{
  "best_metric": 4.050783157348633,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/existential-there-quantifier/transformer/1/checkpoints/checkpoint-152640",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 152640,
  "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": 10.9108,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8267,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.2163,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.0127,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.8444,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.7417,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.6306,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5716,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4939,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.4304,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3863,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.3488,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 5.3048,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.2514,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.217,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1811,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1465,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.1238,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0945,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0646,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.0527,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 5.0194,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 5.0016,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.9744,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9643,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 4.9331,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.9124,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8991,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.877,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756873037409844e-05,
      "loss": 4.854,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.8489,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740117521192533e-05,
      "loss": 4.8365,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731731573682013e-05,
      "loss": 4.8238,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 4.8043,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.8018,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7759,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 4.768,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.7379,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.7409,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.7269,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.7239,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.7113,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.708,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6873,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 4.682,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6756,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6588,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6673,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 4.642,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 4.6279,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 4.6397,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 4.6192,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 4.6128,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 4.591,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 4.6015,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 4.5786,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 4.6079,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 4.5705,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.5846,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5735,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5489,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5431,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.5464,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.5298,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.5237,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 4.5428,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 4.526,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.5138,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4952,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421582546222625e-05,
      "loss": 4.4949,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.5051,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 4.5054,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 4.4952,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 4.492,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 4.4899,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 4.483,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 4.4707,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 4.4667,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346125397431674e-05,
      "loss": 4.4643,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 4.4524,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 4.4537,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 4.4556,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 4.4556,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 4.4381,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 4.4173,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92874565224655e-05,
      "loss": 4.4385,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 4.4309,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927068462744446e-05,
      "loss": 4.4299,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.926229867993394e-05,
      "loss": 4.4217,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925391273242342e-05,
      "loss": 4.4228,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92455267849129e-05,
      "loss": 4.4127,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923714083740238e-05,
      "loss": 4.4018,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922877126869559e-05,
      "loss": 4.4073,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922038532118507e-05,
      "loss": 4.4027,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921199937367455e-05,
      "loss": 4.4107,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 4.4075,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919522747865351e-05,
      "loss": 4.4001,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 4.3808,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 4.3848,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917006963612195e-05,
      "loss": 4.3865,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.3849,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.3738,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914494455119785e-05,
      "loss": 4.373,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.3626,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.378,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.3464,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91114171399595e-05,
      "loss": 4.3589,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910303119244898e-05,
      "loss": 4.3608,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.909464524493846e-05,
      "loss": 4.3472,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908625929742794e-05,
      "loss": 4.3491,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907787334991742e-05,
      "loss": 4.3502,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90694874024069e-05,
      "loss": 4.3401,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906110145489638e-05,
      "loss": 4.3475,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905271550738586e-05,
      "loss": 4.3424,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904434593867907e-05,
      "loss": 4.3348,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903595999116855e-05,
      "loss": 4.3395,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902757404365803e-05,
      "loss": 4.3441,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901920447495124e-05,
      "loss": 4.3312,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901081852744072e-05,
      "loss": 4.3214,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90024325799302e-05,
      "loss": 4.3255,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994046632419686e-05,
