633082038807282e-05,
      "loss": 3.9224,
      "step": 1549312
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4624696091296766e-05,
      "loss": 3.9098,
      "step": 1549824
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4616326522589975e-05,
      "loss": 3.9094,
      "step": 1550336
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4607940575079455e-05,
      "loss": 3.9041,
      "step": 1550848
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4599554627568935e-05,
      "loss": 3.8939,
      "step": 1551360
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4591168680058415e-05,
      "loss": 3.9212,
      "step": 1551872
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4582782732547895e-05,
      "loss": 3.9085,
      "step": 1552384
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4574396785037375e-05,
      "loss": 3.9104,
      "step": 1552896
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4566010837526855e-05,
      "loss": 3.9073,
      "step": 1553408
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4557624890016335e-05,
      "loss": 3.9069,
      "step": 1553920
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4549238942505815e-05,
      "loss": 3.8969,
      "step": 1554432
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4540852994995295e-05,
      "loss": 3.9089,
      "step": 1554944
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4532467047484774e-05,
      "loss": 3.9059,
      "step": 1555456
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.452408109997425e-05,
      "loss": 3.9181,
      "step": 1555968
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4515711531267467e-05,
      "loss": 3.9086,
      "step": 1556480
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4507325583756947e-05,
      "loss": 3.9036,
      "step": 1556992
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4498939636246423e-05,
      "loss": 3.8974,
      "step": 1557504
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4490553688735903e-05,
      "loss": 3.9167,
      "step": 1558016
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4482184120029116e-05,
      "loss": 3.8945,
      "step": 1558528
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.447381455132233e-05,
      "loss": 3.8928,
      "step": 1559040
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.446542860381181e-05,
      "loss": 3.9104,
      "step": 1559552
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4457042656301288e-05,
      "loss": 3.9132,
      "step": 1560064
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4448656708790768e-05,
      "loss": 3.8969,
      "step": 1560576
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4440287140083977e-05,
      "loss": 3.8977,
      "step": 1561088
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.443191757137719e-05,
      "loss": 3.8929,
      "step": 1561600
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.442353162386667e-05,
      "loss": 3.9119,
      "step": 1562112
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.441514567635615e-05,
      "loss": 3.9108,
      "step": 1562624
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.440675972884563e-05,
      "loss": 3.9062,
      "step": 1563136
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.439837378133511e-05,
      "loss": 3.9102,
      "step": 1563648
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.438998783382459e-05,
      "loss": 3.9101,
      "step": 1564160
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.438160188631407e-05,
      "loss": 3.9187,
      "step": 1564672
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.437321593880355e-05,
      "loss": 3.8992,
      "step": 1565184
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.436482999129303e-05,
      "loss": 3.9076,
      "step": 1565696
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.435644404378251e-05,
      "loss": 3.9091,
      "step": 1566208
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.434805809627199e-05,
      "loss": 3.8967,
      "step": 1566720
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.433967214876147e-05,
      "loss": 3.9202,
      "step": 1567232
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.433130258005468e-05,
      "loss": 3.9062,
      "step": 1567744
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.432291663254416e-05,
      "loss": 3.9135,
      "step": 1568256
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4314530685033642e-05,
      "loss": 3.9197,
      "step": 1568768
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.430614473752312e-05,
      "loss": 3.8992,
      "step": 1569280
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.429777516881633e-05,
      "loss": 3.9054,
      "step": 1569792
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.428938922130581e-05,
      "loss": 3.9028,
      "step": 1570304
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.428100327379529e-05,
      "loss": 3.9069,
      "step": 1570816
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.427261732628477e-05,
      "loss": 3.9101,
      "step": 1571328
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4264247757577983e-05,
      "loss": 3.9063,
      "step": 1571840
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4255861810067463e-05,
      "loss": 3.9112,
      "step": 1572352
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4247475862556943e-05,
      "loss": 3.9141,
      "step": 1572864
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4239089915046423e-05,
      "loss": 3.9033,
      "step": 1573376
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4230703967535903e-05,
      "loss": 3.9082,
      "step": 1573888
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4222334398829112e-05,
      "loss": 3.9089,
      "step": 1574400
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4213948451318595e-05,
      "loss": 3.9158,
      "step": 1574912
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4205562503808075e-05,
      "loss": 3.911,
      "step": 1575424
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4197176556297552e-05,
      "loss": 3.9048,
      "step": 1575936
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4188790608787032e-05,
      "loss": 3.9144,
      "step": 1576448
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4180404661276512e-05,
      "loss": 3.908,
      "step": 1576960
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4172018713765992e-05,
      "loss": 3.9253,
      "step": 1577472
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4163632766255472e-05,
      "loss": 3.9081,
      "step": 1577984
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4155246818744952e-05,
      "loss": 3.9059,
      "step": 1578496
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4146877250038164e-05,
      "loss": 3.9039,
      "step": 1579008
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4138507681331377e-05,
      "loss": 3.9125,
      "step": 1579520
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4130138112624586e-05,
      "loss": 3.8899,
      "step": 1580032
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4121752165114066e-05,
      "loss": 3.8999,
      "step": 1580544
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.411336621760355e-05,
      "loss": 3.9094,
      "step": 1581056
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4104980270093026e-05,
      "loss": 3.9105,
      "step": 1581568
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4096594322582506e-05,
      "loss": 3.9011,
      "step": 1582080
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4088208375071986e-05,
      "loss": 3.8968,
      "step": 1582592
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4079822427561466e-05,
      "loss": 3.9026,
      "step": 1583104
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4071436480050945e-05,
      "loss": 3.9177,
      "step": 1583616
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4063066911344158e-05,
      "loss": 3.9046,
      "step": 1584128
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4054680963833638e-05,
      "loss": 3.9128,
      "step": 1584640
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4046295016323118e-05,
      "loss": 3.9099,
      "step": 1585152
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4037909068812598e-05,
      "loss": 3.9192,
      "step": 1585664
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4029523121302078e-05,
      "loss": 3.9111,
      "step": 1586176
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4021153552595287e-05,
      "loss": 3.8988,
      "step": 1586688
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4012767605084767e-05,
      "loss": 3.9068,
      "step": 1587200
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.400438165757425e-05,
      "loss": 3.9137,
      "step": 1587712
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.399601208886746e-05,
      "loss": 3.8971,
      "step": 1588224
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.398762614135694e-05,
      "loss": 3.898,
      "step": 1588736
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.397924019384642e-05,
      "loss": 3.9024,
      "step": 1589248
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.39708542463359e-05,
      "loss": 3.9059,
      "step": 1589760
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.396246829882538e-05,
      "loss": 3.9068,
      "step": 1590272
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.395408235131486e-05,
      "loss": 3.9169,
      "step": 1590784
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.394569640380434e-05,
      "loss": 3.8918,
      "step": 1591296
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.393731045629382e-05,
      "loss": 3.9064,
      "step": 1591808
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3928940887587028e-05,
      "loss": 3.8948,
      "step": 1592320
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3920554940076508e-05,
      "loss": 3.9223,
      "step": 1592832
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3912168992565988e-05,
      "loss": 3.9107,
      "step": 1593344
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3903783045055468e-05,
      "loss": 3.9144,
      "step": 1593856
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.389539709754495e-05,
      "loss": 3.8993,
      "step": 1594368
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.388701115003443e-05,
      "loss": 3.9125,
      "step": 1594880
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3878657960131373e-05,
      "loss": 3.9157,
      "step": 1595392
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3870272012620853e-05,
      "loss": 3.9081,
      "step": 1595904
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.386188606511033e-05,
      "loss": 3.9066,
      "step": 1596416
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3853500117599813e-05,
      "loss": 3.9141,
      "step": 1596928
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3845114170089293e-05,
      "loss": 3.9066,
      "step": 1597440
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3836728222578773e-05,
      "loss": 3.9182,
      "step": 1597952
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3828342275068253e-05,
      "loss": 3.898,
      "step": 1598464
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3819956327557733e-05,
      "loss": 3.9066,
      "step": 1598976
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.381157038004721e-05,
      "loss": 3.9034,
      "step": 1599488
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.380318443253669e-05,
      "loss": 3.8973,
      "step": 1600000
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3794814863829905e-05,
      "loss": 3.9105,
      "step": 1600512
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.3786428916319385e-05,
      "loss": 3.908,
      "step": 1601024
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.377804296880886e-05,
      "loss": 3.9093,
      "step": 1601536
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.376965702129834e-05,
      "loss": 3.9064,
      "step": 1602048
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.376127107378782e-05,
      "loss": 3.9048,
      "step": 1602560
    },
    {
      "epoch": 0.03,
      "eval_loss": 3.997973680496216,
      "eval_runtime": 291.5436,
      "eval_samples_per_second": 1308.864,
      "eval_steps_per_second": 40.903,
      "step": 1602720
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3752901505081034e-05,
      "loss": 3.9033,
      "step": 1603072
