{
  "best_metric": 3.9601728916168213,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/pp-mod-subj2/transformer/4/checkpoints/checkpoint-305280",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 305280,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.9742,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8299,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.1937,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 5.9777,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.8067,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.6974,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.6015,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5311,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4486,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.395,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3426,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.3029,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 5.2636,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.2093,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.1775,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1231,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.11,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.0736,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0348,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0151,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.0138,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 4.9688,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 4.9538,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.9293,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9034,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 4.8937,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.8584,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8555,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.8372,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 4.8099,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.8026,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 4.7822,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 4.7826,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 4.7628,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.745,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970659010995419e-05,
      "loss": 4.7414,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969820416244367e-05,
      "loss": 4.7434,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.704,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.6983,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.6934,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.6719,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.6665,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.6699,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963951890867376e-05,
      "loss": 4.6565,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 4.6412,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6331,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6248,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6233,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9597605549924894e-05,
      "loss": 4.5941,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 4.5932,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 4.6009,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 4.5926,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956407813868654e-05,
      "loss": 4.5769,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.5596,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 4.5671,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 4.5468,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 4.5659,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952216477993767e-05,
      "loss": 4.5338,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.548,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5273,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5154,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948863736869932e-05,
      "loss": 4.5195,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.5065,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.5026,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.4933,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945510995746097e-05,
      "loss": 4.5031,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 4.4779,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.4742,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4629,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421582546222625e-05,
      "loss": 4.463,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.4808,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 4.4709,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 4.4522,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 4.4504,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 4.4544,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 4.4505,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 4.4334,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 4.4363,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346125397431674e-05,
      "loss": 4.4294,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 4.4282,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 4.4209,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 4.4182,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 4.4215,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 4.4136,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 4.3922,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.928744014366177e-05,
      "loss": 4.4021,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927905419615125e-05,
      "loss": 4.4015,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927066824864073e-05,
      "loss": 4.3966,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.926229867993394e-05,
      "loss": 4.3902,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925391273242342e-05,
      "loss": 4.3798,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92455267849129e-05,
      "loss": 4.3863,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923714083740238e-05,
      "loss": 4.3735,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922877126869559e-05,
      "loss": 4.3823,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922038532118507e-05,
      "loss": 4.3697,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921199937367455e-05,
      "loss": 4.3645,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920362980496776e-05,
      "loss": 4.3822,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919524385745724e-05,
      "loss": 4.3477,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918685790994672e-05,
      "loss": 4.3465,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91784719624362e-05,
      "loss": 4.3567,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 4.3542,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.3595,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.3432,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914494455119785e-05,
