{
  "best_metric": 4.2788496017456055,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/re-irr-sv-agr/lstm/2/checkpoints/checkpoint-305280",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 305280,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8202,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5484,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0528,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9887,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9423,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9024,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.732,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.626,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5233,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4512,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.3812,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.324,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2579,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.1877,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988259673485272e-05,
      "loss": 6.1303,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0634,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0219,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.9809,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9399,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9034,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8747,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8349,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8113,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7679,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979875363855125e-05,
      "loss": 5.7551,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979036769104073e-05,
      "loss": 5.7238,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.978198174353021e-05,
      "loss": 5.7047,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.6762,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6445,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6281,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6031,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 5.5931,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.5668,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.556,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5444,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5233,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.5128,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968980183612942e-05,
      "loss": 5.4687,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96814158886189e-05,
      "loss": 5.4655,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.444,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.431,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.4256,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 5.4178,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.3898,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.3895,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 5.3845,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 5.3711,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3656,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3495,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 5.3285,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 5.33,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.3116,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3131,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.282,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 5.2943,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 5.2674,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 5.2857,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 5.2572,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.2442,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 5.2398,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 5.2354,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 5.2208,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 5.2225,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 5.196,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 5.1905,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 5.2036,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944670763114672e-05,
      "loss": 5.1956,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94383216836362e-05,
      "loss": 5.1782,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942993573612568e-05,
      "loss": 5.1613,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942154978861516e-05,
      "loss": 5.1594,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 5.1609,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 5.1542,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 5.1498,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 5.1459,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 5.1466,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 5.1378,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362880913648985e-05,
      "loss": 5.1114,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354494966138465e-05,
      "loss": 5.1079,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.105,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.0932,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.102,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0841,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 5.1024,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 5.0772,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295809712368554e-05,
      "loss": 5.062,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0656,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0663,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0635,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0498,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0448,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0374,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0354,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0255,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92203525635776e-05,
      "loss": 5.0204,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921196661606708e-05,
      "loss": 5.0195,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920358066855656e-05,
      "loss": 5.0248,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919519472104604e-05,
      "loss": 5.0054,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9993,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9923,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917006963612195e-05,
      "loss": 4.9881,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.9944,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.9798,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.9731,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136542224883596e-05,
      "loss": 4.9661,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128156277373076e-05,
      "loss": 4.9697,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119770329862556e-05,
      "loss": 4.9401,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9568,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9442,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909462886613473e-05,
      "loss": 4.9505,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908624291862421e-05,
