{
  "best_metric": 4.346052169799805,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-sim-ques/lstm/2/checkpoints/checkpoint-228960",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 228960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8183,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5489,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0575,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9912,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9451,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9054,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7488,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6422,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5422,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.463,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.394,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3437,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2742,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.2007,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988259673485272e-05,
      "loss": 6.1518,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.98742107873422e-05,
      "loss": 6.0921,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986582483983168e-05,
      "loss": 6.0583,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985743889232116e-05,
      "loss": 6.003,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984905294481064e-05,
      "loss": 5.9592,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984066699730012e-05,
      "loss": 5.9277,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8867,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8583,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8315,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.8007,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7722,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7465,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 5.7163,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 5.6883,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6622,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6488,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6247,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740117521192533e-05,
      "loss": 5.6034,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731731573682013e-05,
      "loss": 5.5806,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 5.561,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 5.5533,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 5.5428,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969820416244367e-05,
      "loss": 5.5181,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 5.4789,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 5.4864,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 5.45,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 5.4447,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.4349,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 5.4261,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.4039,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 5.3995,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 5.3925,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 5.3801,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3701,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9597605549924894e-05,
      "loss": 5.3509,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 5.3372,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 5.3406,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.3213,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3104,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.2914,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.296,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2761,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 5.2905,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 5.262,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.2589,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 5.2526,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 5.2303,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 5.2338,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 5.2184,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.2085,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.2025,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 5.2191,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 5.1953,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 5.1825,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 5.1629,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 5.1616,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941318021990837e-05,
      "loss": 5.1757,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404794272397856e-05,
      "loss": 5.1713,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1457,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1612,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1531,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1385,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.1181,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354494966138465e-05,
      "loss": 5.123,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346109018627945e-05,
      "loss": 5.1065,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.1038,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 5.1046,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 5.0941,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 5.0968,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 5.0916,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295809712368554e-05,
      "loss": 5.0698,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0726,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0717,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.065,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0531,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.0506,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9245494027305433e-05,
      "loss": 5.0492,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9237108079794913e-05,
      "loss": 5.036,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922872213228439e-05,
      "loss": 5.0395,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92203525635776e-05,
      "loss": 5.0249,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921196661606708e-05,
      "loss": 5.0264,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920358066855656e-05,
      "loss": 5.023,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919519472104604e-05,
      "loss": 5.0142,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9944,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9935,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.993,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 5.0033,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915328136229718e-05,
      "loss": 4.99,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.9787,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9698,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.9755,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9503,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9518,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9103014813645245e-05,
      "loss": 4.9484,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909462886613473e-05,
      "loss": 4.9505,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908624291862421e-05,
      "loss": 4.9401,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907785697111369e-05,
      "loss": 4.9479,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906947102360317e-05,
      "loss": 4.926,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906108507609265e-05,
