{
  "best_metric": 4.024856090545654,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-only/transformer/0/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": 11.0588,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8253,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.1936,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 5.9813,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.8216,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.713,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.6086,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.5394,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4501,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.4083,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3582,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.3178,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 5.252,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.2149,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.1821,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1327,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1165,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.0838,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0435,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0226,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.006,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 4.9758,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 4.9579,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.924,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9175,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 4.8957,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.8727,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8537,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.8275,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756873037409844e-05,
      "loss": 4.819,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.8016,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 4.7967,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 4.7803,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 4.7663,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.7574,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7454,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 4.7346,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.7047,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.7049,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.6774,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.6839,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.6639,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.6668,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6458,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 4.6426,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.6324,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6267,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.615,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 4.5899,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 4.5954,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 4.5974,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 4.5919,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 4.5707,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.5576,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954732262246923e-05,
      "loss": 4.5604,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953893667495871e-05,
      "loss": 4.5479,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953055072744819e-05,
      "loss": 4.5611,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952216477993767e-05,
      "loss": 4.5227,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.5403,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5301,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5091,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5199,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.503,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.495,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.4884,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 4.5,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 4.4795,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.4754,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4684,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 4.4614,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9413196598712105e-05,
      "loss": 4.4675,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.940482703000532e-05,
      "loss": 4.4724,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396441082494794e-05,
      "loss": 4.4502,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388055134984274e-05,
      "loss": 4.4592,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379669187473754e-05,
      "loss": 4.463,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371283239963234e-05,
      "loss": 4.4468,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936291367125644e-05,
      "loss": 4.4333,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935452772374592e-05,
      "loss": 4.4358,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93461417762354e-05,
      "loss": 4.4326,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933775582872488e-05,
      "loss": 4.4202,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93293862600181e-05,
      "loss": 4.4169,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932100031250758e-05,
      "loss": 4.4264,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931261436499706e-05,
      "loss": 4.427,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930422841748654e-05,
      "loss": 4.4087,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929584246997602e-05,
      "loss": 4.3873,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92874565224655e-05,
      "loss": 4.4003,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 4.4045,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927068462744446e-05,
      "loss": 4.3923,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.926231505873767e-05,
      "loss": 4.397,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925392911122715e-05,
      "loss": 4.3801,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924554316371663e-05,
      "loss": 4.3782,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923715721620611e-05,
      "loss": 4.3741,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922878764749932e-05,
      "loss": 4.3698,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92204016999888e-05,
      "loss": 4.3758,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921201575247828e-05,
      "loss": 4.366,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920362980496776e-05,
      "loss": 4.3804,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919524385745724e-05,
      "loss": 4.3638,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918685790994672e-05,
      "loss": 4.3481,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917848834123993e-05,
      "loss": 4.3505,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917010239372941e-05,
      "loss": 4.3578,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916171644621889e-05,
      "loss": 4.3477,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915333049870837e-05,
      "loss": 4.3465,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914494455119785e-05,
      "loss": 4.345,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913657498249106e-05,
      "loss": 4.3333,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912818903498054e-05,
      "loss": 4.3372,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911980308747002e-05,
      "loss": 4.3157,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91114171399595e-05,
      "loss": 4.3219,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910304757125272e-05,
      "loss": 4.3278,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90946616237422e-05,
      "loss": 4.3213,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908627567623168e-05,
      "loss": 4.3146,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907788972872115e-05,
      "loss": 4.3235,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906950378121063e-05,
      "loss": 4.299,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9061134212503846e-05,
      "loss": 4.3095,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9052748264993326e-05,
      "loss": 4.3043,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90443623174828e-05,
      "loss": 4.3085,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903597636997228e-05,
      "loss": 4.3151,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902759042246176e-05,