      "loss": 4.338,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.3357,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.3387,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.3206,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91114171399595e-05,
      "loss": 4.3197,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910303119244898e-05,
      "loss": 4.3327,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909464524493846e-05,
      "loss": 4.3207,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908625929742794e-05,
      "loss": 4.3133,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907788972872115e-05,
      "loss": 4.3085,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906950378121063e-05,
      "loss": 4.3068,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906111783370011e-05,
      "loss": 4.3187,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905273188618959e-05,
      "loss": 4.3032,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904434593867907e-05,
      "loss": 4.302,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903595999116855e-05,
      "loss": 4.3067,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902757404365803e-05,
      "loss": 4.3055,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901920447495124e-05,
      "loss": 4.2941,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901081852744072e-05,
      "loss": 4.2974,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90024325799302e-05,
      "loss": 4.2963,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994046632419686e-05,
      "loss": 4.2938,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985660684909166e-05,
      "loss": 4.2762,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977274737398646e-05,
      "loss": 4.2887,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968888789888125e-05,
      "loss": 4.2845,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960519221181335e-05,
      "loss": 4.2872,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952133273670815e-05,
      "loss": 4.2716,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.2812,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.2678,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926991809942984e-05,
      "loss": 4.2735,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918605862432463e-05,
      "loss": 4.2535,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910219914921943e-05,
      "loss": 4.2793,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8901833967411423e-05,
      "loss": 4.2781,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.2609,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.2506,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766925036836e-05,
      "loss": 4.2719,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.2568,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.2584,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885153466115204e-05,
      "loss": 4.2516,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.2585,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883477914493473e-05,
      "loss": 4.25,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.2482,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881800724991369e-05,
      "loss": 4.2338,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880962130240316e-05,
      "loss": 4.2477,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880123535489264e-05,
      "loss": 4.2439,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.2363,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878447983867534e-05,
      "loss": 4.2521,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877609389116482e-05,
      "loss": 4.2296,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.2355,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.2415,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.2297,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.243252277374268,
      "eval_runtime": 314.558,
      "eval_samples_per_second": 1213.102,
      "eval_steps_per_second": 37.91,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8742566479926466e-05,
      "loss": 4.2163,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8734180532415946e-05,
      "loss": 4.2204,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8725794584905426e-05,
      "loss": 4.2318,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717408637394906e-05,
      "loss": 4.2209,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709022689884386e-05,
      "loss": 4.2208,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700653121177595e-05,
      "loss": 4.2066,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692267173667075e-05,
      "loss": 4.2082,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683881226156555e-05,
      "loss": 4.2039,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867549527864604e-05,
      "loss": 4.2137,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866710933113552e-05,
      "loss": 4.1992,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865873976242873e-05,
      "loss": 4.2077,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865037019372194e-05,
      "loss": 4.2118,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864198424621142e-05,
      "loss": 4.194,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86335982987009e-05,
      "loss": 4.189,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862521235119038e-05,
      "loss": 4.1804,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861682640367986e-05,
      "loss": 4.1886,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860844045616934e-05,
      "loss": 4.1841,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860005450865882e-05,
      "loss": 4.1801,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85916685611483e-05,
      "loss": 4.1796,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.2044,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.1765,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566527097420475e-05,
      "loss": 4.18,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558141149909955e-05,
      "loss": 4.1835,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549755202399435e-05,
      "loss": 4.1764,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541385633692644e-05,
      "loss": 4.1714,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532999686182124e-05,
      "loss": 4.1642,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524613738671604e-05,
      "loss": 4.1706,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8516227791161084e-05,
      "loss": 4.1666,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850785822245429e-05,
      "loss": 4.1533,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849947227494377e-05,
      "loss": 4.1667,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849108632743325e-05,
      "loss": 4.1562,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848270037992273e-05,
      "loss": 4.1686,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847431443241221e-05,