      "loss": 4.3268,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985677063712895e-05,
      "loss": 4.3187,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977291116202375e-05,
      "loss": 4.3089,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968905168691855e-05,
      "loss": 4.3166,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960519221181335e-05,
      "loss": 4.3193,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952133273670815e-05,
      "loss": 4.3092,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943747326160294e-05,
      "loss": 4.3065,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935361378649774e-05,
      "loss": 4.298,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926991809942984e-05,
      "loss": 4.3017,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918605862432463e-05,
      "loss": 4.2959,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910219914921943e-05,
      "loss": 4.3112,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8901833967411423e-05,
      "loss": 4.3009,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889346439870464e-05,
      "loss": 4.2811,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888507845119412e-05,
      "loss": 4.2911,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766925036836e-05,
      "loss": 4.2957,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.2862,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885992060866256e-05,
      "loss": 4.2926,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885155103995577e-05,
      "loss": 4.278,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.2863,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883477914493473e-05,
      "loss": 4.2894,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.2789,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881802362871742e-05,
      "loss": 4.2746,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88096376812069e-05,
      "loss": 4.2757,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880125173369638e-05,
      "loss": 4.2645,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879286578618586e-05,
      "loss": 4.2625,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878447983867534e-05,
      "loss": 4.2768,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877609389116482e-05,
      "loss": 4.2552,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.2612,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875933837494751e-05,
      "loss": 4.2659,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8750952427436986e-05,
      "loss": 4.2616,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.224004745483398,
      "eval_runtime": 304.7973,
      "eval_samples_per_second": 1251.95,
      "eval_steps_per_second": 39.124,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8742566479926466e-05,
      "loss": 4.2434,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8734180532415946e-05,
      "loss": 4.245,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8725794584905426e-05,
      "loss": 4.2625,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717425016198635e-05,
      "loss": 4.2449,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709039068688115e-05,
      "loss": 4.2573,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700653121177595e-05,
      "loss": 4.2314,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692267173667075e-05,
      "loss": 4.2477,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683881226156555e-05,
      "loss": 4.2341,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.867551165744977e-05,
      "loss": 4.2328,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.866712570993925e-05,
      "loss": 4.2359,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865873976242873e-05,
      "loss": 4.2355,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865035381491821e-05,
      "loss": 4.238,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8642000625015155e-05,
      "loss": 4.223,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633614677504635e-05,
      "loss": 4.2201,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862522872999411e-05,
      "loss": 4.2154,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861684278248359e-05,
      "loss": 4.2132,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860845683497307e-05,
      "loss": 4.2133,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860007088746255e-05,
      "loss": 4.2165,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.2127,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.235,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574929423734724e-05,
      "loss": 4.2104,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566543476224204e-05,
      "loss": 4.2166,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558157528713684e-05,
      "loss": 4.2089,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.2144,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854140201249637e-05,
      "loss": 4.2031,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853301606498585e-05,
      "loss": 4.2024,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852463011747533e-05,
      "loss": 4.2014,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851624416996481e-05,
      "loss": 4.1934,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850787460125802e-05,
      "loss": 4.1794,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84994886537475e-05,
      "loss": 4.1956,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849110270623698e-05,
      "loss": 4.1916,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848271675872646e-05,
      "loss": 4.1974,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847433081121595e-05,
      "loss": 4.1882,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.1935,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845757529499864e-05,
      "loss": 4.1835,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844918934748812e-05,
      "loss": 4.1815,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84408033999776e-05,
      "loss": 4.1672,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.1764,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842404788376029e-05,
      "loss": 4.1718,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841566193624977e-05,
      "loss": 4.1735,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840727598873925e-05,
      "loss": 4.1733,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839889004122873e-05,
      "loss": 4.1795,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8390520472521936e-05,
      "loss": 4.1691,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382134525011416e-05,
      "loss": 4.1703,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83737485775009e-05,
      "loss": 4.1717,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836536262999038e-05,
      "loss": 4.1585,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835699306128359e-05,
      "loss": 4.1682,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834860711377307e-05,
      "loss": 4.1573,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834022116626255e-05,
      "loss": 4.1493,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833183521875203e-05,
      "loss": 4.1661,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832346565004524e-05,
      "loss": 4.1524,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831507970253472e-05,
      "loss": 4.1488,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83066937550242e-05,
      "loss": 4.1418,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829830780751368e-05,
      "loss": 4.1498,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828993823880689e-05,
      "loss": 4.1352,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828155229129637e-05,
      "loss": 4.1644,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273166343785856e-05,
      "loss": 4.1409,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264780396275336e-05,
      "loss": 4.1544,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256410827568545e-05,
      "loss": 4.1531,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248024880058025e-05,
      "loss": 4.1324,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8239638932547505e-05,
      "loss": 4.1279,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231252985036985e-05,
