{
  "best_metric": 4.026477336883545,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-sent-neg/transformer/2/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": 11.0274,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8127,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.1879,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 5.9751,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.8136,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.7053,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.5967,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5253,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4508,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.3976,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3482,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.305,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989936862987376e-05,
      "loss": 5.2626,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.1983,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.1694,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1287,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1044,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.076,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0502,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0154,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983231380739706e-05,
      "loss": 5.0069,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 4.9746,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 4.949,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.9317,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9225,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 4.8901,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.8696,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8548,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.8301,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756873037409844e-05,
      "loss": 4.824,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.8072,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740117521192533e-05,
      "loss": 4.8019,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731731573682013e-05,
      "loss": 4.7751,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 4.7681,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.7583,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7458,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 4.7305,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968980183612942e-05,
      "loss": 4.6997,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96814158886189e-05,
      "loss": 4.7022,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967302994110837e-05,
      "loss": 4.6785,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.6833,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.6691,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.6645,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6479,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 4.6453,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6408,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6266,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6306,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 4.5986,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 4.5918,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 4.6005,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 4.5893,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 4.5818,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.5532,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 4.5699,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 4.5471,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 4.5703,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 4.5256,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.5428,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5248,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5093,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5199,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.5068,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.4859,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.4898,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 4.5036,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 4.4808,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.4746,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4703,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9421582546222625e-05,
      "loss": 4.4572,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.47,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 4.4682,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 4.4562,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 4.4659,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 4.4531,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 4.4475,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 4.4295,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 4.4335,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346125397431674e-05,
      "loss": 4.4281,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 4.4189,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932936988121436e-05,
      "loss": 4.4166,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932098393370384e-05,
      "loss": 4.4235,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931259798619332e-05,
      "loss": 4.4205,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930421203868281e-05,
      "loss": 4.4125,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929582609117229e-05,
      "loss": 4.3901,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92874565224655e-05,
      "loss": 4.4021,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 4.4051,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927068462744446e-05,
      "loss": 4.4039,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.926229867993394e-05,
      "loss": 4.3885,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925391273242342e-05,
      "loss": 4.3976,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924554316371663e-05,
      "loss": 4.3851,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923715721620611e-05,
      "loss": 4.3757,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922877126869559e-05,
      "loss": 4.378,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922038532118507e-05,
      "loss": 4.3745,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921201575247828e-05,
      "loss": 4.3655,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920362980496776e-05,
      "loss": 4.3764,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919524385745724e-05,
      "loss": 4.3544,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918685790994672e-05,
      "loss": 4.3509,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91784719624362e-05,
      "loss": 4.3501,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 4.3611,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.3547,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.3384,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914494455119785e-05,
      "loss": 4.3399,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.3299,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128172656176805e-05,
      "loss": 4.3384,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119786708666285e-05,
      "loss": 4.3152,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911143351876323e-05,
      "loss": 4.3275,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910304757125272e-05,
      "loss": 4.3251,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90946616237422e-05,
      "loss": 4.3212,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908627567623168e-05,
      "loss": 4.3259,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907788972872115e-05,
      "loss": 4.3178,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906950378121063e-05,
      "loss": 4.3019,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906111783370011e-05,
      "loss": 4.3151,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9052748264993326e-05,
      "loss": 4.3145,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90443623174828e-05,
      "loss": 4.3081,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903597636997228e-05,
      "loss": 4.3097,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902759042246176e-05,
      "loss": 4.3067,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019220853754975e-05,
      "loss": 4.2998,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010834906244455e-05,