      "loss": 4.1636,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.1474,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845755891619491e-05,
      "loss": 4.1617,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844917296868439e-05,
      "loss": 4.169,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844078702117387e-05,
      "loss": 4.1394,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.1518,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842403150495656e-05,
      "loss": 4.1457,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841564555744604e-05,
      "loss": 4.1369,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840725960993552e-05,
      "loss": 4.1437,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8398873662425e-05,
      "loss": 4.1508,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839050409371821e-05,
      "loss": 4.1479,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382118146207687e-05,
      "loss": 4.1403,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373732198697167e-05,
      "loss": 4.1439,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365346251186646e-05,
      "loss": 4.1345,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835697668247986e-05,
      "loss": 4.1422,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834859073496934e-05,
      "loss": 4.1209,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834020478745882e-05,
      "loss": 4.1254,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8331818839948295e-05,
      "loss": 4.1397,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832344927124151e-05,
      "loss": 4.138,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831506332373099e-05,
      "loss": 4.1256,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830667737622047e-05,
      "loss": 4.1185,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298291428709944e-05,
      "loss": 4.1288,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289905481199424e-05,
      "loss": 4.1111,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828153591249264e-05,
      "loss": 4.1369,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827314996498212e-05,
      "loss": 4.1172,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82647640174716e-05,
      "loss": 4.1325,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825637806996108e-05,
      "loss": 4.119,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248008501254296e-05,
      "loss": 4.1093,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823962255374377e-05,
      "loss": 4.114,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823123660623325e-05,
      "loss": 4.1178,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822285065872273e-05,
      "loss": 4.1062,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214481090015945e-05,
      "loss": 4.1036,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820609514250542e-05,
      "loss": 4.1193,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81977091949949e-05,
      "loss": 4.1034,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818932324748438e-05,
      "loss": 4.0959,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180953678777594e-05,
      "loss": 4.0941,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172567731267074e-05,
      "loss": 4.0946,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164181783756554e-05,
      "loss": 4.1159,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155795836246034e-05,
      "loss": 4.1126,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147409888735514e-05,
      "loss": 4.0974,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813904032002872e-05,
      "loss": 4.1005,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81306543725182e-05,
      "loss": 4.1034,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812226842500768e-05,
      "loss": 4.099,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811388247749716e-05,
      "loss": 4.0922,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810551290879037e-05,
      "loss": 4.0996,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809712696127985e-05,
      "loss": 4.0932,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808874101376933e-05,
      "loss": 4.0983,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808035506625881e-05,
      "loss": 4.0883,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807198549755203e-05,
      "loss": 4.0921,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806359955004151e-05,
      "loss": 4.0943,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805521360253099e-05,
      "loss": 4.093,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046844033824197e-05,
      "loss": 4.0763,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8038458086313677e-05,
      "loss": 4.0885,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030072138803156e-05,
      "loss": 4.0867,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021686191292636e-05,
      "loss": 4.0865,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013300243782116e-05,
      "loss": 4.0836,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8004930675075325e-05,
      "loss": 4.0718,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996544727564805e-05,
      "loss": 4.0837,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7988158780054285e-05,
      "loss": 4.0733,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7979789211347494e-05,
      "loss": 4.0832,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797140326383698e-05,
      "loss": 4.071,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796301731632646e-05,
      "loss": 4.0742,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795464774761967e-05,
      "loss": 4.0837,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794626180010915e-05,
      "loss": 4.0737,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793787585259863e-05,
      "loss": 4.0556,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792948990508811e-05,
      "loss": 4.0698,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792110395757759e-05,
      "loss": 4.0723,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791271801006707e-05,
      "loss": 4.0762,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790433206255655e-05,
      "loss": 4.0689,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789594611504603e-05,
      "loss": 4.0622,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788757654633924e-05,
      "loss": 4.0584,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787919059882872e-05,
      "loss": 4.0686,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78708046513182e-05,
      "loss": 4.0493,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862435082611415e-05,
      "loss": 4.057,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7854049135100895e-05,
      "loss": 4.0632,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845663187590375e-05,
      "loss": 4.0639,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837293618883584e-05,
      "loss": 4.0412,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828907671373064e-05,
      "loss": 4.0563,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820521723862544e-05,