      "loss": 4.9473,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905269912858213e-05,
      "loss": 4.9266,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904431318107161e-05,
      "loss": 4.9303,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903592723356109e-05,
      "loss": 4.9301,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.9286,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.9137,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901078576983326e-05,
      "loss": 4.904,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900239982232274e-05,
      "loss": 4.9116,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.9022,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.9033,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8983,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.886,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8983,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.8775,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.873,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.8678,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926975431139254e-05,
      "loss": 4.8738,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918589483628734e-05,
      "loss": 4.8581,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910203536118214e-05,
      "loss": 4.8763,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890181758860769e-05,
      "loss": 4.867,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.8559,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.8472,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766925036836e-05,
      "loss": 4.8502,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.8479,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.8503,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885153466115204e-05,
      "loss": 4.8473,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884314871364151e-05,
      "loss": 4.8448,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883476276613099e-05,
      "loss": 4.8353,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882637681862047e-05,
      "loss": 4.8306,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881799087110995e-05,
      "loss": 4.8197,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.8183,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.8236,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.8049,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8185,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.8176,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.8064,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875930561734005e-05,
      "loss": 4.8111,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875091966982953e-05,
      "loss": 4.806,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.7636847496032715,
      "eval_runtime": 287.1664,
      "eval_samples_per_second": 1328.815,
      "eval_steps_per_second": 41.526,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874253372231901e-05,
      "loss": 4.7863,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873414777480849e-05,
      "loss": 4.7912,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872576182729797e-05,
      "loss": 4.8017,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7838,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709022689884386e-05,
      "loss": 4.7911,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700653121177595e-05,
      "loss": 4.7784,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692267173667075e-05,
      "loss": 4.7806,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683881226156555e-05,
      "loss": 4.7683,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867549527864604e-05,
      "loss": 4.7765,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.764,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658723383625e-05,
      "loss": 4.7869,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865033743611448e-05,
      "loss": 4.7759,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864195148860396e-05,
      "loss": 4.7565,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863356554109344e-05,
      "loss": 4.7552,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862517959358292e-05,
      "loss": 4.7508,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86167936460724e-05,
      "loss": 4.7525,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860842407736561e-05,
      "loss": 4.7462,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860003812985509e-05,
      "loss": 4.7448,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859165218234457e-05,
      "loss": 4.7411,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858326623483405e-05,
      "loss": 4.7515,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857488028732352e-05,
      "loss": 4.7425,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856651071861674e-05,
      "loss": 4.7463,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558124771106226e-05,
      "loss": 4.7417,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85497388235957e-05,
      "loss": 4.7484,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854135287608518e-05,
      "loss": 4.7249,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853296692857466e-05,
      "loss": 4.7254,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852458098106414e-05,
      "loss": 4.7203,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851619503355362e-05,
      "loss": 4.7187,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85078090860431e-05,
      "loss": 4.7188,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849942313853258e-05,
      "loss": 4.7158,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849105356982579e-05,
      "loss": 4.7174,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8482684001119e-05,
      "loss": 4.7135,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847429805360848e-05,
      "loss": 4.7035,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.7163,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8457526158587444e-05,
      "loss": 4.7168,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449140211076924e-05,
      "loss": 4.7033,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440754263566404e-05,
      "loss": 4.6867,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432368316055884e-05,
      "loss": 4.7012,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423982368545363e-05,
      "loss": 4.6791,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415596421034843e-05,
      "loss": 4.6895,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840724323113178e-05,
      "loss": 4.6889,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8398873662425e-05,
      "loss": 4.6926,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839048771491447e-05,
      "loss": 4.6803,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838210176740395e-05,
      "loss": 4.6868,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837371581989343e-05,
      "loss": 4.6885,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836532987238292e-05,
      "loss": 4.682,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83569439248724e-05,
      "loss": 4.6828,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834855797736188e-05,
      "loss": 4.6663,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834017202985136e-05,
      "loss": 4.6679,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833178608234084e-05,
      "loss": 4.6763,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832340013483032e-05,
      "loss": 4.6651,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83150141873198e-05,
      "loss": 4.6669,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8306644618613006e-05,