      "loss": 4.3037,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019220853754975e-05,
      "loss": 4.2971,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010834906244455e-05,
      "loss": 4.2927,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002448958733935e-05,
      "loss": 4.2986,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994063011223415e-05,
      "loss": 4.2912,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985693442516624e-05,
      "loss": 4.2895,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977307495006104e-05,
      "loss": 4.2823,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968921547495584e-05,
      "loss": 4.279,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960535599985064e-05,
      "loss": 4.2796,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952149652474544e-05,
      "loss": 4.2698,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943763704964024e-05,
      "loss": 4.2746,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893539413625723e-05,
      "loss": 4.2725,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892700818874671e-05,
      "loss": 4.2672,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891862224123619e-05,
      "loss": 4.2592,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891023629372567e-05,
      "loss": 4.2765,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890185034621515e-05,
      "loss": 4.2761,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889346439870464e-05,
      "loss": 4.2645,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888507845119412e-05,
      "loss": 4.2548,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887670888248733e-05,
      "loss": 4.2609,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886832293497681e-05,
      "loss": 4.2609,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885993698746629e-05,
      "loss": 4.2617,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885155103995577e-05,
      "loss": 4.2565,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.2554,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883479552373846e-05,
      "loss": 4.2497,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882640957622794e-05,
      "loss": 4.2492,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881802362871742e-05,
      "loss": 4.2419,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88096376812069e-05,
      "loss": 4.2454,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8801268112500106e-05,
      "loss": 4.239,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8792882164989586e-05,
      "loss": 4.2321,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878449621747907e-05,
      "loss": 4.2446,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877611026996855e-05,
      "loss": 4.231,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876772432245803e-05,
      "loss": 4.2397,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875933837494751e-05,
      "loss": 4.227,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8750952427436986e-05,
      "loss": 4.2219,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.193910598754883,
      "eval_runtime": 303.1016,
      "eval_samples_per_second": 1258.954,
      "eval_steps_per_second": 39.343,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8742566479926466e-05,
      "loss": 4.2132,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873419691121968e-05,
      "loss": 4.2137,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872581096370916e-05,
      "loss": 4.2258,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717425016198635e-05,
      "loss": 4.2145,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709039068688115e-05,
      "loss": 4.2196,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.870066949998133e-05,
      "loss": 4.2088,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.869228355247081e-05,
      "loss": 4.2042,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.868389760496029e-05,
      "loss": 4.1964,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675528036253507e-05,
      "loss": 4.2113,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667142088742986e-05,
      "loss": 4.2027,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865875614123246e-05,
      "loss": 4.2141,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865037019372194e-05,
      "loss": 4.1951,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8642000625015155e-05,
      "loss": 4.1929,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633614677504635e-05,
      "loss": 4.1871,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862522872999411e-05,
      "loss": 4.1823,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861684278248359e-05,
      "loss": 4.1827,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860845683497307e-05,
      "loss": 4.1908,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860007088746255e-05,
      "loss": 4.1794,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.1764,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583315371245244e-05,
      "loss": 4.1955,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574929423734724e-05,
      "loss": 4.1839,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566543476224204e-05,
      "loss": 4.1832,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558157528713684e-05,
      "loss": 4.1714,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.1842,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854140201249637e-05,
      "loss": 4.1696,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853301606498585e-05,
      "loss": 4.1701,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852463011747533e-05,
      "loss": 4.1693,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851624416996481e-05,
      "loss": 4.1616,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850785822245429e-05,
      "loss": 4.1558,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84994886537475e-05,
      "loss": 4.1558,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849110270623698e-05,
      "loss": 4.1648,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848271675872646e-05,
      "loss": 4.1633,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847433081121595e-05,
      "loss": 4.1571,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.1632,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845757529499864e-05,
      "loss": 4.1586,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844918934748812e-05,
      "loss": 4.1571,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84408033999776e-05,
      "loss": 4.145,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.1455,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842403150495656e-05,
      "loss": 4.1361,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841566193624977e-05,
      "loss": 4.1485,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840727598873925e-05,
      "loss": 4.1336,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839889004122873e-05,
      "loss": 4.1485,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839050409371821e-05,
      "loss": 4.134,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382118146207687e-05,
      "loss": 4.1381,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83737485775009e-05,
      "loss": 4.1384,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836536262999038e-05,
      "loss": 4.133,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835697668247986e-05,
      "loss": 4.1318,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834859073496934e-05,
      "loss": 4.1195,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834020478745882e-05,
      "loss": 4.1229,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833183521875203e-05,
      "loss": 4.1351,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832344927124151e-05,
      "loss": 4.1296,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831506332373099e-05,
      "loss": 4.1211,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830667737622047e-05,
      "loss": 4.1115,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829830780751368e-05,
      "loss": 4.1214,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828992186000316e-05,
      "loss": 4.1086,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828153591249264e-05,
      "loss": 4.1306,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827314996498212e-05,
      "loss": 4.1028,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264780396275336e-05,
      "loss": 4.1242,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256394448764816e-05,
      "loss": 4.1154,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248008501254296e-05,