      "loss": 4.0438,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812135776352024e-05,
      "loss": 4.0637,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803749828841504e-05,
      "loss": 4.0489,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795363881330984e-05,
      "loss": 4.0428,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786977933820464e-05,
      "loss": 4.0618,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778591986309944e-05,
      "loss": 4.0564,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777022241760315e-05,
      "loss": 4.0478,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776183647009263e-05,
      "loss": 4.0533,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775345052258212e-05,
      "loss": 4.0492,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77450645750716e-05,
      "loss": 4.0512,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773667862756108e-05,
      "loss": 4.0317,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772829268005055e-05,
      "loss": 4.0539,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771990673254003e-05,
      "loss": 4.0414,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771152078502951e-05,
      "loss": 4.0496,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770315121632272e-05,
      "loss": 4.0408,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76947652688122e-05,
      "loss": 4.0446,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768637932130168e-05,
      "loss": 4.0377,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767799337379116e-05,
      "loss": 4.0405,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766962380508437e-05,
      "loss": 4.0222,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766123785757386e-05,
      "loss": 4.0548,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765285191006334e-05,
      "loss": 4.045,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764448234135655e-05,
      "loss": 4.0374,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636096393846026e-05,
      "loss": 4.0268,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7627710446335506e-05,
      "loss": 4.0465,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7619324498824986e-05,
      "loss": 4.0397,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610954930118195e-05,
      "loss": 4.0337,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602568982607675e-05,
      "loss": 4.03,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7594183035097155e-05,
      "loss": 4.0399,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7585797087586635e-05,
      "loss": 4.0328,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7577427518879844e-05,
      "loss": 4.0414,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7569041571369324e-05,
      "loss": 4.0188,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756065562385881e-05,
      "loss": 4.034,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755226967634829e-05,
      "loss": 4.032,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75439001076415e-05,
      "loss": 4.0241,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535530538934716e-05,
      "loss": 4.0435,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527144591424196e-05,
      "loss": 4.0232,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751875864391367e-05,
      "loss": 4.0274,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751037269640315e-05,
      "loss": 4.0291,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750198674889263e-05,
      "loss": 4.0275,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.074976444244385,
      "eval_runtime": 334.4678,
      "eval_samples_per_second": 1140.89,
      "eval_steps_per_second": 35.654,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749360080138211e-05,
      "loss": 4.0116,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748521485387159e-05,
      "loss": 4.0201,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747682890636107e-05,
      "loss": 4.028,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746844295885055e-05,
      "loss": 4.0222,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746005701134003e-05,
      "loss": 4.0262,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745167106382951e-05,
      "loss": 4.0053,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744328511631899e-05,
      "loss": 4.013,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7434899168808475e-05,
      "loss": 4.0084,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7426513221297955e-05,
      "loss": 4.0205,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7418127273787435e-05,
      "loss": 4.0084,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740974132627691e-05,
      "loss": 4.0178,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401371757570124e-05,
      "loss": 4.0194,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.739300218886333e-05,
      "loss": 4.0025,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.738461624135281e-05,
      "loss": 4.0057,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737623029384229e-05,
      "loss": 3.9909,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736784434633177e-05,
      "loss": 4.0027,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735945839882125e-05,
      "loss": 4.0014,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.735107245131073e-05,
      "loss": 3.9984,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.734268650380021e-05,
      "loss": 3.9981,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.733431693509343e-05,
      "loss": 4.0208,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.732593098758291e-05,
      "loss": 3.9988,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.731754504007238e-05,
      "loss": 3.9992,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730915909256186e-05,
      "loss": 4.0052,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730078952385508e-05,
      "loss": 4.0029,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.729240357634456e-05,
      "loss": 3.9918,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728401762883403e-05,
      "loss": 3.9883,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727563168132351e-05,
      "loss": 3.9998,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726726211261673e-05,
      "loss": 3.9877,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258892543909936e-05,
      "loss": 3.9834,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250506596399416e-05,
      "loss": 3.9967,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7242120648888896e-05,
      "loss": 3.9867,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723373470137838e-05,
      "loss": 3.9983,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7225348753867856e-05,
      "loss": 3.9926,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7216962806357336e-05,
      "loss": 3.9816,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7208576858846816e-05,
      "loss": 3.9933,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7200190911336296e-05,
