{
  "best_metric": 4.335472106933594,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/binding-c-command/lstm/0/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": 10.8209,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5535,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0546,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9931,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9443,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.894,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.727,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6298,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5262,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 6.4594,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 6.3927,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 6.3337,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 6.2739,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989098268236324e-05,
      "loss": 6.1987,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1549,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0934,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0554,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.0083,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9618,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9254,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8891,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8527,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8223,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7851,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979875363855125e-05,
      "loss": 5.7676,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.979036769104073e-05,
      "loss": 5.7383,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7084,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6882,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.97652262273129e-05,
      "loss": 5.6597,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.975684027980238e-05,
      "loss": 5.6424,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6238,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 5.6022,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.58,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5579,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5529,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5431,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 5.5157,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968980183612942e-05,
      "loss": 5.4749,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96814158886189e-05,
      "loss": 5.4795,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 5.4488,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 5.4497,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.4329,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 5.4199,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 5.3947,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 5.3985,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 5.3857,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 5.3831,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.366,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3456,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 5.338,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 5.3396,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.3168,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3083,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.2929,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.2902,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2727,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2862,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 5.2498,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.2553,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.247,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949699055860238e-05,
      "loss": 5.2258,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 5.2278,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9480267799992536e-05,
      "loss": 5.2148,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9471881852482016e-05,
      "loss": 5.1968,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946349590497149e-05,
      "loss": 5.19,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945510995746097e-05,
      "loss": 5.2081,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 5.1821,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9438354441243665e-05,
      "loss": 5.1719,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9429968493733145e-05,
      "loss": 5.1546,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421582546222625e-05,
      "loss": 5.1635,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 5.1683,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 5.1514,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 5.1452,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 5.149,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 5.1401,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 5.1275,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 5.1061,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 5.1133,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346125397431674e-05,
      "loss": 5.0958,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 5.0941,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 5.0872,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932098393370384e-05,
      "loss": 5.0844,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931259798619332e-05,
      "loss": 5.093,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930421203868281e-05,
      "loss": 5.0819,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 5.0549,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.928744014366177e-05,
      "loss": 5.0561,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 5.0596,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927068462744446e-05,
      "loss": 5.0558,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.926229867993394e-05,
      "loss": 5.0461,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925391273242342e-05,
      "loss": 5.0399,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92455267849129e-05,
      "loss": 5.0327,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923715721620611e-05,
      "loss": 5.0186,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922877126869559e-05,
      "loss": 5.0237,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922038532118507e-05,
      "loss": 5.0162,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921199937367455e-05,
      "loss": 5.0074,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 5.0146,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919522747865351e-05,
      "loss": 4.9952,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 4.9911,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 4.982,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917010239372941e-05,
      "loss": 4.9771,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916171644621889e-05,
      "loss": 4.9846,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915333049870837e-05,
      "loss": 4.9692,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914494455119785e-05,
      "loss": 4.9635,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.9596,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.9585,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.9404,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91114171399595e-05,
      "loss": 4.9307,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910303119244898e-05,
      "loss": 4.9368,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909464524493846e-05,
      "loss": 4.9431,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908625929742794e-05,
      "loss": 4.9329,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907787334991742e-05,