      "loss": 4.2918,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002448958733935e-05,
      "loss": 4.2937,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994063011223415e-05,
      "loss": 4.2919,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985677063712895e-05,
      "loss": 4.2802,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977291116202375e-05,
      "loss": 4.283,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968905168691855e-05,
      "loss": 4.2808,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960535599985064e-05,
      "loss": 4.2889,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952149652474544e-05,
      "loss": 4.2701,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943763704964024e-05,
      "loss": 4.2737,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935377757453504e-05,
      "loss": 4.2662,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892700818874671e-05,
      "loss": 4.2722,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891862224123619e-05,
      "loss": 4.2615,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891023629372567e-05,
      "loss": 4.2785,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890185034621515e-05,
      "loss": 4.2633,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889346439870464e-05,
      "loss": 4.2591,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888507845119412e-05,
      "loss": 4.2604,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88766925036836e-05,
      "loss": 4.2623,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886830655617308e-05,
      "loss": 4.2648,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885995336627002e-05,
      "loss": 4.2638,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515674187595e-05,
      "loss": 4.2564,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884318147124898e-05,
      "loss": 4.2571,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883479552373846e-05,
      "loss": 4.2533,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8826425955031666e-05,
      "loss": 4.2468,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8818040007521146e-05,
      "loss": 4.2377,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8809654060010626e-05,
      "loss": 4.2442,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8801268112500106e-05,
      "loss": 4.2371,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879289854379332e-05,
      "loss": 4.2296,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87845125962828e-05,
      "loss": 4.2341,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877612664877228e-05,
      "loss": 4.2306,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876774070126176e-05,
      "loss": 4.2357,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875937113255497e-05,
      "loss": 4.2357,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875098518504445e-05,
      "loss": 4.2269,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.195883750915527,
      "eval_runtime": 349.4311,
      "eval_samples_per_second": 1092.035,
      "eval_steps_per_second": 34.127,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874259923753393e-05,
      "loss": 4.21,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873421329002341e-05,
      "loss": 4.2122,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872582734251289e-05,
      "loss": 4.2253,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87174577738061e-05,
      "loss": 4.216,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870907182629558e-05,
      "loss": 4.2197,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870068587878506e-05,
      "loss": 4.2054,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869229993127454e-05,
      "loss": 4.2011,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683913983764027e-05,
      "loss": 4.2027,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675560793860965e-05,
      "loss": 4.2093,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667174846350445e-05,
      "loss": 4.2044,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8658788898839925e-05,
      "loss": 4.2021,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8650402951329405e-05,
      "loss": 4.2106,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8642017003818885e-05,
      "loss": 4.1792,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633631056308365e-05,
      "loss": 4.1883,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625245108797845e-05,
      "loss": 4.1854,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616859161287324e-05,
      "loss": 4.1753,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608473213776804e-05,
      "loss": 4.1893,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600103645070014e-05,
      "loss": 4.1821,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591717697559493e-05,
      "loss": 4.1764,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858333175004898e-05,
      "loss": 4.2025,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857494580253846e-05,
      "loss": 4.1802,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856657623383167e-05,
      "loss": 4.1769,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855819028632115e-05,
      "loss": 4.179,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854980433881063e-05,
      "loss": 4.1878,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854141839130011e-05,
      "loss": 4.1691,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853303244378958e-05,
      "loss": 4.1674,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85246628750828e-05,
      "loss": 4.1695,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851627692757228e-05,
      "loss": 4.1636,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850789098006175e-05,
      "loss": 4.1646,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849950503255123e-05,
      "loss": 4.1578,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849115184264818e-05,
      "loss": 4.1767,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848276589513766e-05,
      "loss": 4.1624,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847437994762714e-05,
      "loss": 4.161,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846599400011662e-05,
      "loss": 4.158,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84576080526061e-05,
      "loss": 4.1632,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844922210509558e-05,
      "loss": 4.1557,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440836157585056e-05,
      "loss": 4.1396,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432450210074536e-05,
      "loss": 4.1436,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8424064262564016e-05,
      "loss": 4.1352,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415694693857225e-05,
      "loss": 4.1487,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8407308746346705e-05,
      "loss": 4.1383,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8398922798836185e-05,
      "loss": 4.1467,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839053685132567e-05,
      "loss": 4.1383,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838216728261888e-05,
      "loss": 4.1385,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.837378133510836e-05,
      "loss": 4.1451,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836539538759784e-05,
      "loss": 4.1344,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835700944008732e-05,
      "loss": 4.1403,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834863987138053e-05,
      "loss": 4.127,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834025392387001e-05,
      "loss": 4.1252,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833186797635949e-05,
      "loss": 4.1329,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832348202884897e-05,
      "loss": 4.1299,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831511246014218e-05,
      "loss": 4.1353,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830672651263166e-05,
      "loss": 4.1064,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829834056512114e-05,
      "loss": 4.128,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289954617610625e-05,
      "loss": 4.108,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281585048903834e-05,
      "loss": 4.1378,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273199101393314e-05,
      "loss": 4.1065,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264813153882794e-05,
      "loss": 4.1226,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256427206372274e-05,
      "loss": 4.1144,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248057637665483e-05,
