{
  "best_metric": 4.023011207580566,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/full2/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": 10.9422,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 6.8087,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 6.1703,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 5.9608,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 5.7996,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 5.6916,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 5.5901,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 5.523,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 5.4358,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992452647240532e-05,
      "loss": 5.3874,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99161405248948e-05,
      "loss": 5.3369,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990775457738428e-05,
      "loss": 5.314,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 5.2569,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 5.1946,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 5.1689,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 5.1321,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 5.1015,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.0742,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.0484,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.0067,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.0017,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9823927859886547e-05,
      "loss": 4.9672,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9815541912376026e-05,
      "loss": 4.9451,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 4.9183,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 4.9163,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790400448648195e-05,
      "loss": 4.8894,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9782014501137675e-05,
      "loss": 4.8614,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773628553627155e-05,
      "loss": 4.8452,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 4.8264,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756873037409844e-05,
      "loss": 4.8155,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 4.7933,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 4.7857,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 4.7723,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 4.7592,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 4.7553,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 4.7416,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969820416244367e-05,
      "loss": 4.7268,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968981821493315e-05,
      "loss": 4.7002,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 4.6953,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 4.6778,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 4.6755,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 4.66,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964788847738054e-05,
      "loss": 4.6553,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963950252987002e-05,
      "loss": 4.6391,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 4.6446,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 4.628,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 4.6295,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 4.6205,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 4.5924,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9589219602414374e-05,
      "loss": 4.5849,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958083365490385e-05,
      "loss": 4.5966,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 4.5827,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 4.5687,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 4.5534,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954732262246923e-05,
      "loss": 4.5586,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953893667495871e-05,
      "loss": 4.541,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953055072744819e-05,
      "loss": 4.5533,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.952216477993767e-05,
      "loss": 4.5209,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951377883242715e-05,
      "loss": 4.5315,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.950539288491663e-05,
      "loss": 4.5306,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 4.5065,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 4.5231,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94802514211888e-05,
      "loss": 4.4931,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947186547367828e-05,
      "loss": 4.4943,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946347952616776e-05,
      "loss": 4.4831,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 4.499,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944670763114672e-05,
      "loss": 4.4698,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 4.4754,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 4.4598,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 4.4567,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941318021990837e-05,
      "loss": 4.4699,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404810651201585e-05,
      "loss": 4.4622,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396424703691065e-05,
      "loss": 4.4519,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388038756180545e-05,
      "loss": 4.4515,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379652808670025e-05,
      "loss": 4.4598,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371283239963234e-05,
      "loss": 4.4314,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.936291367125644e-05,
      "loss": 4.4479,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935452772374592e-05,
      "loss": 4.4221,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.93461417762354e-05,
      "loss": 4.4205,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.933775582872488e-05,
      "loss": 4.4192,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932936988121436e-05,
      "loss": 4.4218,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932098393370384e-05,
      "loss": 4.4141,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931259798619332e-05,
      "loss": 4.4203,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930422841748654e-05,
      "loss": 4.405,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.929584246997602e-05,
      "loss": 4.3855,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92874565224655e-05,
      "loss": 4.3888,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927907057495498e-05,
      "loss": 4.3976,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.927068462744446e-05,
      "loss": 4.3846,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.92623314375414e-05,
      "loss": 4.384,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925394549003088e-05,
      "loss": 4.3887,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924555954252036e-05,
      "loss": 4.3706,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923717359500984e-05,
      "loss": 4.3766,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9228804026303046e-05,
      "loss": 4.3697,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922041807879253e-05,
      "loss": 4.3729,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921203213128201e-05,
      "loss": 4.3717,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920364618377149e-05,