      "loss": 4.9311,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906950378121063e-05,
      "loss": 4.9164,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906111783370011e-05,
      "loss": 4.9208,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905273188618959e-05,
      "loss": 4.9078,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904434593867907e-05,
      "loss": 4.913,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903595999116855e-05,
      "loss": 4.9062,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902759042246176e-05,
      "loss": 4.9046,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901920447495124e-05,
      "loss": 4.8971,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901081852744072e-05,
      "loss": 4.8926,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90024325799302e-05,
      "loss": 4.8958,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994046632419686e-05,
      "loss": 4.8948,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985660684909166e-05,
      "loss": 4.8828,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977274737398646e-05,
      "loss": 4.8833,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968888789888125e-05,
      "loss": 4.8743,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960519221181335e-05,
      "loss": 4.8772,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952133273670815e-05,
      "loss": 4.8568,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.8531,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.8574,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926991809942984e-05,
      "loss": 4.8574,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918605862432463e-05,
      "loss": 4.8475,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910219914921943e-05,
      "loss": 4.8597,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8901833967411423e-05,
      "loss": 4.8512,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889346439870464e-05,
      "loss": 4.8355,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888507845119412e-05,
      "loss": 4.8408,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766925036836e-05,
      "loss": 4.844,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.835,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885993698746629e-05,
      "loss": 4.8312,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515674187595e-05,
      "loss": 4.823,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884318147124898e-05,
      "loss": 4.826,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883479552373846e-05,
      "loss": 4.8298,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882640957622794e-05,
      "loss": 4.8112,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881802362871742e-05,
      "loss": 4.8051,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88096376812069e-05,
      "loss": 4.8115,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8801268112500106e-05,
      "loss": 4.8033,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879289854379332e-05,
      "loss": 4.8001,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87845125962828e-05,
      "loss": 4.8024,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877612664877228e-05,
      "loss": 4.7947,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876774070126176e-05,
      "loss": 4.7939,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875935475375124e-05,
      "loss": 4.7902,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875096880624072e-05,
      "loss": 4.7849,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.752007484436035,
      "eval_runtime": 290.964,
      "eval_samples_per_second": 1311.471,
      "eval_steps_per_second": 40.984,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87425828587302e-05,
      "loss": 4.7788,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873419691121968e-05,
      "loss": 4.7817,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872581096370916e-05,
      "loss": 4.7889,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717425016198635e-05,
      "loss": 4.7769,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709039068688115e-05,
      "loss": 4.7747,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870066949998133e-05,
      "loss": 4.77,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869228355247081e-05,
      "loss": 4.7722,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868389760496029e-05,
      "loss": 4.7517,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675528036253507e-05,
      "loss": 4.7741,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667142088742986e-05,
      "loss": 4.7594,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865875614123246e-05,
      "loss": 4.767,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865037019372194e-05,
      "loss": 4.7652,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8642000625015155e-05,
      "loss": 4.7458,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633614677504635e-05,
      "loss": 4.7478,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862522872999411e-05,
      "loss": 4.7412,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861684278248359e-05,
      "loss": 4.7365,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860845683497307e-05,
      "loss": 4.7441,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860007088746255e-05,
      "loss": 4.7316,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.7296,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.7462,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.7306,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566527097420475e-05,
      "loss": 4.7251,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558141149909955e-05,
      "loss": 4.7174,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.7329,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854140201249637e-05,
      "loss": 4.7128,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853301606498585e-05,
      "loss": 4.7066,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852463011747533e-05,
      "loss": 4.7081,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851624416996481e-05,
      "loss": 4.7102,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850785822245429e-05,
      "loss": 4.7027,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84994886537475e-05,
      "loss": 4.7035,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849110270623698e-05,
      "loss": 4.704,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848271675872646e-05,
      "loss": 4.7031,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847433081121595e-05,
      "loss": 4.6921,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.7028,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845755891619491e-05,
      "loss": 4.7052,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844918934748812e-05,
      "loss": 4.6961,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84408033999776e-05,
      "loss": 4.6691,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.6793,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842403150495656e-05,
      "loss": 4.6753,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841564555744604e-05,
      "loss": 4.6816,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840725960993552e-05,
      "loss": 4.6726,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8398873662425e-05,
      "loss": 4.6763,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839048771491447e-05,
      "loss": 4.6633,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382118146207687e-05,
      "loss": 4.6776,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373732198697167e-05,
      "loss": 4.6726,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365346251186646e-05,
      "loss": 4.6724,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8356960303676126e-05,
      "loss": 4.6726,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834859073496934e-05,