      "loss": 4.0054,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7191821342629505e-05,
      "loss": 3.9737,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7183435395118985e-05,
      "loss": 3.9908,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7175049447608465e-05,
      "loss": 3.9774,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7166663500097945e-05,
      "loss": 3.9782,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7158293931391154e-05,
      "loss": 3.9774,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7149907983880634e-05,
      "loss": 3.9919,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714152203637012e-05,
      "loss": 3.9874,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71331360888596e-05,
      "loss": 3.9807,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712476652015281e-05,
      "loss": 3.9837,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711638057264229e-05,
      "loss": 3.9748,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710799462513177e-05,
      "loss": 3.9878,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709962505642498e-05,
      "loss": 3.967,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709123910891446e-05,
      "loss": 3.9692,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708285316140394e-05,
      "loss": 3.9817,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.707446721389342e-05,
      "loss": 3.9803,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70660812663829e-05,
      "loss": 3.9726,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705769531887238e-05,
      "loss": 3.9682,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704932575016559e-05,
      "loss": 3.9778,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7040939802655074e-05,
      "loss": 3.9578,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7032553855144554e-05,
      "loss": 3.9831,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7024167907634034e-05,
      "loss": 3.9691,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7015781960123514e-05,
      "loss": 3.981,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7007396012612994e-05,
      "loss": 3.9747,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6999010065102474e-05,
      "loss": 3.9638,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6990624117591954e-05,
      "loss": 3.964,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698225454888516e-05,
      "loss": 3.9714,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.697386860137464e-05,
      "loss": 3.9578,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696549903266785e-05,
      "loss": 3.9586,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695711308515733e-05,
      "loss": 3.9709,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694872713764681e-05,
      "loss": 3.9614,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694034119013629e-05,
      "loss": 3.9486,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693195524262577e-05,
      "loss": 3.9514,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.692358567391899e-05,
      "loss": 3.9507,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.691519972640847e-05,
      "loss": 3.9751,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690681377889795e-05,
      "loss": 3.9668,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689842783138743e-05,
      "loss": 3.9561,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689004188387691e-05,
      "loss": 3.9622,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688165593636639e-05,
      "loss": 3.9597,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687326998885587e-05,
      "loss": 3.9637,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864900420149076e-05,
      "loss": 3.9523,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856514472638556e-05,
      "loss": 3.9644,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6848128525128036e-05,
      "loss": 3.9592,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683974257761751e-05,
      "loss": 3.9585,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6831356630106996e-05,
      "loss": 3.9522,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6822970682596476e-05,
      "loss": 3.9589,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6814584735085956e-05,
      "loss": 3.9575,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6806198787575436e-05,
      "loss": 3.957,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6797829218868645e-05,
      "loss": 3.9453,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678945965016186e-05,
      "loss": 3.9512,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678107370265134e-05,
      "loss": 3.9528,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6772687755140814e-05,
      "loss": 3.9515,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6764301807630294e-05,
      "loss": 3.9544,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755915860119774e-05,
      "loss": 3.9382,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6747529912609254e-05,
      "loss": 3.9509,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6739143965098734e-05,
      "loss": 3.946,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673077439639194e-05,
      "loss": 3.953,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.672238844888143e-05,
      "loss": 3.9374,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.671400250137091e-05,
      "loss": 3.9453,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.670561655386039e-05,
      "loss": 3.9579,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669723060634987e-05,
      "loss": 3.9435,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668884465883935e-05,
      "loss": 3.9279,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668045871132883e-05,
      "loss": 3.9429,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667208914262204e-05,
      "loss": 3.9467,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.666370319511152e-05,
      "loss": 3.949,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655317247601e-05,
      "loss": 3.9442,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664693130009048e-05,
      "loss": 3.9325,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663854535257996e-05,
      "loss": 3.9344,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663017578387317e-05,
      "loss": 3.9445,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662178983636265e-05,
      "loss": 3.928,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661340388885213e-05,
      "loss": 3.9335,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6605017941341614e-05,
      "loss": 3.9378,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6596631993831094e-05,
      "loss": 3.9416,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65882624251243e-05,
      "loss": 3.9192,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657987647761378e-05,
      "loss": 3.935,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657149053010326e-05,