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3744515557570514e-05,
      "loss": 3.8927,
      "step": 1603584
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3736129610059994e-05,
      "loss": 3.9154,
      "step": 1604096
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3727743662549474e-05,
      "loss": 3.9055,
      "step": 1604608
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3719357715038954e-05,
      "loss": 3.9211,
      "step": 1605120
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3710971767528434e-05,
      "loss": 3.9056,
      "step": 1605632
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3702585820017914e-05,
      "loss": 3.8982,
      "step": 1606144
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.369419987250739e-05,
      "loss": 3.8943,
      "step": 1606656
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3685830303800606e-05,
      "loss": 3.9063,
      "step": 1607168
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3677460735093815e-05,
      "loss": 3.9123,
      "step": 1607680
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3669074787583295e-05,
      "loss": 3.9101,
      "step": 1608192
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3660688840072775e-05,
      "loss": 3.9182,
      "step": 1608704
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3652302892562255e-05,
      "loss": 3.9033,
      "step": 1609216
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3643916945051735e-05,
      "loss": 3.9067,
      "step": 1609728
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3635530997541215e-05,
      "loss": 3.9005,
      "step": 1610240
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3627145050030695e-05,
      "loss": 3.8944,
      "step": 1610752
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3618759102520175e-05,
      "loss": 3.9052,
      "step": 1611264
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3610373155009655e-05,
      "loss": 3.9019,
      "step": 1611776
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3601987207499135e-05,
      "loss": 3.9022,
      "step": 1612288
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3593601259988615e-05,
      "loss": 3.9124,
      "step": 1612800
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3585215312478095e-05,
      "loss": 3.9116,
      "step": 1613312
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3576845743771307e-05,
      "loss": 3.9086,
      "step": 1613824
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3568459796260787e-05,
      "loss": 3.9054,
      "step": 1614336
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3560073848750267e-05,
      "loss": 3.9105,
      "step": 1614848
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3551704280043476e-05,
      "loss": 3.9003,
      "step": 1615360
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3543318332532956e-05,
      "loss": 3.9103,
      "step": 1615872
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3534932385022436e-05,
      "loss": 3.8991,
      "step": 1616384
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3526546437511916e-05,
      "loss": 3.9007,
      "step": 1616896
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.35181604900014e-05,
      "loss": 3.9007,
      "step": 1617408
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.3509774542490876e-05,
      "loss": 3.8981,
      "step": 1617920
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.350140497378409e-05,
      "loss": 3.9058,
      "step": 1618432
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.349301902627357e-05,
      "loss": 3.9085,
      "step": 1618944
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3484633078763045e-05,
      "loss": 3.9076,
      "step": 1619456
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3476247131252528e-05,
      "loss": 3.904,
      "step": 1619968
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3467861183742008e-05,
      "loss": 3.8969,
      "step": 1620480
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.345949161503522e-05,
      "loss": 3.9115,
      "step": 1620992
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3451105667524697e-05,
      "loss": 3.9001,
      "step": 1621504
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3442719720014177e-05,
      "loss": 3.8868,
      "step": 1622016
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3434333772503657e-05,
      "loss": 3.896,
      "step": 1622528
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3425947824993137e-05,
      "loss": 3.8939,
      "step": 1623040
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.341756187748262e-05,
      "loss": 3.9017,
      "step": 1623552
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.34091759299721e-05,
      "loss": 3.9098,
      "step": 1624064
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.340080636126531e-05,
      "loss": 3.8999,
      "step": 1624576
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.339242041375479e-05,
      "loss": 3.904,
      "step": 1625088
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.338403446624427e-05,
      "loss": 3.9101,
      "step": 1625600
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3375664897537482e-05,
      "loss": 3.8992,
      "step": 1626112
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3367278950026962e-05,
      "loss": 3.9059,
      "step": 1626624
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3358893002516442e-05,
      "loss": 3.8985,
      "step": 1627136
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3350507055005922e-05,
      "loss": 3.8816,
      "step": 1627648
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3342121107495402e-05,
      "loss": 3.9151,
      "step": 1628160
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3333735159984878e-05,
      "loss": 3.8992,
      "step": 1628672
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3325349212474358e-05,
      "loss": 3.9041,
      "step": 1629184
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3316963264963838e-05,
      "loss": 3.8984,
      "step": 1629696
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.330857731745332e-05,
      "loss": 3.8965,
      "step": 1630208
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.330020774874653e-05,
      "loss": 3.8925,
      "step": 1630720
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3291838180039743e-05,
      "loss": 3.8972,
      "step": 1631232
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3283452232529223e-05,
      "loss": 3.9001,
      "step": 1631744
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.32750662850187e-05,
      "loss": 3.9093,
      "step": 1632256
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3266680337508183e-05,
      "loss": 3.9022,
      "step": 1632768
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3258294389997663e-05,
      "loss": 3.9018,
      "step": 1633280
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3249908442487143e-05,
      "loss": 3.8874,
      "step": 1633792
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3241522494976623e-05,
      "loss": 3.9125,
      "step": 1634304
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3233136547466103e-05,
      "loss": 3.8863,
      "step": 1634816
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3224750599955583e-05,
      "loss": 3.8853,
      "step": 1635328
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.321636465244506e-05,
      "loss": 3.9043,
      "step": 1635840
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3207995083738275e-05,
      "loss": 3.9048,
      "step": 1636352
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3199609136227755e-05,
      "loss": 3.886,
      "step": 1636864
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3191223188717232e-05,
      "loss": 3.8951,
      "step": 1637376
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3182837241206712e-05,
      "loss": 3.8818,
      "step": 1637888
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.317445129369619e-05,
      "loss": 3.9066,
      "step": 1638400
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.316606534618567e-05,
      "loss": 3.9019,
      "step": 1638912
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.315767939867515e-05,
      "loss": 3.8999,
      "step": 1639424
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.314929345116463e-05,
      "loss": 3.9005,
      "step": 1639936
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3140923882457844e-05,
      "loss": 3.9021,
      "step": 1640448
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3132537934947324e-05,
      "loss": 3.9151,
      "step": 1640960
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3124151987436804e-05,
      "loss": 3.889,
      "step": 1641472
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3115782418730013e-05,
      "loss": 3.9042,
      "step": 1641984
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3107396471219493e-05,
      "loss": 3.8998,
      "step": 1642496
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3099010523708976e-05,
      "loss": 3.8899,
      "step": 1643008
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3090624576198456e-05,
      "loss": 3.9071,
      "step": 1643520
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3082255007491665e-05,
      "loss": 3.9052,
      "step": 1644032
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3073869059981145e-05,
      "loss": 3.9044,
      "step": 1644544
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3065483112470625e-05,
      "loss": 3.9136,
      "step": 1645056
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3057113543763838e-05,
      "loss": 3.892,
      "step": 1645568
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3048727596253318e-05,
      "loss": 3.8961,
      "step": 1646080
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3040341648742798e-05,
      "loss": 3.8926,
      "step": 1646592
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3031955701232278e-05,
      "loss": 3.8991,
      "step": 1647104
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3023569753721758e-05,
      "loss": 3.9032,
      "step": 1647616
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.3015183806211238e-05,
      "loss": 3.8999,
      "step": 1648128
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.3006797858700714e-05,
      "loss": 3.8994,
      "step": 1648640
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2998411911190197e-05,
      "loss": 3.91,
      "step": 1649152
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2990025963679677e-05,
      "loss": 3.8975,
      "step": 1649664
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2981640016169157e-05,
      "loss": 3.899,
      "step": 1650176
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2973254068658637e-05,
      "loss": 3.9037,
      "step": 1650688
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2964868121148117e-05,
      "loss": 3.9064,
      "step": 1651200
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2956498552441326e-05,
      "loss": 3.9069,
      "step": 1651712
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2948112604930806e-05,
      "loss": 3.8929,
      "step": 1652224
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.293974303622402e-05,
      "loss": 3.9054,
      "step": 1652736
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.29313570887135e-05,
      "loss": 3.9022,
      "step": 1653248
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.292297114120298e-05,
      "loss": 3.9112,
      "step": 1653760
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.291458519369246e-05,
      "loss": 3.9019,
      "step": 1654272
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.290619924618194e-05,
      "loss": 3.9007,
      "step": 1654784
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2897813298671415e-05,
      "loss": 3.8931,
      "step": 1655296
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.288944372996463e-05,
      "loss": 3.9116,
      "step": 1655808
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.288105778245411e-05,
      "loss": 3.8795,
      "step": 1656320
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.287267183494359e-05,
      "loss": 3.8955,
      "step": 1656832
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2864285887433068e-05,
      "loss": 3.9006,
      "step": 1657344
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.285591631872628e-05,
      "loss": 3.8995,
      "step": 1657856
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.284753037121576e-05,
      "loss": 3.8942,
      "step": 1658368