      "loss": 4.6532,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834020478745882e-05,
      "loss": 4.6607,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8331818839948295e-05,
      "loss": 4.666,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832344927124151e-05,
      "loss": 4.6579,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831506332373099e-05,
      "loss": 4.6552,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830667737622047e-05,
      "loss": 4.6445,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298291428709944e-05,
      "loss": 4.6515,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289905481199424e-05,
      "loss": 4.6349,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281519533688904e-05,
      "loss": 4.6645,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273133586178384e-05,
      "loss": 4.6361,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826474763866787e-05,
      "loss": 4.6477,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825637806996108e-05,
      "loss": 4.6449,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.824799212245056e-05,
      "loss": 4.6229,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823960617494004e-05,
      "loss": 4.637,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823122022742952e-05,
      "loss": 4.6308,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222834279919e-05,
      "loss": 4.6157,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821446471121221e-05,
      "loss": 4.6161,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820607876370169e-05,
      "loss": 4.6352,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819769281619117e-05,
      "loss": 4.6229,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818930686868065e-05,
      "loss": 4.6132,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818092092117013e-05,
      "loss": 4.6081,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817253497365961e-05,
      "loss": 4.6222,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816414902614909e-05,
      "loss": 4.629,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155779457442305e-05,
      "loss": 4.6211,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147393509931785e-05,
      "loss": 4.6145,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8139007562421265e-05,
      "loss": 4.6289,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8130621614910745e-05,
      "loss": 4.6209,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8122235667400224e-05,
      "loss": 4.6188,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113866098693434e-05,
      "loss": 4.5985,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105480151182914e-05,
      "loss": 4.6106,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8097094203672393e-05,
      "loss": 4.604,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8088708256161873e-05,
      "loss": 4.599,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808033868745508e-05,
      "loss": 4.6007,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807195273994456e-05,
      "loss": 4.6002,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806356679243404e-05,
      "loss": 4.6134,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805518084492352e-05,
      "loss": 4.6092,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046794897413e-05,
      "loss": 4.5839,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803842532870622e-05,
      "loss": 4.5904,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80300393811957e-05,
      "loss": 4.5927,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802165343368518e-05,
      "loss": 4.5941,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801326748617466e-05,
      "loss": 4.5926,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800488153866413e-05,
      "loss": 4.5898,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799649559115361e-05,
      "loss": 4.5838,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798810964364309e-05,
      "loss": 4.5761,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79797400749363e-05,
      "loss": 4.5862,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797135412742578e-05,
      "loss": 4.5799,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796296817991526e-05,
      "loss": 4.5793,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795458223240474e-05,
      "loss": 4.5857,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794619628489423e-05,
      "loss": 4.5763,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793781033738371e-05,
      "loss": 4.5693,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792942438987319e-05,
      "loss": 4.5674,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7921054821166396e-05,
      "loss": 4.5705,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912668873655876e-05,
      "loss": 4.5754,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7904282926145356e-05,
      "loss": 4.5621,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895913357438565e-05,
      "loss": 4.5606,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887527409928045e-05,
      "loss": 4.5617,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879141462417525e-05,
      "loss": 4.5627,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870755514907005e-05,
      "loss": 4.5517,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862369567396485e-05,
      "loss": 4.5395,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853983619885965e-05,
      "loss": 4.5551,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845597672375445e-05,
      "loss": 4.557,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837211724864925e-05,
      "loss": 4.5516,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782884215615814e-05,
      "loss": 4.5529,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782045620864762e-05,
      "loss": 4.545,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78120702611371e-05,
      "loss": 4.5516,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780368431362658e-05,
      "loss": 4.5394,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779531474491979e-05,
      "loss": 4.5431,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778692879740927e-05,
      "loss": 4.5485,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777854284989875e-05,
      "loss": 4.5483,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777015690238823e-05,
      "loss": 4.5397,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776177095487771e-05,
      "loss": 4.5397,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775338500736719e-05,
      "loss": 4.5462,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774499905985666e-05,
      "loss": 4.5458,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773661311234615e-05,
      "loss": 4.53,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772822716483563e-05,
      "loss": 4.5428,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771984121732511e-05,
      "loss": 4.53,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771147164861832e-05,
      "loss": 4.5439,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77030857011078e-05,
      "loss": 4.5209,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769469975359728e-05,
      "loss": 4.5201,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686330184890494e-05,
      "loss": 4.5238,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767794423737997e-05,
      "loss": 4.5312,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766955828986945e-05,
      "loss": 4.5214,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766117234235893e-05,
      "loss": 4.5378,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765278639484841e-05,
      "loss": 4.5275,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764440044733789e-05,
      "loss": 4.5198,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763601449982737e-05,