      "loss": 4.9377,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9447,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.9302,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9255,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9329,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.9102,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.927,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.9161,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901918809614751e-05,
      "loss": 4.9135,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901080214863699e-05,
      "loss": 4.9007,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900241620112647e-05,
      "loss": 4.9015,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.9044,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8946,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8946,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8845,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896048646357387e-05,
      "loss": 4.886,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.8675,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8699,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8532,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.8594,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918573104825e-05,
      "loss": 4.8463,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891018715731448e-05,
      "loss": 4.8679,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890180120980396e-05,
      "loss": 4.8729,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889341526229344e-05,
      "loss": 4.8578,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888504569358665e-05,
      "loss": 4.8355,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8876659746076134e-05,
      "loss": 4.8457,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8868273798565614e-05,
      "loss": 4.8458,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8859887851055094e-05,
      "loss": 4.8391,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.835,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8343,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8307,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8201,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881799087110995e-05,
      "loss": 4.8209,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.821,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.8095,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879283302857839e-05,
      "loss": 4.8089,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878447983867534e-05,
      "loss": 4.8163,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877609389116482e-05,
      "loss": 4.8046,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.8039,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.8146,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7913,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.780590534210205,
      "eval_runtime": 282.4544,
      "eval_samples_per_second": 1350.982,
      "eval_steps_per_second": 42.219,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7797,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8734180532415946e-05,
      "loss": 4.7866,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8725794584905426e-05,
      "loss": 4.8002,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717408637394906e-05,
      "loss": 4.7879,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709022689884386e-05,
      "loss": 4.7927,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700636742373866e-05,
      "loss": 4.7681,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692250794863346e-05,
      "loss": 4.7842,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683864847352826e-05,
      "loss": 4.7566,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675478899842306e-05,
      "loss": 4.7726,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.759,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7738,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865033743611448e-05,
      "loss": 4.7684,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864196786740769e-05,
      "loss": 4.7567,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86335982987009e-05,
      "loss": 4.7508,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862521235119038e-05,
      "loss": 4.739,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861682640367986e-05,
      "loss": 4.7393,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860844045616934e-05,
      "loss": 4.7491,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860005450865882e-05,
      "loss": 4.7449,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85916685611483e-05,
      "loss": 4.7371,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.755,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.7461,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566527097420475e-05,
      "loss": 4.7382,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558141149909955e-05,
      "loss": 4.7267,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549755202399435e-05,
      "loss": 4.7464,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541369254888915e-05,
      "loss": 4.7218,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532983307378395e-05,
      "loss": 4.7312,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524597359867875e-05,
      "loss": 4.7283,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851621141235735e-05,
      "loss": 4.7209,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850782546484683e-05,
      "loss": 4.7103,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8499455896140044e-05,
      "loss": 4.7113,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8491069948629524e-05,
      "loss": 4.7122,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848270037992273e-05,
      "loss": 4.7097,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847431443241221e-05,
      "loss": 4.71,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846592848490169e-05,
      "loss": 4.7156,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845754253739117e-05,
      "loss": 4.705,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844915658988065e-05,
      "loss": 4.704,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844077064237013e-05,
      "loss": 4.6835,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843238469485961e-05,
      "loss": 4.6906,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842399874734909e-05,
      "loss": 4.6804,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841561279983857e-05,
      "loss": 4.6773,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840722685232805e-05,
      "loss": 4.6885,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839884090481753e-05,
      "loss": 4.6889,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839047133611074e-05,
      "loss": 4.6728,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838210176740395e-05,
      "loss": 4.68,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837371581989343e-05,
      "loss": 4.6779,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836532987238292e-05,
      "loss": 4.6805,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83569439248724e-05,
      "loss": 4.6829,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834855797736188e-05,
      "loss": 4.6733,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.6626,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833178608234084e-05,