      "loss": 4.6486,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298258671102486e-05,
      "loss": 4.661,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289872723591966e-05,
      "loss": 4.6485,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281503154885175e-05,
      "loss": 4.6664,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273117207374655e-05,
      "loss": 4.6502,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264731259864135e-05,
      "loss": 4.6555,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256345312353615e-05,
      "loss": 4.656,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8247959364843095e-05,
      "loss": 4.6334,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823957341733258e-05,
      "loss": 4.6454,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823118746982206e-05,
      "loss": 4.6441,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222801522311535e-05,
      "loss": 4.6283,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214415574801015e-05,
      "loss": 4.6323,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206029627290495e-05,
      "loss": 4.6524,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197643679779975e-05,
      "loss": 4.637,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189274111073184e-05,
      "loss": 4.623,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180888163562664e-05,
      "loss": 4.6199,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172502216052144e-05,
      "loss": 4.6232,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164116268541624e-05,
      "loss": 4.6398,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155730321031104e-05,
      "loss": 4.6394,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147344373520584e-05,
      "loss": 4.6222,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138958426010064e-05,
      "loss": 4.6385,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813058885730328e-05,
      "loss": 4.6365,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812220290979276e-05,
      "loss": 4.627,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811381696228224e-05,
      "loss": 4.6114,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810543101477172e-05,
      "loss": 4.6254,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80970450672612e-05,
      "loss": 4.6135,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808865911975068e-05,
      "loss": 4.6097,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808028955104389e-05,
      "loss": 4.6136,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807190360353337e-05,
      "loss": 4.6068,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806351765602285e-05,
      "loss": 4.622,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805513170851233e-05,
      "loss": 4.608,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804674576100181e-05,
      "loss": 4.5992,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803835981349129e-05,
      "loss": 4.6056,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802997386598077e-05,
      "loss": 4.6053,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802158791847025e-05,
      "loss": 4.6042,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801320197095973e-05,
      "loss": 4.6,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8004832402252944e-05,
      "loss": 4.5936,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996446454742424e-05,
      "loss": 4.6036,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79880605072319e-05,
      "loss": 4.5906,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797967455972138e-05,
      "loss": 4.5984,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797130499101459e-05,
      "loss": 4.583,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796291904350407e-05,
      "loss": 4.5969,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7954533095993546e-05,
      "loss": 4.5933,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7946147148483026e-05,
      "loss": 4.5866,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793777757977624e-05,
      "loss": 4.5724,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792940801106946e-05,
      "loss": 4.5779,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792102206355894e-05,
      "loss": 4.5811,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791263611604842e-05,
      "loss": 4.5902,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79042501685379e-05,
      "loss": 4.5828,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789586422102737e-05,
      "loss": 4.5724,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788747827351685e-05,
      "loss": 4.5685,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787909232600633e-05,
      "loss": 4.5772,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870722757299546e-05,
      "loss": 4.551,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786233680978902e-05,
      "loss": 4.564,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78539508622785e-05,
      "loss": 4.5616,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784556491476798e-05,
      "loss": 4.5647,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783717896725746e-05,
      "loss": 4.557,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782879301974694e-05,
      "loss": 4.5645,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820423451040155e-05,
      "loss": 4.5532,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781205388233337e-05,
      "loss": 4.5739,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803667934822844e-05,
      "loss": 4.5579,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795281987312324e-05,
      "loss": 4.5558,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786896039801804e-05,
      "loss": 4.5717,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778510092291284e-05,
      "loss": 4.5686,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7770124144780764e-05,
      "loss": 4.5532,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7761738197270244e-05,
      "loss": 4.5459,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775336862856345e-05,
      "loss": 4.56,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774498268105293e-05,
      "loss": 4.5499,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773659673354241e-05,
      "loss": 4.5496,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772821078603189e-05,
      "loss": 4.557,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771982483852137e-05,
      "loss": 4.5417,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771143889101086e-05,
      "loss": 4.556,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770305294350034e-05,
      "loss": 4.5384,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769466699598982e-05,
      "loss": 4.541,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76862810484793e-05,
      "loss": 4.5307,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767789510096878e-05,
      "loss": 4.5426,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766950915345826e-05,
      "loss": 4.5333,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766112320594773e-05,
      "loss": 4.5483,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765273725843721e-05,
      "loss": 4.5465,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764436768973043e-05,
      "loss": 4.5361,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76359817422199e-05,
      "loss": 4.522,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762759579470938e-05,
      "loss": 4.5371,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761920984719886e-05,
      "loss": 4.532,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761082389968834e-05,
      "loss": 4.5388,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760243795217783e-05,