      "loss": 4.5229,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762762855231685e-05,
      "loss": 4.5313,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761924260480633e-05,
      "loss": 4.5215,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610856657295813e-05,
      "loss": 4.5207,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602470709785293e-05,
      "loss": 4.5145,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75941011410785e-05,
      "loss": 4.5191,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758571519356798e-05,
      "loss": 4.5227,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757732924605746e-05,
      "loss": 4.511,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756894329854694e-05,
      "loss": 4.5087,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756055735103642e-05,
      "loss": 4.514,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75521714035259e-05,
      "loss": 4.5045,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543785456015375e-05,
      "loss": 4.5049,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535399508504855e-05,
      "loss": 4.5124,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752702993979807e-05,
      "loss": 4.504,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751864399228755e-05,
      "loss": 4.503,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751025804477703e-05,
      "loss": 4.5034,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750187209726651e-05,
      "loss": 4.4971,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.470280647277832,
      "eval_runtime": 291.1643,
      "eval_samples_per_second": 1310.569,
      "eval_steps_per_second": 40.956,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749348614975599e-05,
      "loss": 4.491,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748510020224547e-05,
      "loss": 4.4977,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747671425473495e-05,
      "loss": 4.5105,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746832830722443e-05,
      "loss": 4.4992,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745994235971391e-05,
      "loss": 4.4995,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745155641220339e-05,
      "loss": 4.4931,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744317046469287e-05,
      "loss": 4.4997,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743478451718235e-05,
      "loss": 4.4793,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742639856967183e-05,
      "loss": 4.5019,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741801262216131e-05,
      "loss": 4.4983,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740962667465079e-05,
      "loss": 4.4946,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401240727140264e-05,
      "loss": 4.5055,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.739287115843348e-05,
      "loss": 4.4785,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384485210922966e-05,
      "loss": 4.4836,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376099263412446e-05,
      "loss": 4.4894,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736771331590192e-05,
      "loss": 4.4716,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73593273683914e-05,
      "loss": 4.4851,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735094142088088e-05,
      "loss": 4.4748,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734255547337036e-05,
      "loss": 4.4754,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733416952585984e-05,
      "loss": 4.4976,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732578357834932e-05,
      "loss": 4.4773,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.73173976308388e-05,
      "loss": 4.4797,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730901168332828e-05,
      "loss": 4.4679,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730064211462149e-05,
      "loss": 4.4857,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292272545914704e-05,
      "loss": 4.4616,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7283886598404184e-05,
      "loss": 4.4642,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7275500650893664e-05,
      "loss": 4.4627,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7267114703383144e-05,
      "loss": 4.4729,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258728755872624e-05,
      "loss": 4.4609,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250342808362104e-05,
      "loss": 4.4615,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7241956860851584e-05,
      "loss": 4.4688,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723358729214479e-05,
      "loss": 4.4705,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722520134463427e-05,
      "loss": 4.4535,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721681539712375e-05,
      "loss": 4.4689,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720842944961323e-05,
      "loss": 4.4709,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720005988090644e-05,
      "loss": 4.4634,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719169031219965e-05,
      "loss": 4.4411,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718330436468914e-05,
      "loss": 4.4541,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717491841717862e-05,
      "loss": 4.4425,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71665324696681e-05,
      "loss": 4.4587,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715816290096131e-05,
      "loss": 4.4488,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714977695345079e-05,
      "loss": 4.454,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714139100594027e-05,
      "loss": 4.4429,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713300505842975e-05,
      "loss": 4.4539,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712461911091923e-05,
      "loss": 4.4526,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7116249542212436e-05,
      "loss": 4.4532,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7107863594701916e-05,
      "loss": 4.455,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7099477647191396e-05,
      "loss": 4.4335,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7091091699680876e-05,
      "loss": 4.442,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082705752170356e-05,
      "loss": 4.457,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074319804659836e-05,
      "loss": 4.4402,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706593385714932e-05,
      "loss": 4.4406,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70575479096388e-05,
      "loss": 4.4359,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704917834093201e-05,
      "loss": 4.4403,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704080877222522e-05,
      "loss": 4.4288,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703243920351843e-05,
      "loss": 4.4481,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702405325600791e-05,
      "loss": 4.4335,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701566730849739e-05,
      "loss": 4.4374,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700728136098687e-05,
      "loss": 4.4444,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699889541347635e-05,
      "loss": 4.416,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699050946596583e-05,
      "loss": 4.4332,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698212351845531e-05,
      "loss": 4.4286,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.697373757094479e-05,
      "loss": 4.4184,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965368002238005e-05,
      "loss": 4.4183,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956982054727485e-05,
      "loss": 4.4298,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948596107216965e-05,
      "loss": 4.4278,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940210159706445e-05,