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.283914442370524e-05,
      "loss": 3.8902,
      "step": 1658880
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.283075847619472e-05,
      "loss": 3.8969,
      "step": 1659392
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.28223725286842e-05,
      "loss": 3.91,
      "step": 1659904
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.281398658117368e-05,
      "loss": 3.9006,
      "step": 1660416
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.280560063366316e-05,
      "loss": 3.8993,
      "step": 1660928
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.279721468615264e-05,
      "loss": 3.9048,
      "step": 1661440
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2788845117445852e-05,
      "loss": 3.9133,
      "step": 1661952
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2780459169935332e-05,
      "loss": 3.9058,
      "step": 1662464
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2772073222424812e-05,
      "loss": 3.8919,
      "step": 1662976
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2763687274914292e-05,
      "loss": 3.9005,
      "step": 1663488
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.27553177062075e-05,
      "loss": 3.9031,
      "step": 1664000
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.274693175869698e-05,
      "loss": 3.8892,
      "step": 1664512
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.273854581118646e-05,
      "loss": 3.8913,
      "step": 1665024
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2730159863675944e-05,
      "loss": 3.8978,
      "step": 1665536
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2721773916165424e-05,
      "loss": 3.8977,
      "step": 1666048
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.27133879686549e-05,
      "loss": 3.8998,
      "step": 1666560
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.270500202114438e-05,
      "loss": 3.908,
      "step": 1667072
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.269661607363386e-05,
      "loss": 3.8886,
      "step": 1667584
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2688262883730806e-05,
      "loss": 3.8966,
      "step": 1668096
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2679876936220286e-05,
      "loss": 3.8902,
      "step": 1668608
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2671490988709766e-05,
      "loss": 3.9116,
      "step": 1669120
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2663105041199246e-05,
      "loss": 3.9041,
      "step": 1669632
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2654719093688722e-05,
      "loss": 3.9059,
      "step": 1670144
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2646333146178202e-05,
      "loss": 3.8966,
      "step": 1670656
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2637963577471415e-05,
      "loss": 3.9044,
      "step": 1671168
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2629577629960895e-05,
      "loss": 3.9076,
      "step": 1671680
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2621208061254107e-05,
      "loss": 3.9028,
      "step": 1672192
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2612822113743587e-05,
      "loss": 3.9013,
      "step": 1672704
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2604436166233067e-05,
      "loss": 3.904,
      "step": 1673216
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2596050218722544e-05,
      "loss": 3.8984,
      "step": 1673728
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2587664271212024e-05,
      "loss": 3.9129,
      "step": 1674240
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2579278323701507e-05,
      "loss": 3.8914,
      "step": 1674752
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2570892376190987e-05,
      "loss": 3.9013,
      "step": 1675264
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2562506428680467e-05,
      "loss": 3.8906,
      "step": 1675776
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2554120481169947e-05,
      "loss": 3.8928,
      "step": 1676288
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2545750912463156e-05,
      "loss": 3.906,
      "step": 1676800
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2537364964952636e-05,
      "loss": 3.8958,
      "step": 1677312
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.2528979017442116e-05,
      "loss": 3.9105,
      "step": 1677824
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.25205930699316e-05,
      "loss": 3.8924,
      "step": 1678336
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.251220712242108e-05,
      "loss": 3.9047,
      "step": 1678848
    },
    {
      "epoch": 1.03,
      "eval_loss": 3.994616985321045,
      "eval_runtime": 289.612,
      "eval_samples_per_second": 1317.594,
      "eval_steps_per_second": 41.176,
      "step": 1679040
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2503821174910556e-05,
      "loss": 3.9104,
      "step": 1679360
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2495435227400036e-05,
      "loss": 3.8849,
      "step": 1679872
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2487049279889516e-05,
      "loss": 3.9063,
      "step": 1680384
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2478663332378996e-05,
      "loss": 3.9013,
      "step": 1680896
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2470277384868476e-05,
      "loss": 3.9132,
      "step": 1681408
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2461891437357955e-05,
      "loss": 3.8982,
      "step": 1681920
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2453505489847435e-05,
      "loss": 3.8879,
      "step": 1682432
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2445119542336915e-05,
      "loss": 3.891,
      "step": 1682944
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2436733594826395e-05,
      "loss": 3.8989,
      "step": 1683456
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2428347647315875e-05,
      "loss": 3.9051,
      "step": 1683968
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2419961699805355e-05,
      "loss": 3.9045,
      "step": 1684480
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2411575752294835e-05,
      "loss": 3.9091,
      "step": 1684992
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2403206183588048e-05,
      "loss": 3.8998,
      "step": 1685504
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2394820236077528e-05,
      "loss": 3.8988,
      "step": 1686016
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2386434288567008e-05,
      "loss": 3.892,
      "step": 1686528
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2378048341056488e-05,
      "loss": 3.8856,
      "step": 1687040
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2369662393545968e-05,
      "loss": 3.9009,
      "step": 1687552
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2361292824839177e-05,
      "loss": 3.8927,
      "step": 1688064
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2352906877328657e-05,
      "loss": 3.8954,
      "step": 1688576
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.234452092981814e-05,
      "loss": 3.9054,
      "step": 1689088
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2336134982307616e-05,
      "loss": 3.9081,
      "step": 1689600
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2327749034797096e-05,
      "loss": 3.9019,
      "step": 1690112
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2319363087286576e-05,
      "loss": 3.8978,
      "step": 1690624
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2310977139776056e-05,
      "loss": 3.9013,
      "step": 1691136
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2302591192265536e-05,
      "loss": 3.8979,
      "step": 1691648
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2294205244755016e-05,
      "loss": 3.8993,
      "step": 1692160
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.228583567604823e-05,
      "loss": 3.8904,
      "step": 1692672
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.227744972853771e-05,
      "loss": 3.8963,
      "step": 1693184
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.226906378102719e-05,
      "loss": 3.8952,
      "step": 1693696
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.226067783351667e-05,
      "loss": 3.896,
      "step": 1694208
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2252291886006145e-05,
      "loss": 3.896,
      "step": 1694720
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.224392231729936e-05,
      "loss": 3.9021,
      "step": 1695232
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.223553636978884e-05,
      "loss": 3.8991,
      "step": 1695744
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.222715042227832e-05,
      "loss": 3.9012,
      "step": 1696256
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2218764474767798e-05,
      "loss": 3.8904,
      "step": 1696768
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.221039490606101e-05,
      "loss": 3.9021,
      "step": 1697280
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.220200895855049e-05,
      "loss": 3.8942,
      "step": 1697792
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.219362301103997e-05,
      "loss": 3.8815,
      "step": 1698304
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.218523706352945e-05,
      "loss": 3.8866,
      "step": 1698816
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.217685111601893e-05,
      "loss": 3.8869,
      "step": 1699328
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.216846516850841e-05,
      "loss": 3.8919,
      "step": 1699840
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.216007922099789e-05,
      "loss": 3.9016,
      "step": 1700352
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.215169327348737e-05,
      "loss": 3.8947,
      "step": 1700864
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.214330732597685e-05,
      "loss": 3.9005,
      "step": 1701376
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2134921378466326e-05,
      "loss": 3.9046,
      "step": 1701888
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.212653543095581e-05,
      "loss": 3.8925,
      "step": 1702400
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2118165862249022e-05,
      "loss": 3.8965,
      "step": 1702912
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2109779914738502e-05,
      "loss": 3.8945,
      "step": 1703424
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.210139396722798e-05,
      "loss": 3.8685,
      "step": 1703936
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.209300801971746e-05,
      "loss": 3.9129,
      "step": 1704448
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.208463845101067e-05,
      "loss": 3.8903,
      "step": 1704960
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2076252503500154e-05,
      "loss": 3.9063,
      "step": 1705472
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.206786655598963e-05,
      "loss": 3.8875,
      "step": 1705984
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.205948060847911e-05,
      "loss": 3.893,
      "step": 1706496
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2051111039772323e-05,
      "loss": 3.8863,
      "step": 1707008
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.20427250922618e-05,
      "loss": 3.8878,
      "step": 1707520
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.203433914475128e-05,
      "loss": 3.8939,
      "step": 1708032
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2025953197240763e-05,
      "loss": 3.9037,
      "step": 1708544
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2017567249730243e-05,
      "loss": 3.9007,
      "step": 1709056
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2009181302219723e-05,
      "loss": 3.8869,
      "step": 1709568
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2000795354709203e-05,
      "loss": 3.8805,
      "step": 1710080
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1992409407198683e-05,
      "loss": 3.9091,
      "step": 1710592
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1984039838491892e-05,
      "loss": 3.8781,
      "step": 1711104
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1975653890981372e-05,
      "loss": 3.874,
      "step": 1711616
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1967267943470855e-05,
      "loss": 3.899,
      "step": 1712128
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1958881995960335e-05,
      "loss": 3.9031,