      "loss": 4.6686,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832340013483032e-05,
      "loss": 4.6564,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83150141873198e-05,
      "loss": 4.6658,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830662823980928e-05,
      "loss": 4.6535,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829824229229876e-05,
      "loss": 4.6639,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289872723591966e-05,
      "loss": 4.6424,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281486776081446e-05,
      "loss": 4.6662,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273117207374655e-05,
      "loss": 4.6492,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264731259864135e-05,
      "loss": 4.6446,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256345312353615e-05,
      "loss": 4.6516,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247959364843095e-05,
      "loss": 4.6417,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823957341733258e-05,
      "loss": 4.6336,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823118746982206e-05,
      "loss": 4.6505,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822281790111527e-05,
      "loss": 4.6192,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821443195360475e-05,
      "loss": 4.626,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820604600609423e-05,
      "loss": 4.6375,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819766005858371e-05,
      "loss": 4.6374,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189274111073184e-05,
      "loss": 4.6271,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180888163562664e-05,
      "loss": 4.6189,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817251859485588e-05,
      "loss": 4.618,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816413264734535e-05,
      "loss": 4.6244,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815576307863857e-05,
      "loss": 4.6292,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814737713112805e-05,
      "loss": 4.6237,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138991183617535e-05,
      "loss": 4.6251,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813060523610701e-05,
      "loss": 4.6295,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812221928859649e-05,
      "loss": 4.6262,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811383334108597e-05,
      "loss": 4.6079,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810544739357545e-05,
      "loss": 4.6072,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809706144606493e-05,
      "loss": 4.6145,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808869187735814e-05,
      "loss": 4.6,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808030592984762e-05,
      "loss": 4.6113,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80719199823371e-05,
      "loss": 4.5977,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806353403482658e-05,
      "loss": 4.6223,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805514808731606e-05,
      "loss": 4.6043,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804677851860927e-05,
      "loss": 4.5888,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803839257109875e-05,
      "loss": 4.598,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803000662358823e-05,
      "loss": 4.5993,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802162067607771e-05,
      "loss": 4.6047,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801323472856719e-05,
      "loss": 4.5938,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80048651598604e-05,
      "loss": 4.5937,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799647921234988e-05,
      "loss": 4.5965,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798809326483936e-05,
      "loss": 4.585,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797970731732884e-05,
      "loss": 4.5846,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797132136981832e-05,
      "loss": 4.5832,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79629354223078e-05,
      "loss": 4.5913,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795456585360101e-05,
      "loss": 4.5881,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794617990609049e-05,
      "loss": 4.5855,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793779395857997e-05,
      "loss": 4.5725,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792940801106946e-05,
      "loss": 4.5729,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792102206355894e-05,
      "loss": 4.5776,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791265249485215e-05,
      "loss": 4.5758,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790426654734163e-05,
      "loss": 4.5728,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895880599831107e-05,
      "loss": 4.5676,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887494652320587e-05,
      "loss": 4.5619,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879108704810067e-05,
      "loss": 4.5731,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870739136103276e-05,
      "loss": 4.541,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862353188592756e-05,
      "loss": 4.5658,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853967241082236e-05,
      "loss": 4.5568,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845581293571715e-05,
      "loss": 4.5639,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837195346061195e-05,
      "loss": 4.5512,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828809398550675e-05,
      "loss": 4.564,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782043982984389e-05,
      "loss": 4.5522,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781205388233337e-05,
      "loss": 4.55,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803667934822844e-05,
      "loss": 4.561,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795281987312324e-05,
      "loss": 4.5442,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778691241860554e-05,
      "loss": 4.5592,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777852647109502e-05,
      "loss": 4.5559,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777014052358449e-05,
      "loss": 4.5514,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776175457607397e-05,
      "loss": 4.5425,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775336862856345e-05,
      "loss": 4.5456,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774499905985666e-05,
      "loss": 4.551,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773661311234615e-05,
      "loss": 4.5362,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772822716483563e-05,
      "loss": 4.5545,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771984121732511e-05,
      "loss": 4.5361,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771145526981459e-05,
      "loss": 4.5416,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770306932230407e-05,
      "loss": 4.5353,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769468337479355e-05,
      "loss": 4.5333,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768629742728303e-05,
      "loss": 4.5149,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767794423737997e-05,
      "loss": 4.5284,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766955828986945e-05,