      "loss": 3.9233,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656310458259274e-05,
      "loss": 3.9383,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655471863508222e-05,
      "loss": 3.9322,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65463326875717e-05,
      "loss": 3.9254,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6537946740061176e-05,
      "loss": 3.9413,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529560792550656e-05,
      "loss": 3.9392,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65212076026476e-05,
      "loss": 3.9272,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.651282165513708e-05,
      "loss": 3.9322,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.650443570762657e-05,
      "loss": 3.933,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.649604976011605e-05,
      "loss": 3.9304,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.648766381260553e-05,
      "loss": 3.9177,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479277865095e-05,
      "loss": 3.9334,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647089191758448e-05,
      "loss": 3.9255,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.646250597007396e-05,
      "loss": 3.9313,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645412002256344e-05,
      "loss": 3.9285,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644573407505292e-05,
      "loss": 3.9308,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.64373481275424e-05,
      "loss": 3.9172,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642897855883561e-05,
      "loss": 3.9246,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642059261132509e-05,
      "loss": 3.9105,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641220666381457e-05,
      "loss": 3.9387,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640382071630405e-05,
      "loss": 3.9263,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6395434768793536e-05,
      "loss": 3.9259,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6387065200086746e-05,
      "loss": 3.9162,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6378679252576226e-05,
      "loss": 3.9344,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6370293305065705e-05,
      "loss": 3.9289,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6361907357555185e-05,
      "loss": 3.92,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6353521410044665e-05,
      "loss": 3.9209,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6345151841337874e-05,
      "loss": 3.9272,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6336765893827354e-05,
      "loss": 3.9212,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6328379946316834e-05,
      "loss": 3.9321,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6319993998806314e-05,
      "loss": 3.9123,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6311624430099523e-05,
      "loss": 3.9232,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6303238482589003e-05,
      "loss": 3.9183,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629485253507848e-05,
      "loss": 3.9138,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628646658756797e-05,
      "loss": 3.9366,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.627808064005745e-05,
      "loss": 3.9122,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626969469254693e-05,
      "loss": 3.9199,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626130874503641e-05,
      "loss": 3.9203,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625292279752589e-05,
      "loss": 3.9187,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.001308441162109,
      "eval_runtime": 301.2019,
      "eval_samples_per_second": 1266.895,
      "eval_steps_per_second": 39.591,
      "step": 228960
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.62445532288191e-05,
      "loss": 3.9121,
      "step": 229376
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.623618366011231e-05,
      "loss": 3.9115,
      "step": 229888
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.622779771260179e-05,
      "loss": 3.9206,
      "step": 230400
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.621941176509127e-05,
      "loss": 3.9177,
      "step": 230912
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.621102581758075e-05,
      "loss": 3.9211,
      "step": 231424
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.620263987007023e-05,
      "loss": 3.9031,
      "step": 231936
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.619427030136344e-05,
      "loss": 3.911,
      "step": 232448
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6185884353852924e-05,
      "loss": 3.8983,
      "step": 232960
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6177498406342404e-05,
      "loss": 3.9228,
      "step": 233472
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6169112458831884e-05,
      "loss": 3.9034,
      "step": 233984
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.616074289012509e-05,
      "loss": 3.913,
      "step": 234496
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.615235694261457e-05,
      "loss": 3.9161,
      "step": 235008
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.614398737390778e-05,
      "loss": 3.9047,
      "step": 235520
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.613560142639726e-05,
      "loss": 3.8971,
      "step": 236032
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.612721547888674e-05,
      "loss": 3.8961,
      "step": 236544
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.611882953137622e-05,
      "loss": 3.8946,
      "step": 237056
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.61104435838657e-05,
      "loss": 3.9007,
      "step": 237568
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.610205763635518e-05,
      "loss": 3.8975,
      "step": 238080
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.609367168884466e-05,
      "loss": 3.8967,
      "step": 238592
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.608528574133414e-05,
      "loss": 3.9184,
      "step": 239104
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.607691617262736e-05,
      "loss": 3.9003,
      "step": 239616
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.606853022511684e-05,
      "loss": 3.899,
      "step": 240128
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.606014427760631e-05,
      "loss": 3.9063,
      "step": 240640
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.605175833009579e-05,
      "loss": 3.903,
      "step": 241152
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6043388761389006e-05,
      "loss": 3.8944,
      "step": 241664
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.603500281387848e-05,
      "loss": 3.89,
      "step": 242176
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.602661686636796e-05,
      "loss": 3.9059,
      "step": 242688
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.601823091885744e-05,
      "loss": 3.8844,
      "step": 243200
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6009861350150655e-05,
      "loss": 3.8887,