      "loss": 4.4157,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931824212195925e-05,
      "loss": 4.409,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923454643489134e-05,
      "loss": 4.4285,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6915068695978614e-05,
      "loss": 4.4371,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906682748468094e-05,
      "loss": 4.4251,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898296800957574e-05,
      "loss": 4.4232,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688991085344705e-05,
      "loss": 4.433,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688152490593653e-05,
      "loss": 4.4297,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873138958426014e-05,
      "loss": 4.4296,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864753010915494e-05,
      "loss": 4.4057,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.685639982101244e-05,
      "loss": 4.4219,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684801387350192e-05,
      "loss": 4.4213,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68396279259914e-05,
      "loss": 4.4094,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683125835728461e-05,
      "loss": 4.4189,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682287240977409e-05,
      "loss": 4.4155,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681448646226357e-05,
      "loss": 4.429,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680610051475305e-05,
      "loss": 4.4213,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679773094604626e-05,
      "loss": 4.4018,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678934499853574e-05,
      "loss": 4.4087,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678095905102522e-05,
      "loss": 4.4101,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6772573103514697e-05,
      "loss": 4.4145,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6764187156004177e-05,
      "loss": 4.4137,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755801208493657e-05,
      "loss": 4.408,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6747415260983136e-05,
      "loss": 4.4055,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6739029313472616e-05,
      "loss": 4.3976,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6730659744765826e-05,
      "loss": 4.4107,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6722273797255305e-05,
      "loss": 4.3994,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6713887849744785e-05,
      "loss": 4.407,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705501902234265e-05,
      "loss": 4.41,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6697115954723745e-05,
      "loss": 4.4033,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6688730007213225e-05,
      "loss": 4.3971,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6680344059702705e-05,
      "loss": 4.3985,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667197449099592e-05,
      "loss": 4.3991,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66635885434854e-05,
      "loss": 4.4031,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665520259597488e-05,
      "loss": 4.3946,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664681664846436e-05,
      "loss": 4.3914,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663843070095384e-05,
      "loss": 4.3905,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663004475344332e-05,
      "loss": 4.3948,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66216588059328e-05,
      "loss": 4.3856,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661327285842228e-05,
      "loss": 4.3752,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660490328971549e-05,
      "loss": 4.3863,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659651734220497e-05,
      "loss": 4.3932,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658813139469445e-05,
      "loss": 4.3841,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657976182598766e-05,
      "loss": 4.3936,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657137587847714e-05,
      "loss": 4.3801,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656298993096662e-05,
      "loss": 4.3899,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65546039834561e-05,
      "loss": 4.3805,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6546218035945586e-05,
      "loss": 4.3775,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6537848467238795e-05,
      "loss": 4.392,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529462519728275e-05,
      "loss": 4.3889,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6521076572217755e-05,
      "loss": 4.3819,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512690624707234e-05,
      "loss": 4.3834,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650430467719671e-05,
      "loss": 4.3833,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649591872968619e-05,
      "loss": 4.3879,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648753278217567e-05,
      "loss": 4.3754,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647914683466515e-05,
      "loss": 4.3877,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647077726595836e-05,
      "loss": 4.3749,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6462391318447837e-05,
      "loss": 4.3873,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645400537093732e-05,
      "loss": 4.3633,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644563580223053e-05,
      "loss": 4.3725,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643724985472001e-05,
      "loss": 4.3672,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642886390720949e-05,
      "loss": 4.3772,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642047795969897e-05,
      "loss": 4.3692,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641209201218845e-05,
      "loss": 4.3884,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640370606467793e-05,
      "loss": 4.3768,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639532011716741e-05,
      "loss": 4.3721,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638695054846062e-05,
      "loss": 4.3654,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637858097975383e-05,
      "loss": 4.3805,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637019503224331e-05,
      "loss": 4.3738,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636180908473279e-05,
      "loss": 4.3753,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635342313722228e-05,
      "loss": 4.3681,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634503718971176e-05,
      "loss": 4.3686,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633665124220124e-05,
      "loss": 4.3768,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6328281673494446e-05,
      "loss": 4.3653,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6319895725983926e-05,
      "loss": 4.3619,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6311509778473406e-05,
      "loss": 4.3729,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6303123830962886e-05,
      "loss": 4.3624,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6294737883452366e-05,
      "loss": 4.3586,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6286351935941846e-05,
      "loss": 4.3633,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277965988431326e-05,
      "loss": 4.3673,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269580040920806e-05,
      "loss": 4.358,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261226851017744e-05,
      "loss": 4.3579,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625284090350723e-05,
      "loss": 4.3536,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.335472106933594,
      "eval_runtime": 293.6337,
      "eval_samples_per_second": 1299.548,
      "eval_steps_per_second": 40.612,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.49904359122191e+16,
  "trial_name": null,
  "trial_params": null
}