      "loss": 4.5167,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766117234235893e-05,
      "loss": 4.5398,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765278639484841e-05,
      "loss": 4.5509,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764440044733789e-05,
      "loss": 4.5334,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763601449982737e-05,
      "loss": 4.5174,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762762855231685e-05,
      "loss": 4.5294,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761924260480633e-05,
      "loss": 4.5235,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761087303609954e-05,
      "loss": 4.5267,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760248708858902e-05,
      "loss": 4.5227,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759411751988223e-05,
      "loss": 4.522,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758573157237171e-05,
      "loss": 4.5181,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757734562486119e-05,
      "loss": 4.5183,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756895967735067e-05,
      "loss": 4.5179,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756057372984015e-05,
      "loss": 4.5224,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755218778232963e-05,
      "loss": 4.5142,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754381821362284e-05,
      "loss": 4.5073,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753543226611232e-05,
      "loss": 4.522,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75270463186018e-05,
      "loss": 4.5149,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751866037109128e-05,
      "loss": 4.5058,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751027442358077e-05,
      "loss": 4.5237,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750188847607025e-05,
      "loss": 4.504,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.497745037078857,
      "eval_runtime": 283.3858,
      "eval_samples_per_second": 1346.542,
      "eval_steps_per_second": 42.08,
      "step": 152640
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.749350252855973e-05,
      "loss": 4.4937,
      "step": 153088
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.748511658104921e-05,
      "loss": 4.4989,
      "step": 153600
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7476747012342416e-05,
      "loss": 4.5153,
      "step": 154112
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7468361064831896e-05,
      "loss": 4.5065,
      "step": 154624
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7459975117321376e-05,
      "loss": 4.5109,
      "step": 155136
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.745158916981085e-05,
      "loss": 4.4893,
      "step": 155648
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7443219601104065e-05,
      "loss": 4.5082,
      "step": 156160
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7434850032397274e-05,
      "loss": 4.4799,
      "step": 156672
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7426464084886754e-05,
      "loss": 4.5036,
      "step": 157184
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7418078137376234e-05,
      "loss": 4.4864,
      "step": 157696
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.740969218986572e-05,
      "loss": 4.5039,
      "step": 158208
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.74013062423552e-05,
      "loss": 4.5015,
      "step": 158720
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.739295305245214e-05,
      "loss": 4.4891,
      "step": 159232
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.738456710494162e-05,
      "loss": 4.4856,
      "step": 159744
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.73761811574311e-05,
      "loss": 4.4793,
      "step": 160256
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.736779520992058e-05,
      "loss": 4.4794,
      "step": 160768
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.735940926241006e-05,
      "loss": 4.4848,
      "step": 161280
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.735102331489954e-05,
      "loss": 4.486,
      "step": 161792
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.734263736738902e-05,
      "loss": 4.4857,
      "step": 162304
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.73342514198785e-05,
      "loss": 4.4974,
      "step": 162816
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.732586547236797e-05,
      "loss": 4.4937,
      "step": 163328
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.731747952485746e-05,
      "loss": 4.4879,
      "step": 163840
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.730909357734694e-05,
      "loss": 4.4778,
      "step": 164352
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.730070762983642e-05,
      "loss": 4.4909,
      "step": 164864
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.729233806112963e-05,
      "loss": 4.4761,
      "step": 165376
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7283968492422844e-05,
      "loss": 4.4833,
      "step": 165888
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7275582544912323e-05,
      "loss": 4.4812,
      "step": 166400
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.72671965974018e-05,
      "loss": 4.4772,
      "step": 166912
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.7258810649891277e-05,
      "loss": 4.4678,
      "step": 167424
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7250424702380757e-05,
      "loss": 4.4693,
      "step": 167936
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7242038754870237e-05,
      "loss": 4.4715,
      "step": 168448
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7233669186163446e-05,
      "loss": 4.4747,
      "step": 168960
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7225283238652926e-05,
      "loss": 4.4771,
      "step": 169472
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.721689729114241e-05,
      "loss": 4.4728,
      "step": 169984
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.720852772243563e-05,
      "loss": 4.4694,
      "step": 170496
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.72001417749251e-05,
      "loss": 4.4699,
      "step": 171008
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.719175582741458e-05,
      "loss": 4.4514,
      "step": 171520
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.718336987990406e-05,
      "loss": 4.4627,
      "step": 172032
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.717498393239354e-05,
      "loss": 4.4524,
      "step": 172544
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.716661436368675e-05,
      "loss": 4.4528,
      "step": 173056
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.715822841617623e-05,
      "loss": 4.4585,
      "step": 173568
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.714984246866571e-05,
      "loss": 4.466,
      "step": 174080
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.714145652115519e-05,
      "loss": 4.4517,
      "step": 174592
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.713307057364467e-05,
      "loss": 4.4543,
      "step": 175104
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.712468462613415e-05,
      "loss": 4.4626,
      "step": 175616
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.711629867862363e-05,
      "loss": 4.454,