      "loss": 4.3673,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919526023626097e-05,
      "loss": 4.3456,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918687428875045e-05,
      "loss": 4.3532,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917848834123993e-05,
      "loss": 4.3469,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917010239372941e-05,
      "loss": 4.3496,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.916171644621889e-05,
      "loss": 4.3527,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915333049870837e-05,
      "loss": 4.3321,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914494455119785e-05,
      "loss": 4.3376,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136558603687325e-05,
      "loss": 4.3388,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912818903498054e-05,
      "loss": 4.331,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911980308747002e-05,
      "loss": 4.3203,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911143351876323e-05,
      "loss": 4.3241,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.910304757125272e-05,
      "loss": 4.3132,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90946616237422e-05,
      "loss": 4.3117,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908627567623168e-05,
      "loss": 4.3103,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907788972872115e-05,
      "loss": 4.3093,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906950378121063e-05,
      "loss": 4.3054,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906111783370011e-05,
      "loss": 4.3098,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905273188618959e-05,
      "loss": 4.3007,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90443623174828e-05,
      "loss": 4.3037,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903597636997228e-05,
      "loss": 4.3106,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9027606801265495e-05,
      "loss": 4.3055,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9019220853754975e-05,
      "loss": 4.3007,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9010834906244455e-05,
      "loss": 4.2906,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9002448958733935e-05,
      "loss": 4.2898,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8994063011223415e-05,
      "loss": 4.2885,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985677063712895e-05,
      "loss": 4.276,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977307495006104e-05,
      "loss": 4.2818,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968921547495584e-05,
      "loss": 4.2757,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960535599985064e-05,
      "loss": 4.2805,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952149652474544e-05,
      "loss": 4.2717,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943763704964024e-05,
      "loss": 4.2796,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8935377757453504e-05,
      "loss": 4.2732,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926991809942984e-05,
      "loss": 4.2653,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891862224123619e-05,
      "loss": 4.2604,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.891023629372567e-05,
      "loss": 4.2761,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890185034621515e-05,
      "loss": 4.2671,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.889346439870464e-05,
      "loss": 4.2593,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888507845119412e-05,
      "loss": 4.261,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887670888248733e-05,
      "loss": 4.2641,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886832293497681e-05,
      "loss": 4.2588,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885993698746629e-05,
      "loss": 4.2614,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885155103995577e-05,
      "loss": 4.2507,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884316509244525e-05,
      "loss": 4.2497,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883477914493473e-05,
      "loss": 4.2491,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882639319742421e-05,
      "loss": 4.2409,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881800724991369e-05,
      "loss": 4.2378,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88096376812069e-05,
      "loss": 4.2372,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880125173369638e-05,
      "loss": 4.2402,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879286578618586e-05,
      "loss": 4.231,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.878449621747907e-05,
      "loss": 4.2402,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877611026996855e-05,
      "loss": 4.2315,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876772432245803e-05,
      "loss": 4.2336,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875933837494751e-05,
      "loss": 4.2379,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8750952427436986e-05,
      "loss": 4.22,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.190673351287842,
      "eval_runtime": 313.8813,
      "eval_samples_per_second": 1215.718,
      "eval_steps_per_second": 37.992,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87425828587302e-05,
      "loss": 4.2087,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873419691121968e-05,
      "loss": 4.2056,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.872581096370916e-05,
      "loss": 4.2306,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8717425016198635e-05,
      "loss": 4.213,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709039068688115e-05,
      "loss": 4.2179,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700653121177595e-05,
      "loss": 4.2089,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692267173667075e-05,
      "loss": 4.2014,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683881226156555e-05,
      "loss": 4.1951,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675528036253507e-05,
      "loss": 4.2024,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667142088742986e-05,
      "loss": 4.195,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865875614123246e-05,
      "loss": 4.2182,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865037019372194e-05,
      "loss": 4.204,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864198424621142e-05,
      "loss": 4.1856,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86335982987009e-05,
      "loss": 4.1869,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862521235119038e-05,
      "loss": 4.1886,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.861682640367986e-05,
      "loss": 4.1748,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860845683497307e-05,
      "loss": 4.1901,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.860007088746255e-05,
      "loss": 4.1849,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.859168493995203e-05,
      "loss": 4.1734,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858329899244151e-05,
      "loss": 4.1979,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574913044930995e-05,
      "loss": 4.1772,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8566543476224204e-05,
      "loss": 4.1787,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558157528713684e-05,
      "loss": 4.1701,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549771581203164e-05,
      "loss": 4.1879,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541385633692644e-05,