      "loss": 4.5353,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759406838347104e-05,
      "loss": 4.5326,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758569881476425e-05,
      "loss": 4.5292,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757731286725373e-05,
      "loss": 4.5342,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7568926919743206e-05,
      "loss": 4.516,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7560540972232686e-05,
      "loss": 4.517,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7552155024722166e-05,
      "loss": 4.5246,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543785456015375e-05,
      "loss": 4.5109,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535399508504855e-05,
      "loss": 4.5199,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527013560994335e-05,
      "loss": 4.5305,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518627613483815e-05,
      "loss": 4.5167,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510241665973295e-05,
      "loss": 4.5154,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750185571846278e-05,
      "loss": 4.5179,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.481045722961426,
      "eval_runtime": 332.9874,
      "eval_samples_per_second": 1145.962,
      "eval_steps_per_second": 35.812,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749346977095226e-05,
      "loss": 4.5024,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748508382344174e-05,
      "loss": 4.506,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747669787593122e-05,
      "loss": 4.52,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.74683119284207e-05,
      "loss": 4.5059,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745992598091018e-05,
      "loss": 4.5133,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745154003339966e-05,
      "loss": 4.4983,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744315408588914e-05,
      "loss": 4.5113,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743476813837862e-05,
      "loss": 4.4917,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426382190868095e-05,
      "loss": 4.5049,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7417996243357575e-05,
      "loss": 4.4983,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7409610295847055e-05,
      "loss": 4.5128,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401224348336535e-05,
      "loss": 4.5127,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.739285477962975e-05,
      "loss": 4.4881,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.738446883211923e-05,
      "loss": 4.4937,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737608288460871e-05,
      "loss": 4.4951,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736769693709819e-05,
      "loss": 4.4896,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735931098958767e-05,
      "loss": 4.4896,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735094142088088e-05,
      "loss": 4.4865,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734255547337036e-05,
      "loss": 4.4846,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733416952585984e-05,
      "loss": 4.5061,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732578357834932e-05,
      "loss": 4.4866,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73173976308388e-05,
      "loss": 4.4988,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730902806213201e-05,
      "loss": 4.4938,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730064211462149e-05,
      "loss": 4.4998,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729225616711097e-05,
      "loss": 4.4727,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728387021960045e-05,
      "loss": 4.4842,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727548427208993e-05,
      "loss": 4.4736,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7267098324579415e-05,
      "loss": 4.481,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258712377068895e-05,
      "loss": 4.4741,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250326429558375e-05,
      "loss": 4.476,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7241956860851584e-05,
      "loss": 4.483,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7233570913341064e-05,
      "loss": 4.4766,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7225184965830544e-05,
      "loss": 4.4663,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7216799018320024e-05,
      "loss": 4.4829,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720842944961323e-05,
      "loss": 4.477,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720004350210271e-05,
      "loss": 4.4731,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719167393339592e-05,
      "loss": 4.4601,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71832879858854e-05,
      "loss": 4.4713,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717490203837488e-05,
      "loss": 4.4514,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716651609086437e-05,
      "loss": 4.4664,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715814652215758e-05,
      "loss": 4.4598,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714976057464706e-05,
      "loss": 4.4672,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714137462713654e-05,
      "loss": 4.4614,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713298867962602e-05,
      "loss": 4.4643,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71246027321155e-05,
      "loss": 4.4672,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711621678460498e-05,
      "loss": 4.4632,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710783083709446e-05,
      "loss": 4.4666,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709944488958393e-05,
      "loss": 4.4484,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709105894207341e-05,
      "loss": 4.4481,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082689373366627e-05,
      "loss": 4.4632,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074303425856106e-05,
      "loss": 4.4508,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7065917478345586e-05,
      "loss": 4.4518,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7057531530835066e-05,
      "loss": 4.4383,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704916196212828e-05,
      "loss": 4.4514,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7040776014617755e-05,
      "loss": 4.4367,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703240644591097e-05,
      "loss": 4.4539,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702402049840045e-05,
      "loss": 4.4426,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7015634550889924e-05,
      "loss": 4.4515,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007248603379404e-05,
      "loss": 4.4479,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6998862655868884e-05,
      "loss": 4.4315,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990476708358364e-05,
      "loss": 4.4398,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982090760847844e-05,
      "loss": 4.44,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973704813337324e-05,
      "loss": 4.4274,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696533524463054e-05,
      "loss": 4.4377,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695694929712002e-05,
      "loss": 4.4455,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.69485633496095e-05,
      "loss": 4.4406,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694017740209898e-05,
      "loss": 4.4274,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693179145458846e-05,
      "loss": 4.4208,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.692342188588167e-05,