      "step": 176128
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7107929109916846e-05,
      "loss": 4.4667,
      "step": 176640
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7099543162406326e-05,
      "loss": 4.4512,
      "step": 177152
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7091157214895806e-05,
      "loss": 4.4427,
      "step": 177664
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7082771267385286e-05,
      "loss": 4.454,
      "step": 178176
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7074385319874766e-05,
      "loss": 4.4441,
      "step": 178688
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7065999372364246e-05,
      "loss": 4.4531,
      "step": 179200
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7057613424853726e-05,
      "loss": 4.4413,
      "step": 179712
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7049227477343206e-05,
      "loss": 4.4471,
      "step": 180224
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7040857908636415e-05,
      "loss": 4.436,
      "step": 180736
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7032471961125895e-05,
      "loss": 4.4506,
      "step": 181248
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7024086013615375e-05,
      "loss": 4.4397,
      "step": 181760
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7015700066104855e-05,
      "loss": 4.4392,
      "step": 182272
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.7007330497398064e-05,
      "loss": 4.4455,
      "step": 182784
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.6998944549887544e-05,
      "loss": 4.4334,
      "step": 183296
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.699055860237703e-05,
      "loss": 4.4284,
      "step": 183808
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.698217265486651e-05,
      "loss": 4.4508,
      "step": 184320
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.697380308615972e-05,
      "loss": 4.4174,
      "step": 184832
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.69654171386492e-05,
      "loss": 4.4214,
      "step": 185344
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.695703119113868e-05,
      "loss": 4.4303,
      "step": 185856
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.694864524362816e-05,
      "loss": 4.4419,
      "step": 186368
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.694025929611763e-05,
      "loss": 4.4278,
      "step": 186880
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.693187334860711e-05,
      "loss": 4.4216,
      "step": 187392
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.692348740109659e-05,
      "loss": 4.4224,
      "step": 187904
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.691511783238981e-05,
      "loss": 4.4258,
      "step": 188416
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.690673188487928e-05,
      "loss": 4.4339,
      "step": 188928
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.689834593736877e-05,
      "loss": 4.4302,
      "step": 189440
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.688995998985825e-05,
      "loss": 4.429,
      "step": 189952
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.688157404234773e-05,
      "loss": 4.4337,
      "step": 190464
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.687320447364094e-05,
      "loss": 4.4356,
      "step": 190976
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.686481852613042e-05,
      "loss": 4.4165,
      "step": 191488
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.68564325786199e-05,
      "loss": 4.4163,
      "step": 192000
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.684804663110938e-05,
      "loss": 4.4266,
      "step": 192512
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.683966068359886e-05,
      "loss": 4.4093,
      "step": 193024
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.683127473608834e-05,
      "loss": 4.4248,
      "step": 193536
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.6822905167381546e-05,
      "loss": 4.4113,
      "step": 194048
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.6814519219871026e-05,
      "loss": 4.432,
      "step": 194560
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.6806133272360506e-05,
      "loss": 4.4198,
      "step": 195072
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.6797747324849986e-05,
      "loss": 4.4058,
      "step": 195584
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.6789361377339466e-05,
      "loss": 4.4094,
      "step": 196096
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.678097542982895e-05,
      "loss": 4.4168,
      "step": 196608
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.677258948231843e-05,
      "loss": 4.4254,
      "step": 197120
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.676420353480791e-05,
      "loss": 4.4111,
      "step": 197632
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.675583396610112e-05,
      "loss": 4.4151,
      "step": 198144
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.67474480185906e-05,
      "loss": 4.4066,
      "step": 198656
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.673906207108008e-05,
      "loss": 4.4155,
      "step": 199168
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.673067612356956e-05,
      "loss": 4.4066,
      "step": 199680
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.672229017605904e-05,
      "loss": 4.403,
      "step": 200192
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.671390422854852e-05,
      "loss": 4.4145,
      "step": 200704
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.670553465984173e-05,
      "loss": 4.4124,
      "step": 201216
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.669714871233121e-05,
      "loss": 4.4114,
      "step": 201728
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.668876276482069e-05,
      "loss": 4.397,
      "step": 202240
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.668037681731017e-05,
      "loss": 4.4006,
      "step": 202752
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.667199086979965e-05,
      "loss": 4.4029,
      "step": 203264
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.666360492228913e-05,
      "loss": 4.4032,
      "step": 203776
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6655235353582346e-05,
      "loss": 4.403,
      "step": 204288
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.664684940607182e-05,
      "loss": 4.3936,
      "step": 204800
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.66384634585613e-05,
      "loss": 4.3938,
      "step": 205312
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.663007751105078e-05,
      "loss": 4.4027,
      "step": 205824
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.662169156354026e-05,
      "loss": 4.3747,
      "step": 206336
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.661330561602974e-05,
      "loss": 4.3935,
      "step": 206848
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.660493604732295e-05,
      "loss": 4.3854,
      "step": 207360
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.659655009981243e-05,
      "loss": 4.3997,
      "step": 207872
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.658816415230191e-05,
      "loss": 4.3887,
      "step": 208384
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.657977820479139e-05,
      "loss": 4.3962,