      "step": 1712640
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1950496048449812e-05,
      "loss": 3.8775,
      "step": 1713152
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1942110100939292e-05,
      "loss": 3.888,
      "step": 1713664
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1933740532232504e-05,
      "loss": 3.8792,
      "step": 1714176
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.192535458472198e-05,
      "loss": 3.8981,
      "step": 1714688
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1916968637211464e-05,
      "loss": 3.8972,
      "step": 1715200
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1908582689700944e-05,
      "loss": 3.8948,
      "step": 1715712
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1900213120994157e-05,
      "loss": 3.8893,
      "step": 1716224
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1891827173483633e-05,
      "loss": 3.8956,
      "step": 1716736
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1883441225973113e-05,
      "loss": 3.9079,
      "step": 1717248
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1875055278462593e-05,
      "loss": 3.8807,
      "step": 1717760
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1866669330952073e-05,
      "loss": 3.8974,
      "step": 1718272
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1858299762245286e-05,
      "loss": 3.8942,
      "step": 1718784
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1849913814734766e-05,
      "loss": 3.8835,
      "step": 1719296
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1841527867224246e-05,
      "loss": 3.8985,
      "step": 1719808
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1833141919713726e-05,
      "loss": 3.9005,
      "step": 1720320
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1824755972203206e-05,
      "loss": 3.8985,
      "step": 1720832
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1816370024692685e-05,
      "loss": 3.9081,
      "step": 1721344
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1808000455985898e-05,
      "loss": 3.8829,
      "step": 1721856
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1799630887279107e-05,
      "loss": 3.8907,
      "step": 1722368
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1791244939768587e-05,
      "loss": 3.8865,
      "step": 1722880
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1782858992258067e-05,
      "loss": 3.8954,
      "step": 1723392
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1774473044747547e-05,
      "loss": 3.895,
      "step": 1723904
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.176610347604076e-05,
      "loss": 3.9003,
      "step": 1724416
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.175771752853024e-05,
      "loss": 3.8874,
      "step": 1724928
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.174933158101972e-05,
      "loss": 3.9046,
      "step": 1725440
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.17409456335092e-05,
      "loss": 3.8934,
      "step": 1725952
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.173255968599868e-05,
      "loss": 3.8916,
      "step": 1726464
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.172417373848816e-05,
      "loss": 3.8985,
      "step": 1726976
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.171578779097764e-05,
      "loss": 3.8969,
      "step": 1727488
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.170740184346712e-05,
      "loss": 3.9032,
      "step": 1728000
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.16990158959566e-05,
      "loss": 3.8864,
      "step": 1728512
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.169062994844608e-05,
      "loss": 3.8973,
      "step": 1729024
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.168224400093556e-05,
      "loss": 3.8982,
      "step": 1729536
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.167385805342504e-05,
      "loss": 3.9036,
      "step": 1730048
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.166547210591452e-05,
      "loss": 3.8986,
      "step": 1730560
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1657086158403995e-05,
      "loss": 3.8932,
      "step": 1731072
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.164870021089348e-05,
      "loss": 3.891,
      "step": 1731584
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.164031426338296e-05,
      "loss": 3.9018,
      "step": 1732096
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1631944694676168e-05,
      "loss": 3.8744,
      "step": 1732608
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1623558747165648e-05,
      "loss": 3.8907,
      "step": 1733120
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1615172799655128e-05,
      "loss": 3.8901,
      "step": 1733632
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.160680323094834e-05,
      "loss": 3.8938,
      "step": 1734144
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.159841728343782e-05,
      "loss": 3.8868,
      "step": 1734656
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1590047714731033e-05,
      "loss": 3.8866,
      "step": 1735168
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1581661767220513e-05,
      "loss": 3.8884,
      "step": 1735680
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1573275819709993e-05,
      "loss": 3.903,
      "step": 1736192
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.156488987219947e-05,
      "loss": 3.8972,
      "step": 1736704
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.155650392468895e-05,
      "loss": 3.8901,
      "step": 1737216
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1548117977178432e-05,
      "loss": 3.8985,
      "step": 1737728
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1539732029667912e-05,
      "loss": 3.9082,
      "step": 1738240
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1531346082157392e-05,
      "loss": 3.9033,
      "step": 1738752
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1522960134646872e-05,
      "loss": 3.8846,
      "step": 1739264
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.151457418713635e-05,
      "loss": 3.8946,
      "step": 1739776
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.150618823962583e-05,
      "loss": 3.8983,
      "step": 1740288
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.149780229211531e-05,
      "loss": 3.882,
      "step": 1740800
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1489432723408525e-05,
      "loss": 3.8863,
      "step": 1741312
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1481046775898e-05,
      "loss": 3.8928,
      "step": 1741824
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.147266082838748e-05,
      "loss": 3.8889,
      "step": 1742336
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.146427488087696e-05,
      "loss": 3.8963,
      "step": 1742848
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.145588893336644e-05,
      "loss": 3.9015,
      "step": 1743360
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.144750298585592e-05,
      "loss": 3.884,
      "step": 1743872
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1439133417149134e-05,
      "loss": 3.8868,
      "step": 1744384
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1430747469638613e-05,
      "loss": 3.8923,
      "step": 1744896
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1422361522128093e-05,
      "loss": 3.9029,
      "step": 1745408
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1413975574617573e-05,
      "loss": 3.8973,
      "step": 1745920
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1405589627107053e-05,
      "loss": 3.9015,
      "step": 1746432
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.139720367959653e-05,
      "loss": 3.8887,
      "step": 1746944
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.138881773208601e-05,
      "loss": 3.8982,
      "step": 1747456
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.138043178457549e-05,
      "loss": 3.9016,
      "step": 1747968
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1372078594672435e-05,
      "loss": 3.8956,
      "step": 1748480
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1363692647161915e-05,
      "loss": 3.8963,
      "step": 1748992
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1355306699651395e-05,
      "loss": 3.8955,
      "step": 1749504
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1346920752140875e-05,
      "loss": 3.8932,
      "step": 1750016
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1338534804630355e-05,
      "loss": 3.9067,
      "step": 1750528
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1330148857119835e-05,
      "loss": 3.8883,
      "step": 1751040
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1321762909609315e-05,
      "loss": 3.8941,
      "step": 1751552
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1313376962098795e-05,
      "loss": 3.8862,
      "step": 1752064
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1305007393392004e-05,
      "loss": 3.8861,
      "step": 1752576
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1296637824685216e-05,
      "loss": 3.8934,
      "step": 1753088
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1288251877174696e-05,
      "loss": 3.8925,
      "step": 1753600
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.127986592966418e-05,
      "loss": 3.9001,
      "step": 1754112
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1271479982153656e-05,
      "loss": 3.8864,
      "step": 1754624
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.1263094034643136e-05,
      "loss": 3.9019,
      "step": 1755136
    },
    {
      "epoch": 0.03,
      "eval_loss": 3.9924521446228027,
      "eval_runtime": 285.4504,
      "eval_samples_per_second": 1336.803,
      "eval_steps_per_second": 41.776,
      "step": 1755360
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.125472446593635e-05,
      "loss": 3.8902,
      "step": 1755648
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.124633851842583e-05,
      "loss": 3.8797,
      "step": 1756160
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1237952570915305e-05,
      "loss": 3.9008,
      "step": 1756672
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1229566623404788e-05,
      "loss": 3.892,
      "step": 1757184
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1221180675894268e-05,
      "loss": 3.9042,
      "step": 1757696
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1212811107187477e-05,
      "loss": 3.8957,
      "step": 1758208
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1204425159676957e-05,
      "loss": 3.8849,
      "step": 1758720
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1196039212166437e-05,
      "loss": 3.8874,
      "step": 1759232
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1187653264655917e-05,
      "loss": 3.892,
      "step": 1759744
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1179267317145397e-05,
      "loss": 3.8958,
      "step": 1760256
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.117089774843861e-05,
      "loss": 3.8961,
      "step": 1760768
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.116251180092809e-05,
      "loss": 3.903,
      "step": 1761280
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1154142232221302e-05,
      "loss": 3.893,
      "step": 1761792
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.114575628471078e-05,
      "loss": 3.8926,
      "step": 1762304
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.113737033720026e-05,
      "loss": 3.8875,
      "step": 1762816
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1128984389689742e-05,
      "loss": 3.8778,
      "step": 1763328
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1120598442179222e-05,
      "loss": 3.8959,
      "step": 1763840
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1112212494668702e-05,
      "loss": 3.8896,
      "step": 1764352
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1103826547158182e-05,
      "loss": 3.8889,
      "step": 1764864
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.109544059964766e-05,
      "loss": 3.8911,
      "step": 1765376
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.108705465213714e-05,
      "loss": 3.9077,
      "step": 1765888
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.107866870462662e-05,
      "loss": 3.8952,
      "step": 1766400
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1070299135919834e-05,
      "loss": 3.8925,