      "loss": 4.0984,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823967169015496e-05,
      "loss": 4.111,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823128574264444e-05,
      "loss": 4.1089,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.822289979513392e-05,
      "loss": 4.0913,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821453022642713e-05,
      "loss": 4.098,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820614427891661e-05,
      "loss": 4.1142,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819775833140609e-05,
      "loss": 4.1007,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818937238389558e-05,
      "loss": 4.0982,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818100281518879e-05,
      "loss": 4.0969,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817261686767827e-05,
      "loss": 4.0838,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816423092016775e-05,
      "loss": 4.1035,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815584497265723e-05,
      "loss": 4.1021,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814747540395044e-05,
      "loss": 4.1003,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813908945643992e-05,
      "loss": 4.1085,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81307035089294e-05,
      "loss": 4.0999,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812231756141888e-05,
      "loss": 4.1005,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113947992712086e-05,
      "loss": 4.0832,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105562045201566e-05,
      "loss": 4.0963,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8097176097691046e-05,
      "loss": 4.0883,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808879015018053e-05,
      "loss": 4.083,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808042058147374e-05,
      "loss": 4.0856,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807203463396322e-05,
      "loss": 4.0911,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80636486864527e-05,
      "loss": 4.097,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805526273894218e-05,
      "loss": 4.0897,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804687679143166e-05,
      "loss": 4.0711,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803850722272487e-05,
      "loss": 4.0766,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803012127521435e-05,
      "loss": 4.0924,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802173532770383e-05,
      "loss": 4.0881,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801334938019331e-05,
      "loss": 4.0807,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800497981148652e-05,
      "loss": 4.0869,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996610242779736e-05,
      "loss": 4.084,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7988224295269215e-05,
      "loss": 4.0693,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7979838347758695e-05,
      "loss": 4.0761,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7971452400248175e-05,
      "loss": 4.0793,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7963066452737655e-05,
      "loss": 4.0702,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7954696884030864e-05,
      "loss": 4.0856,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7946310936520344e-05,
      "loss": 4.0653,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7937924989009824e-05,
      "loss": 4.0622,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7929539041499304e-05,
      "loss": 4.0678,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7921153093988784e-05,
      "loss": 4.0754,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912767146478264e-05,
      "loss": 4.074,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7904381198967744e-05,
      "loss": 4.0623,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895995251457224e-05,
      "loss": 4.0597,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788762568275043e-05,
      "loss": 4.0541,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787923973523992e-05,
      "loss": 4.0717,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78708537877294e-05,
      "loss": 4.0459,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786248421902261e-05,
      "loss": 4.0525,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785409827151209e-05,
      "loss": 4.0661,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.784571232400157e-05,
      "loss": 4.058,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783734275529478e-05,
      "loss": 4.0561,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782895680778426e-05,
      "loss": 4.057,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782057086027374e-05,
      "loss": 4.0476,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.781218491276322e-05,
      "loss": 4.0591,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78037989652527e-05,
      "loss": 4.0539,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779541301774218e-05,
      "loss": 4.0557,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778702707023166e-05,
      "loss": 4.0579,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777864112272114e-05,
      "loss": 4.06,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777028793281808e-05,
      "loss": 4.0505,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776190198530756e-05,
      "loss": 4.0433,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775351603779704e-05,
      "loss": 4.0498,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774513009028652e-05,
      "loss": 4.0472,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7736744142776e-05,
      "loss": 4.0357,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772835819526548e-05,
      "loss": 4.0433,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719972247754956e-05,
      "loss": 4.0397,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771160267904817e-05,
      "loss": 4.0509,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770321673153765e-05,
      "loss": 4.0341,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769483078402713e-05,
      "loss": 4.043,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768644483651661e-05,
      "loss": 4.0243,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767807526780983e-05,
      "loss": 4.042,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766968932029931e-05,
      "loss": 4.0318,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766130337278878e-05,
      "loss": 4.05,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765291742527826e-05,
      "loss": 4.0403,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7644547856571476e-05,
      "loss": 4.0349,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636161909060956e-05,
      "loss": 4.0288,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762777596155043e-05,
      "loss": 4.0364,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761939001403991e-05,
      "loss": 4.0444,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7611020445333125e-05,
      "loss": 4.0452,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602634497822605e-05,
      "loss": 4.0398,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759424855031208e-05,
      "loss": 4.0392,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7585862602801565e-05,
      "loss": 4.0309,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757749303409478e-05,
      "loss": 4.0321,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7569107086584254e-05,
      "loss": 4.0244,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7560721139073734e-05,
      "loss": 4.0306,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7552335191563214e-05,
      "loss": 4.0252,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754398200166016e-05,
      "loss": 4.0176,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753559605414964e-05,
      "loss": 4.0285,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752721010663912e-05,
      "loss": 4.024,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75188241591286e-05,
      "loss": 4.0271,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751043821161808e-05,
      "loss": 4.0251,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750206864291129e-05,
      "loss": 4.0235,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.026477336883545,
      "eval_runtime": 327.3586,
      "eval_samples_per_second": 1165.667,
      "eval_steps_per_second": 36.428,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.055480250628178e+17,
  "trial_name": null,
  "trial_params": null
}