      "step": 208896
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.657139225728087e-05,
      "loss": 4.3898,
      "step": 209408
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6563006309770355e-05,
      "loss": 4.3851,
      "step": 209920
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6554620362259835e-05,
      "loss": 4.3973,
      "step": 210432
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6546250793553044e-05,
      "loss": 4.386,
      "step": 210944
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6537864846042524e-05,
      "loss": 4.3954,
      "step": 211456
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6529478898532004e-05,
      "loss": 4.3925,
      "step": 211968
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.652110932982521e-05,
      "loss": 4.3919,
      "step": 212480
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.651272338231469e-05,
      "loss": 4.3831,
      "step": 212992
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.650433743480417e-05,
      "loss": 4.3796,
      "step": 213504
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.649595148729365e-05,
      "loss": 4.3948,
      "step": 214016
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.648756553978313e-05,
      "loss": 4.3772,
      "step": 214528
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.647917959227261e-05,
      "loss": 4.3946,
      "step": 215040
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.647079364476209e-05,
      "loss": 4.3787,
      "step": 215552
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.646240769725157e-05,
      "loss": 4.3835,
      "step": 216064
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.645403812854479e-05,
      "loss": 4.38,
      "step": 216576
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.644565218103427e-05,
      "loss": 4.3778,
      "step": 217088
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.643726623352375e-05,
      "loss": 4.3607,
      "step": 217600
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.642888028601323e-05,
      "loss": 4.3779,
      "step": 218112
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.642051071730644e-05,
      "loss": 4.3655,
      "step": 218624
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.641212476979592e-05,
      "loss": 4.3798,
      "step": 219136
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.64037388222854e-05,
      "loss": 4.3927,
      "step": 219648
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.639535287477488e-05,
      "loss": 4.3891,
      "step": 220160
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.638696692726436e-05,
      "loss": 4.3567,
      "step": 220672
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6378597358557566e-05,
      "loss": 4.3821,
      "step": 221184
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6370211411047046e-05,
      "loss": 4.3757,
      "step": 221696
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6361825463536526e-05,
      "loss": 4.3753,
      "step": 222208
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6353439516026006e-05,
      "loss": 4.3666,
      "step": 222720
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6345053568515486e-05,
      "loss": 4.3765,
      "step": 223232
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6336667621004966e-05,
      "loss": 4.3706,
      "step": 223744
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6328281673494446e-05,
      "loss": 4.3682,
      "step": 224256
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6319912104787655e-05,
      "loss": 4.3717,
      "step": 224768
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6311526157277135e-05,
      "loss": 4.3739,
      "step": 225280
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6303140209766615e-05,
      "loss": 4.3692,
      "step": 225792
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6294754262256095e-05,
      "loss": 4.3661,
      "step": 226304
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6286368314745575e-05,
      "loss": 4.371,
      "step": 226816
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6277982367235055e-05,
      "loss": 4.3722,
      "step": 227328
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6269596419724535e-05,
      "loss": 4.3629,
      "step": 227840
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.6261226851017744e-05,
      "loss": 4.3769,
      "step": 228352
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.625284090350723e-05,
      "loss": 4.3624,
      "step": 228864
    },
    {
      "epoch": 2.03,
      "eval_loss": 4.36259651184082,
      "eval_runtime": 281.8171,
      "eval_samples_per_second": 1354.038,
      "eval_steps_per_second": 42.315,
      "step": 228960
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.624445495599671e-05,
      "loss": 4.3523,
      "step": 229376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.623606900848619e-05,
      "loss": 4.3554,
      "step": 229888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.622768306097567e-05,
      "loss": 4.3738,
      "step": 230400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.621929711346515e-05,
      "loss": 4.3614,
      "step": 230912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.621091116595463e-05,
      "loss": 4.3739,
      "step": 231424
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.620252521844411e-05,
      "loss": 4.3485,
      "step": 231936
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.619413927093359e-05,
      "loss": 4.3677,
      "step": 232448
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.618575332342307e-05,
      "loss": 4.3376,
      "step": 232960
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.6177367375912543e-05,
      "loss": 4.3689,
      "step": 233472
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.6168981428402023e-05,
      "loss": 4.3499,
      "step": 233984
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.61605954808915e-05,
      "loss": 4.3641,
      "step": 234496
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.615220953338098e-05,
      "loss": 4.363,
      "step": 235008
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.61438399646742e-05,
      "loss": 4.3491,
      "step": 235520
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.613545401716368e-05,
      "loss": 4.3518,
      "step": 236032
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.612706806965316e-05,
      "loss": 4.3458,
      "step": 236544
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.611868212214264e-05,
      "loss": 4.3414,
      "step": 237056
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.611029617463212e-05,
      "loss": 4.352,
      "step": 237568
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.61019102271216e-05,
      "loss": 4.3498,
      "step": 238080
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.609352427961108e-05,
      "loss": 4.3521,
      "step": 238592
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.608513833210056e-05,
      "loss": 4.3614,
      "step": 239104
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.607675238459004e-05,
      "loss": 4.3643,
      "step": 239616
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.606836643707952e-05,
      "loss": 4.3536,
      "step": 240128
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.6059980489569e-05,
      "loss": 4.3458,