      "step": 1766912
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.106191318840931e-05,
      "loss": 3.8917,
      "step": 1767424
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.105352724089879e-05,
      "loss": 3.8913,
      "step": 1767936
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.104514129338827e-05,
      "loss": 3.8927,
      "step": 1768448
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.103675534587775e-05,
      "loss": 3.887,
      "step": 1768960
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.102838577717096e-05,
      "loss": 3.8929,
      "step": 1769472
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1019999829660443e-05,
      "loss": 3.8864,
      "step": 1769984
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.1011613882149923e-05,
      "loss": 3.8859,
      "step": 1770496
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.1003227934639403e-05,
      "loss": 3.888,
      "step": 1771008
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0994841987128883e-05,
      "loss": 3.8967,
      "step": 1771520
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0986456039618363e-05,
      "loss": 3.8964,
      "step": 1772032
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0978086470911572e-05,
      "loss": 3.8936,
      "step": 1772544
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0969700523401052e-05,
      "loss": 3.8857,
      "step": 1773056
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0961330954694264e-05,
      "loss": 3.8905,
      "step": 1773568
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0952945007183744e-05,
      "loss": 3.8885,
      "step": 1774080
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0944559059673224e-05,
      "loss": 3.8795,
      "step": 1774592
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0936173112162704e-05,
      "loss": 3.8809,
      "step": 1775104
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0927787164652184e-05,
      "loss": 3.8803,
      "step": 1775616
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0919401217141664e-05,
      "loss": 3.8892,
      "step": 1776128
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0911015269631144e-05,
      "loss": 3.8921,
      "step": 1776640
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0902629322120624e-05,
      "loss": 3.8881,
      "step": 1777152
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0894243374610104e-05,
      "loss": 3.8975,
      "step": 1777664
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0885873805903313e-05,
      "loss": 3.896,
      "step": 1778176
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0877487858392793e-05,
      "loss": 3.8871,
      "step": 1778688
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0869101910882273e-05,
      "loss": 3.8942,
      "step": 1779200
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0860715963371753e-05,
      "loss": 3.8875,
      "step": 1779712
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0852330015861236e-05,
      "loss": 3.8635,
      "step": 1780224
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0843960447154445e-05,
      "loss": 3.9019,
      "step": 1780736
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0835574499643925e-05,
      "loss": 3.8833,
      "step": 1781248
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0827188552133405e-05,
      "loss": 3.8998,
      "step": 1781760
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0818802604622885e-05,
      "loss": 3.8856,
      "step": 1782272
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0810433035916098e-05,
      "loss": 3.8868,
      "step": 1782784
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0802047088405578e-05,
      "loss": 3.8776,
      "step": 1783296
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0793661140895058e-05,
      "loss": 3.8825,
      "step": 1783808
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0785275193384538e-05,
      "loss": 3.8916,
      "step": 1784320
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0776889245874018e-05,
      "loss": 3.8924,
      "step": 1784832
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0768503298363494e-05,
      "loss": 3.8937,
      "step": 1785344
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0760117350852974e-05,
      "loss": 3.8852,
      "step": 1785856
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0751731403342458e-05,
      "loss": 3.8734,
      "step": 1786368
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.074336183463567e-05,
      "loss": 3.904,
      "step": 1786880
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0734975887125147e-05,
      "loss": 3.8748,
      "step": 1787392
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0726589939614627e-05,
      "loss": 3.8643,
      "step": 1787904
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0718203992104106e-05,
      "loss": 3.893,
      "step": 1788416
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.070983442339732e-05,
      "loss": 3.8953,
      "step": 1788928
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.070146485469053e-05,
      "loss": 3.8743,
      "step": 1789440
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.069307890718001e-05,
      "loss": 3.8836,
      "step": 1789952
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.068469295966949e-05,
      "loss": 3.8697,
      "step": 1790464
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0676307012158968e-05,
      "loss": 3.8901,
      "step": 1790976
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0667921064648448e-05,
      "loss": 3.8931,
      "step": 1791488
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0659535117137928e-05,
      "loss": 3.8858,
      "step": 1792000
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.065114916962741e-05,
      "loss": 3.8852,
      "step": 1792512
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.064276322211689e-05,
      "loss": 3.8915,
      "step": 1793024
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.063437727460637e-05,
      "loss": 3.9014,
      "step": 1793536
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.062599132709585e-05,
      "loss": 3.8756,
      "step": 1794048
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0617605379585328e-05,
      "loss": 3.8942,
      "step": 1794560
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0609219432074808e-05,
      "loss": 3.8848,
      "step": 1795072
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.060084986336802e-05,
      "loss": 3.8828,
      "step": 1795584
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0592480294661233e-05,
      "loss": 3.8914,
      "step": 1796096
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0584094347150712e-05,
      "loss": 3.8948,
      "step": 1796608
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0575708399640192e-05,
      "loss": 3.8876,
      "step": 1797120
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0567322452129672e-05,
      "loss": 3.9054,
      "step": 1797632
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.055895288342288e-05,
      "loss": 3.8761,
      "step": 1798144
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0550566935912365e-05,
      "loss": 3.8815,
      "step": 1798656
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0542180988401845e-05,
      "loss": 3.8825,
      "step": 1799168
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.053379504089132e-05,
      "loss": 3.8895,
      "step": 1799680
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.05254090933808e-05,
      "loss": 3.8866,
      "step": 1800192
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.0517039524674014e-05,
      "loss": 3.8957,
      "step": 1800704
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0508653577163494e-05,
      "loss": 3.8798,
      "step": 1801216
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0500267629652974e-05,
      "loss": 3.899,
      "step": 1801728
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0491881682142454e-05,
      "loss": 3.8906,
      "step": 1802240
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0483495734631934e-05,
      "loss": 3.8831,
      "step": 1802752
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0475109787121414e-05,
      "loss": 3.8926,
      "step": 1803264
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0466740218414623e-05,
      "loss": 3.8938,
      "step": 1803776
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0458354270904103e-05,
      "loss": 3.8978,
      "step": 1804288
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0449968323393583e-05,
      "loss": 3.8825,
      "step": 1804800
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0441598754686795e-05,
      "loss": 3.8881,
      "step": 1805312
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0433212807176275e-05,
      "loss": 3.8917,
      "step": 1805824
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0424826859665755e-05,
      "loss": 3.9006,
      "step": 1806336
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0416440912155235e-05,
      "loss": 3.894,
      "step": 1806848
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0408054964644715e-05,
      "loss": 3.8847,
      "step": 1807360
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0399669017134195e-05,
      "loss": 3.8864,
      "step": 1807872
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0391283069623675e-05,
      "loss": 3.898,
      "step": 1808384
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0382897122113155e-05,
      "loss": 3.8661,
      "step": 1808896
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0374511174602635e-05,
      "loss": 3.8855,
      "step": 1809408
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0366141605895847e-05,
      "loss": 3.8827,
      "step": 1809920
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0357755658385327e-05,
      "loss": 3.8893,
      "step": 1810432
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0349369710874804e-05,
      "loss": 3.879,
      "step": 1810944
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0340983763364284e-05,
      "loss": 3.8837,
      "step": 1811456
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.03326141946575e-05,
      "loss": 3.8875,
      "step": 1811968
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0324228247146976e-05,
      "loss": 3.8912,
      "step": 1812480
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0315842299636456e-05,
      "loss": 3.8874,
      "step": 1812992
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0307456352125936e-05,
      "loss": 3.8862,
      "step": 1813504
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0299070404615416e-05,
      "loss": 3.8936,
      "step": 1814016
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0290684457104896e-05,
      "loss": 3.9019,
      "step": 1814528
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0282298509594376e-05,
      "loss": 3.895,
      "step": 1815040
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.027391256208386e-05,
      "loss": 3.8791,
      "step": 1815552
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.026554299337707e-05,
      "loss": 3.8875,
      "step": 1816064
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0257157045866548e-05,
      "loss": 3.8922,
      "step": 1816576
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0248771098356028e-05,
      "loss": 3.8763,
      "step": 1817088
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0240401529649237e-05,
      "loss": 3.8807,
      "step": 1817600
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.023201558213872e-05,
      "loss": 3.8863,
      "step": 1818112
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.02236296346282e-05,
      "loss": 3.8849,
      "step": 1818624
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.021524368711768e-05,
      "loss": 3.884,
      "step": 1819136
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.020687411841089e-05,
      "loss": 3.8951,
      "step": 1819648
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.019848817090037e-05,
      "loss": 3.8838,
      "step": 1820160
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.019010222338985e-05,
      "loss": 3.8831,
      "step": 1820672
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.018171627587933e-05,
      "loss": 3.881,
      "step": 1821184
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0173346707172542e-05,
      "loss": 3.9017,
      "step": 1821696
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0164960759662022e-05,