      "step": 243712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.600147540264013e-05,
      "loss": 3.8946,
      "step": 244224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5993089455129615e-05,
      "loss": 3.889,
      "step": 244736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5984703507619095e-05,
      "loss": 3.9031,
      "step": 245248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5976317560108575e-05,
      "loss": 3.8922,
      "step": 245760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5967947991401784e-05,
      "loss": 3.8868,
      "step": 246272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5959562043891264e-05,
      "loss": 3.8966,
      "step": 246784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5951176096380744e-05,
      "loss": 3.9063,
      "step": 247296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5942790148870224e-05,
      "loss": 3.8829,
      "step": 247808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.593442058016343e-05,
      "loss": 3.8941,
      "step": 248320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.592603463265291e-05,
      "loss": 3.879,
      "step": 248832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.591764868514239e-05,
      "loss": 3.8926,
      "step": 249344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.590926273763187e-05,
      "loss": 3.879,
      "step": 249856
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.590087679012135e-05,
      "loss": 3.9023,
      "step": 250368
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.589249084261083e-05,
      "loss": 3.892,
      "step": 250880
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.588412127390405e-05,
      "loss": 3.8887,
      "step": 251392
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.587573532639353e-05,
      "loss": 3.8913,
      "step": 251904
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.586734937888301e-05,
      "loss": 3.8838,
      "step": 252416
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.585896343137249e-05,
      "loss": 3.8929,
      "step": 252928
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.58505938626657e-05,
      "loss": 3.8772,
      "step": 253440
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.584220791515518e-05,
      "loss": 3.8739,
      "step": 253952
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.583382196764466e-05,
      "loss": 3.8901,
      "step": 254464
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.582543602013414e-05,
      "loss": 3.8859,
      "step": 254976
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.581706645142735e-05,
      "loss": 3.8872,
      "step": 255488
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580868050391683e-05,
      "loss": 3.8745,
      "step": 256000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580029455640631e-05,
      "loss": 3.8841,
      "step": 256512
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.579190860889579e-05,
      "loss": 3.8681,
      "step": 257024
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5783522661385267e-05,
      "loss": 3.8906,
      "step": 257536
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.577515309267848e-05,
      "loss": 3.8794,
      "step": 258048
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.576676714516796e-05,
      "loss": 3.8944,
      "step": 258560
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.575838119765744e-05,
      "loss": 3.8848,
      "step": 259072
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.574999525014692e-05,
      "loss": 3.8758,
      "step": 259584
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.57416093026364e-05,
      "loss": 3.8732,
      "step": 260096
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.573322335512588e-05,
      "loss": 3.8827,
      "step": 260608
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.572483740761536e-05,
      "loss": 3.8708,
      "step": 261120
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.571646783890857e-05,
      "loss": 3.8703,
      "step": 261632
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.570808189139805e-05,
      "loss": 3.8809,
      "step": 262144
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.569969594388753e-05,
      "loss": 3.8773,
      "step": 262656
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.569130999637701e-05,
      "loss": 3.859,
      "step": 263168
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.568294042767022e-05,
      "loss": 3.8647,
      "step": 263680
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.567455448015971e-05,
      "loss": 3.8611,
      "step": 264192
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.566616853264919e-05,
      "loss": 3.8895,
      "step": 264704
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.565778258513867e-05,
      "loss": 3.8769,
      "step": 265216
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.564939663762814e-05,
      "loss": 3.8719,
      "step": 265728
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5641027068921356e-05,
      "loss": 3.8755,
      "step": 266240
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5632641121410836e-05,
      "loss": 3.8728,
      "step": 266752
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5624255173900316e-05,
      "loss": 3.8842,
      "step": 267264
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.561586922638979e-05,
      "loss": 3.8635,
      "step": 267776
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.560748327887927e-05,
      "loss": 3.885,
      "step": 268288
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5599113710172485e-05,
      "loss": 3.8688,
      "step": 268800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5590727762661965e-05,
      "loss": 3.8735,
      "step": 269312
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5582341815151445e-05,
      "loss": 3.8718,
      "step": 269824
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5573955867640925e-05,
      "loss": 3.8689,
      "step": 270336
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.556558629893414e-05,
      "loss": 3.8788,
      "step": 270848
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5557200351423614e-05,
      "loss": 3.8674,
      "step": 271360
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5548814403913094e-05,
      "loss": 3.8607,
      "step": 271872
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5540428456402574e-05,
      "loss": 3.8692,
      "step": 272384
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5532042508892054e-05,
      "loss": 3.8634,
      "step": 272896
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5523656561381534e-05,
      "loss": 3.8737,
      "step": 273408
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.551528699267474e-05,
      "loss": 3.8703,
      "step": 273920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.550690104516422e-05,
      "loss": 3.8551,
      "step": 274432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.54985150976537e-05,
      "loss": 3.8704,
      "step": 274944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.549012915014318e-05,