      "loss": 4.1416,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222867037526465e-05,
      "loss": 4.1284,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214497468819674e-05,
      "loss": 4.1239,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206111521309154e-05,
      "loss": 4.1438,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197725573798634e-05,
      "loss": 4.1395,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189339626288114e-05,
      "loss": 4.1246,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818097005758132e-05,
      "loss": 4.1166,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817258411007081e-05,
      "loss": 4.1216,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816419816256029e-05,
      "loss": 4.1323,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81558285938535e-05,
      "loss": 4.1314,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814744264634298e-05,
      "loss": 4.1307,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813905669883246e-05,
      "loss": 4.1319,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813067075132194e-05,
      "loss": 4.1314,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812228480381141e-05,
      "loss": 4.1319,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811391523510463e-05,
      "loss": 4.1143,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810552928759411e-05,
      "loss": 4.124,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809714334008359e-05,
      "loss": 4.1209,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808875739257306e-05,
      "loss": 4.1072,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808038782386628e-05,
      "loss": 4.119,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807200187635576e-05,
      "loss": 4.1204,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063615928845243e-05,
      "loss": 4.1248,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055229981334717e-05,
      "loss": 4.1045,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804686041262793e-05,
      "loss": 4.0936,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803847446511741e-05,
      "loss": 4.1109,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030088517606886e-05,
      "loss": 4.1093,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021702570096366e-05,
      "loss": 4.1129,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801333300138958e-05,
      "loss": 4.1049,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800496343268279e-05,
      "loss": 4.1048,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799657748517227e-05,
      "loss": 4.1039,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798819153766175e-05,
      "loss": 4.0931,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797980559015123e-05,
      "loss": 4.1016,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797141964264071e-05,
      "loss": 4.1004,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796303369513019e-05,
      "loss": 4.1079,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795464774761967e-05,
      "loss": 4.1116,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794626180010915e-05,
      "loss": 4.1036,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793789223140236e-05,
      "loss": 4.0903,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792950628389184e-05,
      "loss": 4.0936,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792112033638132e-05,
      "loss": 4.0952,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79127343888708e-05,
      "loss": 4.1024,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790436482016401e-05,
      "loss": 4.0888,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789597887265349e-05,
      "loss": 4.0887,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788759292514297e-05,
      "loss": 4.0819,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787920697763245e-05,
      "loss": 4.1023,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870837408925664e-05,
      "loss": 4.0694,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862451461415144e-05,
      "loss": 4.0859,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785408189270836e-05,
      "loss": 4.0899,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784569594519783e-05,
      "loss": 4.0766,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783730999768731e-05,
      "loss": 4.0785,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782892405017679e-05,
      "loss": 4.0883,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782053810266627e-05,
      "loss": 4.0771,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781215215515575e-05,
      "loss": 4.0813,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780376620764523e-05,
      "loss": 4.0857,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779539663893844e-05,
      "loss": 4.0743,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778701069142792e-05,
      "loss": 4.0844,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77786247439174e-05,
      "loss": 4.0887,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777025517521062e-05,
      "loss": 4.0783,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77618692277001e-05,
      "loss": 4.0708,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775348328018958e-05,
      "loss": 4.0763,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774509733267906e-05,
      "loss": 4.0742,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773672776397227e-05,
      "loss": 4.0693,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7728341816461747e-05,
      "loss": 4.0664,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719955868951227e-05,
      "loss": 4.067,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7711569921440707e-05,
      "loss": 4.0826,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7703183973930186e-05,
      "loss": 4.0687,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694798026419666e-05,
      "loss": 4.069,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686412078909146e-05,
      "loss": 4.0597,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7678026131398626e-05,
      "loss": 4.0625,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766965656269184e-05,
      "loss": 4.0621,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766127061518132e-05,
      "loss": 4.0773,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76528846676708e-05,
      "loss": 4.072,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764449872016028e-05,
      "loss": 4.0486,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763612915145349e-05,
      "loss": 4.0585,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762774320394297e-05,
      "loss": 4.068,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761935725643245e-05,
      "loss": 4.0621,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761097130892193e-05,
      "loss": 4.0689,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760260174021514e-05,
      "loss": 4.0552,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759421579270462e-05,
      "loss": 4.0608,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75858298451941e-05,
      "loss": 4.0677,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757744389768358e-05,
      "loss": 4.0586,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7569074328976796e-05,
      "loss": 4.057,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7560688381466276e-05,
      "loss": 4.0601,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7552302433955756e-05,
      "loss": 4.0486,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7543932865248965e-05,
      "loss": 4.0497,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7535546917738445e-05,
      "loss": 4.0627,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527160970227925e-05,
      "loss": 4.045,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7518775022717405e-05,
      "loss": 4.0518,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510405454010614e-05,
      "loss": 4.0546,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7502019506500094e-05,
      "loss": 4.0508,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.050783157348633,
      "eval_runtime": 305.1552,
      "eval_samples_per_second": 1250.482,
      "eval_steps_per_second": 39.078,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.0565630839056384e+17,
  "trial_name": null,
  "trial_params": null
}