      "loss": 3.8877,
      "step": 1822208
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0156574812151502e-05,
      "loss": 3.8961,
      "step": 1822720
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0148188864640982e-05,
      "loss": 3.882,
      "step": 1823232
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.013980291713046e-05,
      "loss": 3.8938,
      "step": 1823744
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0131433348423674e-05,
      "loss": 3.8931,
      "step": 1824256
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0123047400913154e-05,
      "loss": 3.8916,
      "step": 1824768
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.011466145340263e-05,
      "loss": 3.8901,
      "step": 1825280
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.010627550589211e-05,
      "loss": 3.8881,
      "step": 1825792
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.009788955838159e-05,
      "loss": 3.8888,
      "step": 1826304
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.008950361087107e-05,
      "loss": 3.8994,
      "step": 1826816
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.008111766336055e-05,
      "loss": 3.8845,
      "step": 1827328
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0072748094653763e-05,
      "loss": 3.8889,
      "step": 1827840
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0064362147143243e-05,
      "loss": 3.8845,
      "step": 1828352
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0055976199632723e-05,
      "loss": 3.8824,
      "step": 1828864
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0047590252122203e-05,
      "loss": 3.8822,
      "step": 1829376
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0039220683415412e-05,
      "loss": 3.8869,
      "step": 1829888
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0030834735904892e-05,
      "loss": 3.8973,
      "step": 1830400
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0022448788394375e-05,
      "loss": 3.8816,
      "step": 1830912
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.0014062840883855e-05,
      "loss": 3.8934,
      "step": 1831424
    },
    {
      "epoch": 1.03,
      "eval_loss": 3.9899635314941406,
      "eval_runtime": 289.7192,
      "eval_samples_per_second": 1317.106,
      "eval_steps_per_second": 41.161,
      "step": 1831680
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.0005676893373335e-05,
      "loss": 3.8924,
      "step": 1831936
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9997290945862812e-05,
      "loss": 3.8758,
      "step": 1832448
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9988904998352292e-05,
      "loss": 3.8952,
      "step": 1832960
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9980519050841772e-05,
      "loss": 3.8881,
      "step": 1833472
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9972133103331252e-05,
      "loss": 3.8974,
      "step": 1833984
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9963747155820732e-05,
      "loss": 3.8892,
      "step": 1834496
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9955361208310215e-05,
      "loss": 3.8865,
      "step": 1835008
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9946975260799695e-05,
      "loss": 3.8786,
      "step": 1835520
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.993858931328917e-05,
      "loss": 3.8862,
      "step": 1836032
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.993020336577865e-05,
      "loss": 3.8935,
      "step": 1836544
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.992181741826813e-05,
      "loss": 3.8877,
      "step": 1837056
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.991343147075761e-05,
      "loss": 3.895,
      "step": 1837568
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9905061902050824e-05,
      "loss": 3.8953,
      "step": 1838080
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9896675954540304e-05,
      "loss": 3.8816,
      "step": 1838592
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9888290007029784e-05,
      "loss": 3.8785,
      "step": 1839104
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9879904059519264e-05,
      "loss": 3.8792,
      "step": 1839616
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9871518112008744e-05,
      "loss": 3.8883,
      "step": 1840128
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9863132164498224e-05,
      "loss": 3.8854,
      "step": 1840640
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.98547462169877e-05,
      "loss": 3.8815,
      "step": 1841152
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9846360269477184e-05,
      "loss": 3.889,
      "step": 1841664
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9837974321966664e-05,
      "loss": 3.8992,
      "step": 1842176
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9829604753259876e-05,
      "loss": 3.8911,
      "step": 1842688
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9821218805749353e-05,
      "loss": 3.8883,
      "step": 1843200
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9812832858238833e-05,
      "loss": 3.8844,
      "step": 1843712
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9804446910728313e-05,
      "loss": 3.8858,
      "step": 1844224
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9796060963217793e-05,
      "loss": 3.8889,
      "step": 1844736
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9787675015707276e-05,
      "loss": 3.8774,
      "step": 1845248
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9779305447000485e-05,
      "loss": 3.891,
      "step": 1845760
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9770919499489965e-05,
      "loss": 3.8819,
      "step": 1846272
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9762533551979445e-05,
      "loss": 3.8782,
      "step": 1846784
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9754147604468925e-05,
      "loss": 3.8849,
      "step": 1847296
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9745778035762137e-05,
      "loss": 3.8853,
      "step": 1847808
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9737392088251617e-05,
      "loss": 3.8884,
      "step": 1848320
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9729006140741097e-05,
      "loss": 3.891,
      "step": 1848832
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9720620193230577e-05,
      "loss": 3.8786,
      "step": 1849344
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9712250624523786e-05,
      "loss": 3.8871,
      "step": 1849856
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9703864677013266e-05,
      "loss": 3.8905,
      "step": 1850368
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9695478729502746e-05,
      "loss": 3.8703,
      "step": 1850880
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.968709278199223e-05,
      "loss": 3.8754,
      "step": 1851392
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9678706834481706e-05,
      "loss": 3.8755,
      "step": 1851904
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9670320886971186e-05,
      "loss": 3.8806,
      "step": 1852416
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9661934939460666e-05,
      "loss": 3.8859,
      "step": 1852928
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9653548991950146e-05,
      "loss": 3.884,
      "step": 1853440
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9645163044439626e-05,
      "loss": 3.8961,
      "step": 1853952
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.963679347573284e-05,
      "loss": 3.8837,
      "step": 1854464
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.962840752822232e-05,
      "loss": 3.8852,
      "step": 1854976
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.96200215807118e-05,
      "loss": 3.8848,
      "step": 1855488
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9611635633201278e-05,
      "loss": 3.8835,
      "step": 1856000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9603249685690758e-05,
      "loss": 3.8588,
      "step": 1856512
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9594863738180238e-05,
      "loss": 3.8981,
      "step": 1857024
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9586477790669715e-05,
      "loss": 3.879,
      "step": 1857536
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.957812460076666e-05,
      "loss": 3.8902,
      "step": 1858048
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.956973865325614e-05,
      "loss": 3.8811,
      "step": 1858560
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.956135270574562e-05,
      "loss": 3.8837,
      "step": 1859072
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.95529667582351e-05,
      "loss": 3.8674,
      "step": 1859584
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.954458081072458e-05,
      "loss": 3.8765,
      "step": 1860096
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.953619486321406e-05,
      "loss": 3.8899,
      "step": 1860608
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.952780891570354e-05,
      "loss": 3.8808,
      "step": 1861120
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.951942296819302e-05,
      "loss": 3.8894,
      "step": 1861632
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.95110370206825e-05,
      "loss": 3.8828,
      "step": 1862144
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9502667451975712e-05,
      "loss": 3.8635,
      "step": 1862656
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.949428150446519e-05,
      "loss": 3.897,
      "step": 1863168
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.948589555695467e-05,
      "loss": 3.8714,
      "step": 1863680
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.947750960944415e-05,
      "loss": 3.8604,
      "step": 1864192
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9469123661933632e-05,
      "loss": 3.8888,
      "step": 1864704
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.946073771442311e-05,
      "loss": 3.8867,
      "step": 1865216
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.945235176691259e-05,
      "loss": 3.8681,
      "step": 1865728
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9443965819402068e-05,
      "loss": 3.8777,
      "step": 1866240
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.943559625069528e-05,
      "loss": 3.8704,
      "step": 1866752
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.942721030318476e-05,
      "loss": 3.8806,
      "step": 1867264
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9418840734477973e-05,
      "loss": 3.8865,
      "step": 1867776
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9410454786967453e-05,
      "loss": 3.8798,
      "step": 1868288
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9402085218260662e-05,
      "loss": 3.8826,
      "step": 1868800
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9393699270750142e-05,
      "loss": 3.8835,
      "step": 1869312
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9385313323239622e-05,
      "loss": 3.8955,
      "step": 1869824
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9376927375729102e-05,
      "loss": 3.8711,
      "step": 1870336
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9368541428218585e-05,
      "loss": 3.8862,
      "step": 1870848
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9360155480708065e-05,
      "loss": 3.878,
      "step": 1871360
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9351769533197542e-05,
      "loss": 3.8778,
      "step": 1871872
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9343383585687022e-05,
      "loss": 3.8861,
      "step": 1872384
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9335014016980234e-05,
      "loss": 3.8876,
      "step": 1872896
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9326628069469714e-05,
      "loss": 3.8848,
      "step": 1873408
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9318242121959194e-05,
      "loss": 3.9008,
      "step": 1873920
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9309856174448674e-05,
      "loss": 3.8698,
      "step": 1874432
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9301470226938154e-05,
      "loss": 3.8776,
      "step": 1874944
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9293084279427634e-05,
      "loss": 3.8784,
      "step": 1875456
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9284698331917114e-05,
      "loss": 3.8836,
      "step": 1875968