      "step": 240640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.605159454205848e-05,
      "loss": 4.359,
      "step": 241152
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.604322497335169e-05,
      "loss": 4.3428,
      "step": 241664
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.60348554046449e-05,
      "loss": 4.3568,
      "step": 242176
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.602648583593811e-05,
      "loss": 4.351,
      "step": 242688
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.601809988842759e-05,
      "loss": 4.3422,
      "step": 243200
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.600971394091707e-05,
      "loss": 4.3384,
      "step": 243712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.600132799340655e-05,
      "loss": 4.3436,
      "step": 244224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.599294204589603e-05,
      "loss": 4.3393,
      "step": 244736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.598455609838551e-05,
      "loss": 4.3479,
      "step": 245248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.597617015087499e-05,
      "loss": 4.3482,
      "step": 245760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.596778420336447e-05,
      "loss": 4.3443,
      "step": 246272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.595939825585395e-05,
      "loss": 4.3463,
      "step": 246784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.595101230834343e-05,
      "loss": 4.3437,
      "step": 247296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.594264273963664e-05,
      "loss": 4.3266,
      "step": 247808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.593425679212612e-05,
      "loss": 4.3339,
      "step": 248320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.59258708446156e-05,
      "loss": 4.3267,
      "step": 248832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.591748489710508e-05,
      "loss": 4.3326,
      "step": 249344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.590909894959456e-05,
      "loss": 4.3329,
      "step": 249856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.590071300208404e-05,
      "loss": 4.3378,
      "step": 250368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5892359812180986e-05,
      "loss": 4.3337,
      "step": 250880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5883973864670466e-05,
      "loss": 4.3265,
      "step": 251392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5875587917159946e-05,
      "loss": 4.3394,
      "step": 251904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5867201969649426e-05,
      "loss": 4.3375,
      "step": 252416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5858816022138906e-05,
      "loss": 4.3404,
      "step": 252928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.585043007462838e-05,
      "loss": 4.3315,
      "step": 253440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.584204412711786e-05,
      "loss": 4.3203,
      "step": 253952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5833674558411075e-05,
      "loss": 4.336,
      "step": 254464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5825288610900555e-05,
      "loss": 4.323,
      "step": 254976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5816902663390035e-05,
      "loss": 4.3331,
      "step": 255488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5808516715879515e-05,
      "loss": 4.3224,
      "step": 256000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5800130768368995e-05,
      "loss": 4.322,
      "step": 256512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5791744820858475e-05,
      "loss": 4.3222,
      "step": 257024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5783358873347955e-05,
      "loss": 4.3276,
      "step": 257536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5774972925837435e-05,
      "loss": 4.3248,
      "step": 258048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5766586978326915e-05,
      "loss": 4.3249,
      "step": 258560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5758201030816395e-05,
      "loss": 4.3239,
      "step": 259072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5749831462109604e-05,
      "loss": 4.3161,
      "step": 259584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5741445514599084e-05,
      "loss": 4.3129,
      "step": 260096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5733059567088564e-05,
      "loss": 4.3364,
      "step": 260608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5724673619578044e-05,
      "loss": 4.3026,
      "step": 261120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5716287672067524e-05,
      "loss": 4.3051,
      "step": 261632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5707901724557004e-05,
      "loss": 4.311,
      "step": 262144
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5699515777046484e-05,
      "loss": 4.3285,
      "step": 262656
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5691129829535963e-05,
      "loss": 4.3135,
      "step": 263168
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.568276026082918e-05,
      "loss": 4.31,
      "step": 263680
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.567437431331866e-05,
      "loss": 4.3038,
      "step": 264192
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.566598836580814e-05,
      "loss": 4.3108,
      "step": 264704
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.565760241829762e-05,
      "loss": 4.3221,
      "step": 265216
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.564923284959083e-05,
      "loss": 4.3159,
      "step": 265728
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.564084690208031e-05,
      "loss": 4.3144,
      "step": 266240
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.563246095456979e-05,
      "loss": 4.3249,
      "step": 266752
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5624091385863e-05,
      "loss": 4.3232,
      "step": 267264
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.561570543835248e-05,
      "loss": 4.3053,
      "step": 267776
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.560731949084196e-05,
      "loss": 4.3058,
      "step": 268288
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.559893354333144e-05,
      "loss": 4.3199,
      "step": 268800
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.559054759582092e-05,
      "loss": 4.2946,
      "step": 269312
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.55821616483104e-05,
      "loss": 4.3137,
      "step": 269824
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.557377570079988e-05,
      "loss": 4.3014,
      "step": 270336
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.556540613209309e-05,
      "loss": 4.3239,
      "step": 270848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5557020184582566e-05,
      "loss": 4.3087,
      "step": 271360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5548634237072046e-05,
      "loss": 4.297,
      "step": 271872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5540248289561526e-05,
      "loss": 4.297,
      "step": 272384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5531862342051006e-05,
      "loss": 4.311,
      "step": 272896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5523476394540486e-05,
      "loss": 4.3169,
      "step": 273408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5515106825833695e-05,