      "loss": 3.8606,
      "step": 275456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.548175958143639e-05,
      "loss": 3.8685,
      "step": 275968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.547337363392588e-05,
      "loss": 3.8577,
      "step": 276480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.546498768641536e-05,
      "loss": 3.8658,
      "step": 276992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.545660173890484e-05,
      "loss": 3.8751,
      "step": 277504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.544823217019805e-05,
      "loss": 3.8597,
      "step": 278016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.543984622268753e-05,
      "loss": 3.8508,
      "step": 278528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.543146027517701e-05,
      "loss": 3.861,
      "step": 279040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.542307432766649e-05,
      "loss": 3.8604,
      "step": 279552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5414704758959696e-05,
      "loss": 3.8729,
      "step": 280064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5406318811449176e-05,
      "loss": 3.8619,
      "step": 280576
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5397932863938656e-05,
      "loss": 3.855,
      "step": 281088
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5389546916428136e-05,
      "loss": 3.8556,
      "step": 281600
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5381160968917616e-05,
      "loss": 3.8671,
      "step": 282112
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.537279140021083e-05,
      "loss": 3.8458,
      "step": 282624
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.536440545270031e-05,
      "loss": 3.8574,
      "step": 283136
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.535601950518979e-05,
      "loss": 3.858,
      "step": 283648
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.534763355767927e-05,
      "loss": 3.8635,
      "step": 284160
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.533926398897248e-05,
      "loss": 3.8353,
      "step": 284672
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.533087804146196e-05,
      "loss": 3.856,
      "step": 285184
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.532249209395144e-05,
      "loss": 3.8518,
      "step": 285696
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.531410614644092e-05,
      "loss": 3.8527,
      "step": 286208
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.530573657773413e-05,
      "loss": 3.8587,
      "step": 286720
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.529735063022361e-05,
      "loss": 3.8428,
      "step": 287232
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528896468271309e-05,
      "loss": 3.8654,
      "step": 287744
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528057873520257e-05,
      "loss": 3.8582,
      "step": 288256
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5272225545299515e-05,
      "loss": 3.8498,
      "step": 288768
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5263839597788995e-05,
      "loss": 3.8575,
      "step": 289280
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5255453650278475e-05,
      "loss": 3.8533,
      "step": 289792
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5247067702767955e-05,
      "loss": 3.8547,
      "step": 290304
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5238681755257435e-05,
      "loss": 3.836,
      "step": 290816
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5230295807746915e-05,
      "loss": 3.8608,
      "step": 291328
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5221909860236395e-05,
      "loss": 3.8455,
      "step": 291840
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5213523912725875e-05,
      "loss": 3.8525,
      "step": 292352
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5205154344019084e-05,
      "loss": 3.8513,
      "step": 292864
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5196768396508564e-05,
      "loss": 3.8596,
      "step": 293376
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5188382448998044e-05,
      "loss": 3.8457,
      "step": 293888
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5179996501487524e-05,
      "loss": 3.843,
      "step": 294400
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5171610553977003e-05,
      "loss": 3.8395,
      "step": 294912
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.516324098527022e-05,
      "loss": 3.8582,
      "step": 295424
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.51548550377597e-05,
      "loss": 3.8511,
      "step": 295936
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.514646909024918e-05,
      "loss": 3.8514,
      "step": 296448
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.513808314273866e-05,
      "loss": 3.8414,
      "step": 296960
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.512971357403187e-05,
      "loss": 3.8581,
      "step": 297472
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.512132762652135e-05,
      "loss": 3.8533,
      "step": 297984
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.511294167901083e-05,
      "loss": 3.8446,
      "step": 298496
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.510455573150031e-05,
      "loss": 3.8437,
      "step": 299008
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.509618616279352e-05,
      "loss": 3.852,
      "step": 299520
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5087800215283e-05,
      "loss": 3.8505,
      "step": 300032
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.507941426777248e-05,
      "loss": 3.8559,
      "step": 300544
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.507102832026196e-05,
      "loss": 3.8383,
      "step": 301056
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.506264237275144e-05,
      "loss": 3.8491,
      "step": 301568
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.505427280404465e-05,
      "loss": 3.8429,
      "step": 302080
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.504588685653413e-05,
      "loss": 3.8432,
      "step": 302592
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.503750090902361e-05,
      "loss": 3.8576,
      "step": 303104
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.502911496151309e-05,
      "loss": 3.8413,
      "step": 303616
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.502072901400257e-05,
      "loss": 3.8459,
      "step": 304128
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.501235944529578e-05,
      "loss": 3.8429,
      "step": 304640
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.500397349778526e-05,
      "loss": 3.8499,
      "step": 305152
    },
    {
      "epoch": 1.03,
      "eval_loss": 3.9601728916168213,
      "eval_runtime": 301.1519,
      "eval_samples_per_second": 1267.105,
      "eval_steps_per_second": 39.598,
      "step": 305280
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.9807281118550016e+17,
  "trial_name": null,
  "trial_params": null
}