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9276312384406594e-05,
      "loss": 3.8795,
      "step": 1876480
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9267942815699807e-05,
      "loss": 3.8892,
      "step": 1876992
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9259556868189286e-05,
      "loss": 3.8798,
      "step": 1877504
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9251170920678766e-05,
      "loss": 3.8909,
      "step": 1878016
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9242784973168246e-05,
      "loss": 3.8847,
      "step": 1878528
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9234415404461455e-05,
      "loss": 3.8782,
      "step": 1879040
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9226029456950935e-05,
      "loss": 3.8882,
      "step": 1879552
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9217643509440415e-05,
      "loss": 3.8869,
      "step": 1880064
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9209257561929895e-05,
      "loss": 3.8933,
      "step": 1880576
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9200871614419375e-05,
      "loss": 3.8795,
      "step": 1881088
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9192485666908855e-05,
      "loss": 3.8802,
      "step": 1881600
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9184099719398335e-05,
      "loss": 3.8892,
      "step": 1882112
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9175713771887815e-05,
      "loss": 3.8927,
      "step": 1882624
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9167327824377295e-05,
      "loss": 3.8855,
      "step": 1883136
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9158941876866775e-05,
      "loss": 3.8847,
      "step": 1883648
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9150555929356255e-05,
      "loss": 3.8809,
      "step": 1884160
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9142186360649468e-05,
      "loss": 3.8909,
      "step": 1884672
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9133800413138947e-05,
      "loss": 3.8621,
      "step": 1885184
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9125414465628427e-05,
      "loss": 3.8824,
      "step": 1885696
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9117028518117904e-05,
      "loss": 3.8681,
      "step": 1886208
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9108658949411117e-05,
      "loss": 3.885,
      "step": 1886720
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.91002730019006e-05,
      "loss": 3.8769,
      "step": 1887232
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.909188705439008e-05,
      "loss": 3.884,
      "step": 1887744
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9083501106879556e-05,
      "loss": 3.8794,
      "step": 1888256
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9075115159369036e-05,
      "loss": 3.886,
      "step": 1888768
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.906674559066225e-05,
      "loss": 3.8831,
      "step": 1889280
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.9058359643151725e-05,
      "loss": 3.8818,
      "step": 1889792
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.904997369564121e-05,
      "loss": 3.8885,
      "step": 1890304
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.904158774813069e-05,
      "loss": 3.8928,
      "step": 1890816
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.903320180062017e-05,
      "loss": 3.8887,
      "step": 1891328
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.902481585310965e-05,
      "loss": 3.8797,
      "step": 1891840
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.901642990559913e-05,
      "loss": 3.8796,
      "step": 1892352
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.900804395808861e-05,
      "loss": 3.8875,
      "step": 1892864
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8999674389381818e-05,
      "loss": 3.8762,
      "step": 1893376
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.899130482067503e-05,
      "loss": 3.8718,
      "step": 1893888
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.898291887316451e-05,
      "loss": 3.8789,
      "step": 1894400
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.897453292565399e-05,
      "loss": 3.879,
      "step": 1894912
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.896614697814347e-05,
      "loss": 3.8757,
      "step": 1895424
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.895776103063295e-05,
      "loss": 3.8929,
      "step": 1895936
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8949391461926162e-05,
      "loss": 3.8779,
      "step": 1896448
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8941005514415642e-05,
      "loss": 3.878,
      "step": 1896960
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8932619566905122e-05,
      "loss": 3.8752,
      "step": 1897472
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8924233619394602e-05,
      "loss": 3.8945,
      "step": 1897984
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8915847671884082e-05,
      "loss": 3.8863,
      "step": 1898496
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.890747810317729e-05,
      "loss": 3.8864,
      "step": 1899008
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8899108534470504e-05,
      "loss": 3.8785,
      "step": 1899520
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8890722586959984e-05,
      "loss": 3.8905,
      "step": 1900032
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8882336639449464e-05,
      "loss": 3.8898,
      "step": 1900544
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8873950691938944e-05,
      "loss": 3.8877,
      "step": 1901056
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8865564744428424e-05,
      "loss": 3.8864,
      "step": 1901568
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8857178796917904e-05,
      "loss": 3.8795,
      "step": 1902080
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.884879284940738e-05,
      "loss": 3.8837,
      "step": 1902592
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8840406901896863e-05,
      "loss": 3.8943,
      "step": 1903104
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8832020954386343e-05,
      "loss": 3.8773,
      "step": 1903616
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8823651385679556e-05,
      "loss": 3.8814,
      "step": 1904128
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8815265438169032e-05,
      "loss": 3.8798,
      "step": 1904640
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8806879490658512e-05,
      "loss": 3.8785,
      "step": 1905152
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8798493543147992e-05,
      "loss": 3.8766,
      "step": 1905664
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8790107595637472e-05,
      "loss": 3.8822,
      "step": 1906176
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8781738026930685e-05,
      "loss": 3.8907,
      "step": 1906688
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8773352079420165e-05,
      "loss": 3.8793,
      "step": 1907200
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.8764966131909645e-05,
      "loss": 3.8867,
      "step": 1907712
    },
    {
      "epoch": 0.03,
      "eval_loss": 3.9878811836242676,
      "eval_runtime": 295.9503,
      "eval_samples_per_second": 1289.375,
      "eval_steps_per_second": 40.294,
      "step": 1908000
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8756580184399125e-05,
      "loss": 3.8721,
      "step": 1908224
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8748194236888605e-05,
      "loss": 3.8701,
      "step": 1908736
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8739808289378085e-05,
      "loss": 3.8892,
      "step": 1909248
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8731422341867565e-05,
      "loss": 3.8818,
      "step": 1909760
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8723036394357045e-05,
      "loss": 3.8937,
      "step": 1910272
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8714666825650257e-05,
      "loss": 3.881,
      "step": 1910784
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8706280878139737e-05,
      "loss": 3.8811,
      "step": 1911296
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8697911309432946e-05,
      "loss": 3.872,
      "step": 1911808
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8689525361922426e-05,
      "loss": 3.8813,
      "step": 1912320
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.868113941441191e-05,
      "loss": 3.8876,
      "step": 1912832
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8672753466901386e-05,
      "loss": 3.8834,
      "step": 1913344
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.86643838981946e-05,
      "loss": 3.8899,
      "step": 1913856
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8656014329487808e-05,
      "loss": 3.8955,
      "step": 1914368
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8647628381977287e-05,
      "loss": 3.8724,
      "step": 1914880
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.863924243446677e-05,
      "loss": 3.8764,
      "step": 1915392
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.863085648695625e-05,
      "loss": 3.8722,
      "step": 1915904
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.862247053944573e-05,
      "loss": 3.885,
      "step": 1916416
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.861408459193521e-05,
      "loss": 3.874,
      "step": 1916928
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8605698644424687e-05,
      "loss": 3.8779,
      "step": 1917440
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8597312696914167e-05,
      "loss": 3.884,
      "step": 1917952
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.858894312820738e-05,
      "loss": 3.8924,
      "step": 1918464
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8580573559500592e-05,
      "loss": 3.8869,
      "step": 1918976
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8572187611990072e-05,
      "loss": 3.8818,
      "step": 1919488
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8563801664479552e-05,
      "loss": 3.8817,
      "step": 1920000
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8555415716969032e-05,
      "loss": 3.8796,
      "step": 1920512
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.854702976945851e-05,
      "loss": 3.8842,
      "step": 1921024
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.853864382194799e-05,
      "loss": 3.8678,
      "step": 1921536
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8530257874437472e-05,
      "loss": 3.8873,
      "step": 1922048
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8521871926926952e-05,
      "loss": 3.8777,
      "step": 1922560
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.8513485979416432e-05,
      "loss": 3.8721,
      "step": 1923072
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8505100031905912e-05,
      "loss": 3.884,
      "step": 1923584
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.849673046319912e-05,
      "loss": 3.881,
      "step": 1924096
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.84883445156886e-05,
      "loss": 3.8777,
      "step": 1924608
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.847995856817808e-05,
      "loss": 3.8895,
      "step": 1925120
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8471572620667564e-05,
      "loss": 3.8753,
      "step": 1925632
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8463203051960773e-05,
      "loss": 3.8805,
      "step": 1926144
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8454817104450253e-05,
      "loss": 3.8825,
      "step": 1926656
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8446447535743462e-05,
      "loss": 3.8618,
      "step": 1927168
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8438061588232942e-05,
      "loss": 3.8757,
      "step": 1927680
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8429675640722426e-05,
      "loss": 3.872,
      "step": 1928192
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8421289693211906e-05,
      "loss": 3.8728,
      "step": 1928704
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8412903745701385e-05,
      "loss": 3.8829,
      "step": 1929216
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8404517798190865e-05,
      "loss": 3.8752,
      "step": 1929728
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8396131850680342e-05,
      "loss": 3.8893,
      "step": 1930240