      "loss": 4.3058,
      "step": 273920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5506720878323175e-05,
      "loss": 4.3044,
      "step": 274432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5498334930812655e-05,
      "loss": 4.2982,
      "step": 274944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.548994898330214e-05,
      "loss": 4.3082,
      "step": 275456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.548156303579162e-05,
      "loss": 4.2984,
      "step": 275968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.54731770882811e-05,
      "loss": 4.298,
      "step": 276480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.546479114077058e-05,
      "loss": 4.3085,
      "step": 276992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.545640519326006e-05,
      "loss": 4.3071,
      "step": 277504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.544801924574954e-05,
      "loss": 4.3035,
      "step": 278016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.543964967704275e-05,
      "loss": 4.2908,
      "step": 278528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.543126372953223e-05,
      "loss": 4.2998,
      "step": 279040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.542287778202171e-05,
      "loss": 4.2959,
      "step": 279552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.541449183451119e-05,
      "loss": 4.3019,
      "step": 280064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.540610588700067e-05,
      "loss": 4.2984,
      "step": 280576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.539775269709761e-05,
      "loss": 4.2895,
      "step": 281088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5389366749587095e-05,
      "loss": 4.2926,
      "step": 281600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5380980802076575e-05,
      "loss": 4.2985,
      "step": 282112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5372594854566055e-05,
      "loss": 4.2734,
      "step": 282624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5364208907055535e-05,
      "loss": 4.2933,
      "step": 283136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5355839338348744e-05,
      "loss": 4.28,
      "step": 283648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5347453390838224e-05,
      "loss": 4.2975,
      "step": 284160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5339067443327704e-05,
      "loss": 4.288,
      "step": 284672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5330681495817184e-05,
      "loss": 4.2913,
      "step": 285184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5322295548306664e-05,
      "loss": 4.2864,
      "step": 285696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5313909600796144e-05,
      "loss": 4.2876,
      "step": 286208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5305523653285624e-05,
      "loss": 4.2979,
      "step": 286720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5297137705775104e-05,
      "loss": 4.2815,
      "step": 287232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.528878451587205e-05,
      "loss": 4.2949,
      "step": 287744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.528039856836153e-05,
      "loss": 4.2935,
      "step": 288256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.527201262085101e-05,
      "loss": 4.2891,
      "step": 288768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.526362667334049e-05,
      "loss": 4.292,
      "step": 289280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.525524072582997e-05,
      "loss": 4.277,
      "step": 289792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.524685477831945e-05,
      "loss": 4.297,
      "step": 290304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.523846883080893e-05,
      "loss": 4.2828,
      "step": 290816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.52300828832984e-05,
      "loss": 4.2912,
      "step": 291328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.522171331459162e-05,
      "loss": 4.2811,
      "step": 291840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.52133273670811e-05,
      "loss": 4.2863,
      "step": 292352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5204974177178036e-05,
      "loss": 4.2831,
      "step": 292864
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5196588229667516e-05,
      "loss": 4.278,
      "step": 293376
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5188202282157e-05,
      "loss": 4.267,
      "step": 293888
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.517981633464648e-05,
      "loss": 4.2776,
      "step": 294400
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.517143038713596e-05,
      "loss": 4.2729,
      "step": 294912
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.516304443962544e-05,
      "loss": 4.2836,
      "step": 295424
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.515465849211492e-05,
      "loss": 4.295,
      "step": 295936
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.51462725446044e-05,
      "loss": 4.2923,
      "step": 296448
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5137886597093876e-05,
      "loss": 4.2642,
      "step": 296960
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5129500649583356e-05,
      "loss": 4.2892,
      "step": 297472
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5121114702072836e-05,
      "loss": 4.2801,
      "step": 297984
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5112728754562316e-05,
      "loss": 4.2813,
      "step": 298496
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5104359185855525e-05,
      "loss": 4.2735,
      "step": 299008
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5095973238345005e-05,
      "loss": 4.2833,
      "step": 299520
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5087587290834485e-05,
      "loss": 4.2729,
      "step": 300032
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.507920134332397e-05,
      "loss": 4.281,
      "step": 300544
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.507083177461718e-05,
      "loss": 4.2749,
      "step": 301056
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5062462205910396e-05,
      "loss": 4.278,
      "step": 301568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5054076258399876e-05,
      "loss": 4.2787,
      "step": 302080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.504569031088935e-05,
      "loss": 4.272,
      "step": 302592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.503730436337883e-05,
      "loss": 4.2771,
      "step": 303104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.502891841586831e-05,
      "loss": 4.2835,
      "step": 303616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.502054884716152e-05,
      "loss": 4.2678,
      "step": 304128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.5012162899651e-05,
      "loss": 4.2813,
      "step": 304640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.500377695214048e-05,
      "loss": 4.2755,
      "step": 305152
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.2788496017456055,
      "eval_runtime": 304.6964,
      "eval_samples_per_second": 1252.365,
      "eval_steps_per_second": 39.137,
      "step": 305280
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.2223819719679363e+17,
  "trial_name": null,
  "trial_params": null
}