      "loss": 4.1021,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823962255374377e-05,
      "loss": 4.1127,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823123660623325e-05,
      "loss": 4.1064,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222867037526465e-05,
      "loss": 4.098,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214481090015945e-05,
      "loss": 4.1004,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820609514250542e-05,
      "loss": 4.113,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81977091949949e-05,
      "loss": 4.1012,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189339626288114e-05,
      "loss": 4.0986,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180953678777594e-05,
      "loss": 4.0949,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172567731267074e-05,
      "loss": 4.092,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164181783756554e-05,
      "loss": 4.0999,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155795836246034e-05,
      "loss": 4.1128,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814742626753924e-05,
      "loss": 4.0921,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813904032002872e-05,
      "loss": 4.1084,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81306543725182e-05,
      "loss": 4.1072,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812226842500768e-05,
      "loss": 4.0968,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811389885630089e-05,
      "loss": 4.0871,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810551290879037e-05,
      "loss": 4.0993,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809712696127985e-05,
      "loss": 4.0944,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808874101376933e-05,
      "loss": 4.0855,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808035506625881e-05,
      "loss": 4.0875,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807198549755203e-05,
      "loss": 4.0929,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806359955004151e-05,
      "loss": 4.1,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.805521360253099e-05,
      "loss": 4.0875,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804682765502047e-05,
      "loss": 4.0731,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8038458086313677e-05,
      "loss": 4.0804,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030072138803156e-05,
      "loss": 4.0871,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021686191292636e-05,
      "loss": 4.0819,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8013316622585846e-05,
      "loss": 4.0868,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8004930675075325e-05,
      "loss": 4.0737,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996544727564805e-05,
      "loss": 4.0716,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7988158780054285e-05,
      "loss": 4.0715,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7979772832543765e-05,
      "loss": 4.0712,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7971386885033245e-05,
      "loss": 4.0763,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7963000937522725e-05,
      "loss": 4.0731,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795461499001221e-05,
      "loss": 4.0908,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794624542130542e-05,
      "loss": 4.0768,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79378594737949e-05,
      "loss": 4.0603,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792947352628438e-05,
      "loss": 4.0695,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792108757877386e-05,
      "loss": 4.0712,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791271801006707e-05,
      "loss": 4.0717,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790433206255655e-05,
      "loss": 4.069,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789594611504603e-05,
      "loss": 4.0666,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788756016753551e-05,
      "loss": 4.0592,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787919059882872e-05,
      "loss": 4.071,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78708046513182e-05,
      "loss": 4.0441,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786241870380768e-05,
      "loss": 4.0573,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7854032756297166e-05,
      "loss": 4.0568,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845663187590375e-05,
      "loss": 4.0589,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837277240079855e-05,
      "loss": 4.0492,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828891292569335e-05,
      "loss": 4.0605,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820505345058815e-05,
      "loss": 4.0452,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812135776352024e-05,
      "loss": 4.0514,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803749828841504e-05,
      "loss": 4.0515,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795363881330984e-05,
      "loss": 4.0531,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786977933820464e-05,
      "loss": 4.0668,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777860836511367e-05,
      "loss": 4.0566,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777023879640689e-05,
      "loss": 4.0522,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776185284889637e-05,
      "loss": 4.0446,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775346690138585e-05,
      "loss": 4.0511,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774508095387533e-05,
      "loss": 4.0494,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773669500636481e-05,
      "loss": 4.0468,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772830905885429e-05,
      "loss": 4.0436,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771992311134377e-05,
      "loss": 4.0399,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771153716383325e-05,
      "loss": 4.0425,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770316759512646e-05,
      "loss": 4.033,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769478164761594e-05,
      "loss": 4.0404,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768639570010542e-05,
      "loss": 4.0376,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76780097525949e-05,
      "loss": 4.0384,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669640183888106e-05,
      "loss": 4.027,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661254236377586e-05,
      "loss": 4.0472,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765286828886707e-05,
      "loss": 4.0524,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764448234135655e-05,
      "loss": 4.0356,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763611277264976e-05,
      "loss": 4.0313,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762774320394297e-05,
      "loss": 4.0388,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761935725643245e-05,
      "loss": 4.0405,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761097130892193e-05,
      "loss": 4.0412,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760258536141141e-05,
      "loss": 4.0365,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759419941390089e-05,
      "loss": 4.0363,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758581346639037e-05,
      "loss": 4.0307,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7577427518879844e-05,
      "loss": 4.0375,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7569041571369324e-05,
      "loss": 4.0252,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756067200266254e-05,
      "loss": 4.0352,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755228605515202e-05,
      "loss": 4.0311,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75439001076415e-05,
      "loss": 4.0195,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753551416013098e-05,
      "loss": 4.0369,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7527144591424196e-05,
      "loss": 4.0241,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751875864391367e-05,
      "loss": 4.0323,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751037269640315e-05,
      "loss": 4.0187,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750198674889263e-05,
      "loss": 4.0171,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.024856090545654,
      "eval_runtime": 302.4886,
      "eval_samples_per_second": 1261.505,
      "eval_steps_per_second": 39.423,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.0644055367242752e+17,
  "trial_name": null,
  "trial_params": null
}