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8387745903169822e-05,
      "loss": 3.8802,
      "step": 1930752
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8379359955659302e-05,
      "loss": 3.8805,
      "step": 1931264
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8370974008148782e-05,
      "loss": 3.8798,
      "step": 1931776
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8362588060638265e-05,
      "loss": 3.8801,
      "step": 1932288
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8354202113127745e-05,
      "loss": 3.859,
      "step": 1932800
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8345816165617222e-05,
      "loss": 3.8862,
      "step": 1933312
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8337446596910434e-05,
      "loss": 3.8738,
      "step": 1933824
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8329060649399914e-05,
      "loss": 3.8878,
      "step": 1934336
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8320674701889394e-05,
      "loss": 3.8713,
      "step": 1934848
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8312288754378874e-05,
      "loss": 3.8775,
      "step": 1935360
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8303902806868354e-05,
      "loss": 3.8645,
      "step": 1935872
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8295516859357834e-05,
      "loss": 3.8741,
      "step": 1936384
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8287130911847314e-05,
      "loss": 3.884,
      "step": 1936896
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8278744964336794e-05,
      "loss": 3.8737,
      "step": 1937408
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8270359016826274e-05,
      "loss": 3.8893,
      "step": 1937920
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.826200582692322e-05,
      "loss": 3.8794,
      "step": 1938432
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8253619879412695e-05,
      "loss": 3.8547,
      "step": 1938944
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8245233931902175e-05,
      "loss": 3.8917,
      "step": 1939456
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8236847984391655e-05,
      "loss": 3.8677,
      "step": 1939968
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8228478415684868e-05,
      "loss": 3.8595,
      "step": 1940480
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8220092468174348e-05,
      "loss": 3.8809,
      "step": 1940992
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8211706520663828e-05,
      "loss": 3.8827,
      "step": 1941504
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8203320573153308e-05,
      "loss": 3.8664,
      "step": 1942016
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8194951004446517e-05,
      "loss": 3.8713,
      "step": 1942528
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8186565056935997e-05,
      "loss": 3.8627,
      "step": 1943040
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8178179109425477e-05,
      "loss": 3.8779,
      "step": 1943552
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8169793161914957e-05,
      "loss": 3.8812,
      "step": 1944064
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.816142359320817e-05,
      "loss": 3.8751,
      "step": 1944576
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.815303764569765e-05,
      "loss": 3.8788,
      "step": 1945088
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.814465169818713e-05,
      "loss": 3.8788,
      "step": 1945600
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.813626575067661e-05,
      "loss": 3.8888,
      "step": 1946112
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.812787980316609e-05,
      "loss": 3.8681,
      "step": 1946624
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.81195102344593e-05,
      "loss": 3.877,
      "step": 1947136
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.811112428694878e-05,
      "loss": 3.8723,
      "step": 1947648
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.810273833943826e-05,
      "loss": 3.8737,
      "step": 1948160
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.809435239192774e-05,
      "loss": 3.8774,
      "step": 1948672
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.808596644441722e-05,
      "loss": 3.8803,
      "step": 1949184
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.807759687571043e-05,
      "loss": 3.8836,
      "step": 1949696
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.806921092819991e-05,
      "loss": 3.8903,
      "step": 1950208
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8060824980689394e-05,
      "loss": 3.8735,
      "step": 1950720
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8052439033178874e-05,
      "loss": 3.8707,
      "step": 1951232
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8044069464472083e-05,
      "loss": 3.8717,
      "step": 1951744
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8035683516961563e-05,
      "loss": 3.8815,
      "step": 1952256
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8027297569451043e-05,
      "loss": 3.8738,
      "step": 1952768
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.8018911621940523e-05,
      "loss": 3.8785,
      "step": 1953280
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.8010525674430003e-05,
      "loss": 3.8815,
      "step": 1953792
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.8002156105723215e-05,
      "loss": 3.8829,
      "step": 1954304
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7993770158212695e-05,
      "loss": 3.8804,
      "step": 1954816
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.798538421070217e-05,
      "loss": 3.8741,
      "step": 1955328
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.797699826319165e-05,
      "loss": 3.8832,
      "step": 1955840
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7968628694484864e-05,
      "loss": 3.8818,
      "step": 1956352
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7960242746974344e-05,
      "loss": 3.8867,
      "step": 1956864
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7951856799463824e-05,
      "loss": 3.8708,
      "step": 1957376
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7943470851953304e-05,
      "loss": 3.878,
      "step": 1957888
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7935084904442784e-05,
      "loss": 3.8795,
      "step": 1958400
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7926698956932264e-05,
      "loss": 3.8904,
      "step": 1958912
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7918313009421744e-05,
      "loss": 3.8824,
      "step": 1959424
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7909927061911224e-05,
      "loss": 3.8799,
      "step": 1959936
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7901557493204436e-05,
      "loss": 3.877,
      "step": 1960448
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7893187924497645e-05,
      "loss": 3.8817,
      "step": 1960960
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7884801976987125e-05,
      "loss": 3.8619,
      "step": 1961472
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7876416029476605e-05,
      "loss": 3.8726,
      "step": 1961984
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7868030081966085e-05,
      "loss": 3.8652,
      "step": 1962496
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7859644134455565e-05,
      "loss": 3.8821,
      "step": 1963008
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7851274565748778e-05,
      "loss": 3.8733,
      "step": 1963520
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7842888618238258e-05,
      "loss": 3.8757,
      "step": 1964032
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7834502670727737e-05,
      "loss": 3.8772,
      "step": 1964544
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7826116723217217e-05,
      "loss": 3.8799,
      "step": 1965056
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7817730775706697e-05,
      "loss": 3.8805,
      "step": 1965568
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.780936120699991e-05,
      "loss": 3.8776,
      "step": 1966080
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.780097525948939e-05,
      "loss": 3.8815,
      "step": 1966592
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.779258931197887e-05,
      "loss": 3.8883,
      "step": 1967104
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.778420336446835e-05,
      "loss": 3.8813,
      "step": 1967616
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.777585017456529e-05,
      "loss": 3.8788,
      "step": 1968128
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.776746422705477e-05,
      "loss": 3.8722,
      "step": 1968640
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.775907827954425e-05,
      "loss": 3.8783,
      "step": 1969152
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.775069233203373e-05,
      "loss": 3.879,
      "step": 1969664
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.774230638452321e-05,
      "loss": 3.8605,
      "step": 1970176
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.773392043701269e-05,
      "loss": 3.8743,
      "step": 1970688
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.772553448950217e-05,
      "loss": 3.8742,
      "step": 1971200
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.771714854199165e-05,
      "loss": 3.8767,
      "step": 1971712
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7708778973284864e-05,
      "loss": 3.8855,
      "step": 1972224
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7700393025774344e-05,
      "loss": 3.8748,
      "step": 1972736
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7692007078263823e-05,
      "loss": 3.8738,
      "step": 1973248
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.76836211307533e-05,
      "loss": 3.8705,
      "step": 1973760
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.767523518324278e-05,
      "loss": 3.8899,
      "step": 1974272
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7666865614535992e-05,
      "loss": 3.879,
      "step": 1974784
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7658479667025472e-05,
      "loss": 3.8831,
      "step": 1975296
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7650093719514952e-05,
      "loss": 3.8685,
      "step": 1975808
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7641707772004432e-05,
      "loss": 3.8904,
      "step": 1976320
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7633321824493912e-05,
      "loss": 3.8827,
      "step": 1976832
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7624935876983392e-05,
      "loss": 3.8848,
      "step": 1977344
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7616549929472872e-05,
      "loss": 3.8796,
      "step": 1977856
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7608163981962352e-05,
      "loss": 3.8754,
      "step": 1978368
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7599794413255565e-05,
      "loss": 3.8764,
      "step": 1978880
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7591408465745045e-05,
      "loss": 3.8973,
      "step": 1979392
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7583022518234525e-05,
      "loss": 3.869,
      "step": 1979904
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7574636570724005e-05,
      "loss": 3.8799,
      "step": 1980416
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7566267002017214e-05,
      "loss": 3.8771,
      "step": 1980928
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7557881054506694e-05,
      "loss": 3.8676,
      "step": 1981440
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7549495106996174e-05,
      "loss": 3.8739,
      "step": 1981952
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7541109159485657e-05,
      "loss": 3.8786,
      "step": 1982464
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7532723211975133e-05,
      "loss": 3.8851,
      "step": 1982976
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7524353643268346e-05,
      "loss": 3.8786,
      "step": 1983488
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7515967695757826e-05,
      "loss": 3.8786,
      "step": 1984000
    },
    {
      "epoch": 1.03,
      "eval_loss": 3.986476421356201,
      "eval_runtime": 295.6997,
      "eval_samples_per_second": 1290.468,
      "eval_steps_per_second": 40.328,
      "step": 1984320
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 8.154685597708017e+17,
  "trial_name": null,
  "trial_params": null
}