      "loss": 4.4272,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691503593837115e-05,
      "loss": 4.4427,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690664999086063e-05,
      "loss": 4.4469,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689826404335011e-05,
      "loss": 4.4306,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688987809583959e-05,
      "loss": 4.4431,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688149214832907e-05,
      "loss": 4.4432,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687310620081855e-05,
      "loss": 4.4396,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686472025330803e-05,
      "loss": 4.4153,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856367063404974e-05,
      "loss": 4.436,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6847981115894454e-05,
      "loss": 4.4304,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6839595168383934e-05,
      "loss": 4.4186,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6831209220873414e-05,
      "loss": 4.4308,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682283965216662e-05,
      "loss": 4.4209,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68144537046561e-05,
      "loss": 4.434,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680606775714558e-05,
      "loss": 4.4268,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679768180963506e-05,
      "loss": 4.4144,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678929586212454e-05,
      "loss": 4.4227,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678090991461402e-05,
      "loss": 4.4212,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.67725239671035e-05,
      "loss": 4.4247,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676413801959298e-05,
      "loss": 4.419,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675576845088619e-05,
      "loss": 4.4151,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674738250337568e-05,
      "loss": 4.4223,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673899655586516e-05,
      "loss": 4.4116,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673061060835464e-05,
      "loss": 4.4191,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672224103964785e-05,
      "loss": 4.4084,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671385509213733e-05,
      "loss": 4.4234,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705485523430536e-05,
      "loss": 4.4185,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6697099575920016e-05,
      "loss": 4.4112,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6688713628409496e-05,
      "loss": 4.4026,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6680327680898976e-05,
      "loss": 4.4107,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6671941733388456e-05,
      "loss": 4.4041,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6663555785877936e-05,
      "loss": 4.4221,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655169838367416e-05,
      "loss": 4.4101,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6646783890856896e-05,
      "loss": 4.4037,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6638397943346376e-05,
      "loss": 4.3977,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663002837463959e-05,
      "loss": 4.4062,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6621642427129065e-05,
      "loss": 4.3859,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6613256479618545e-05,
      "loss": 4.3964,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6604870532108025e-05,
      "loss": 4.3962,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6596484584597505e-05,
      "loss": 4.3994,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658813139469445e-05,
      "loss": 4.3888,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657974544718393e-05,
      "loss": 4.403,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657135949967341e-05,
      "loss": 4.3877,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656297355216288e-05,
      "loss": 4.4113,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655458760465237e-05,
      "loss": 4.3956,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654620165714185e-05,
      "loss": 4.3926,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.653781570963133e-05,
      "loss": 4.4121,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652942976212081e-05,
      "loss": 4.4106,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.652106019341402e-05,
      "loss": 4.3946,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65126742459035e-05,
      "loss": 4.3881,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650428829839298e-05,
      "loss": 4.3965,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649590235088246e-05,
      "loss": 4.3928,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648753278217567e-05,
      "loss": 4.3896,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647914683466515e-05,
      "loss": 4.4028,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647076088715463e-05,
      "loss": 4.3825,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646237493964411e-05,
      "loss": 4.4007,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645398899213359e-05,
      "loss": 4.3853,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644560304462307e-05,
      "loss": 4.3885,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6437217097112554e-05,
      "loss": 4.375,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6428831149602034e-05,
      "loss": 4.3923,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642046158089524e-05,
      "loss": 4.3769,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641207563338472e-05,
      "loss": 4.3977,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64036896858742e-05,
      "loss": 4.3908,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639530373836368e-05,
      "loss": 4.3859,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638691779085316e-05,
      "loss": 4.3701,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637854822214637e-05,
      "loss": 4.3877,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637016227463585e-05,
      "loss": 4.3847,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636177632712533e-05,
      "loss": 4.3848,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635340675841854e-05,
      "loss": 4.3823,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634502081090802e-05,
      "loss": 4.3889,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633663486339751e-05,
      "loss": 4.3823,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632824891588699e-05,
      "loss": 4.3889,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631986296837647e-05,
      "loss": 4.3691,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631147702086595e-05,
      "loss": 4.3727,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.630309107335542e-05,
      "loss": 4.3792,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.62947051258449e-05,
      "loss": 4.3657,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628631917833438e-05,
      "loss": 4.3761,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627793323082386e-05,
      "loss": 4.3862,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626954728331334e-05,
      "loss": 4.3708,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626116133580282e-05,
      "loss": 4.3713,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625279176709603e-05,
      "loss": 4.3786,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.346052169799805,
      "eval_runtime": 305.7015,
      "eval_samples_per_second": 1248.247,
      "eval_steps_per_second": 39.009,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.525304855346558e+16,
  "trial_name": null,
  "trial_params": null
}