      "loss": 4.1655,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853301606498585e-05,
      "loss": 4.1663,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852463011747533e-05,
      "loss": 4.1633,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851624416996481e-05,
      "loss": 4.1611,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850785822245429e-05,
      "loss": 4.1591,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849947227494377e-05,
      "loss": 4.1589,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849110270623698e-05,
      "loss": 4.1608,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848271675872646e-05,
      "loss": 4.1613,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847433081121595e-05,
      "loss": 4.1582,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846594486370543e-05,
      "loss": 4.1634,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845757529499864e-05,
      "loss": 4.1562,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844918934748812e-05,
      "loss": 4.156,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.84408033999776e-05,
      "loss": 4.1427,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843241745246708e-05,
      "loss": 4.1452,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842403150495656e-05,
      "loss": 4.1361,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841566193624977e-05,
      "loss": 4.1428,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840727598873925e-05,
      "loss": 4.1361,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839889004122873e-05,
      "loss": 4.1453,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839050409371821e-05,
      "loss": 4.1273,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8382118146207687e-05,
      "loss": 4.145,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83737485775009e-05,
      "loss": 4.1376,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836536262999038e-05,
      "loss": 4.1396,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.835697668247986e-05,
      "loss": 4.1357,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834859073496934e-05,
      "loss": 4.119,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834022116626255e-05,
      "loss": 4.1209,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833183521875203e-05,
      "loss": 4.1367,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832344927124151e-05,
      "loss": 4.1266,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831506332373099e-05,
      "loss": 4.1237,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83066937550242e-05,
      "loss": 4.1132,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829832418631741e-05,
      "loss": 4.1247,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828993823880689e-05,
      "loss": 4.1041,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828155229129637e-05,
      "loss": 4.1307,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273166343785856e-05,
      "loss": 4.1008,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8264780396275336e-05,
      "loss": 4.1183,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8256394448764816e-05,
      "loss": 4.1165,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8248008501254296e-05,
      "loss": 4.1029,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823962255374377e-05,
      "loss": 4.1181,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8231252985036985e-05,
      "loss": 4.1046,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222867037526465e-05,
      "loss": 4.1037,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214481090015945e-05,
      "loss": 4.0921,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820609514250542e-05,
      "loss": 4.1169,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197725573798634e-05,
      "loss": 4.09,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189339626288114e-05,
      "loss": 4.1017,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180953678777594e-05,
      "loss": 4.0822,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172567731267074e-05,
      "loss": 4.095,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816419816256029e-05,
      "loss": 4.1034,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815581221504977e-05,
      "loss": 4.1029,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814742626753924e-05,
      "loss": 4.0965,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813904032002872e-05,
      "loss": 4.1001,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813067075132194e-05,
      "loss": 4.1053,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812228480381141e-05,
      "loss": 4.0903,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811389885630089e-05,
      "loss": 4.0999,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810551290879037e-05,
      "loss": 4.0867,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809714334008359e-05,
      "loss": 4.0878,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808875739257306e-05,
      "loss": 4.0841,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808037144506255e-05,
      "loss": 4.0923,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.807198549755203e-05,
      "loss": 4.0923,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063615928845243e-05,
      "loss": 4.0955,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055229981334717e-05,
      "loss": 4.0859,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046844033824197e-05,
      "loss": 4.0665,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8038458086313677e-05,
      "loss": 4.0713,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8030088517606886e-05,
      "loss": 4.0825,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8021702570096366e-05,
      "loss": 4.0759,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801333300138958e-05,
      "loss": 4.0736,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800494705387906e-05,
      "loss": 4.0844,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7996561106368535e-05,
      "loss": 4.0675,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7988175158858015e-05,
      "loss": 4.0709,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7979789211347494e-05,
      "loss": 4.0735,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797140326383698e-05,
      "loss": 4.0799,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796301731632646e-05,
      "loss": 4.0779,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795464774761967e-05,
      "loss": 4.0819,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794626180010915e-05,
      "loss": 4.057,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793787585259863e-05,
      "loss": 4.0662,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792948990508811e-05,
      "loss": 4.0647,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792110395757759e-05,
      "loss": 4.0667,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791271801006707e-05,
      "loss": 4.0775,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790433206255655e-05,
      "loss": 4.0567,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789594611504603e-05,
      "loss": 4.0598,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788757654633924e-05,
      "loss": 4.0676,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787919059882872e-05,
      "loss": 4.0645,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78708046513182e-05,
      "loss": 4.0471,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862435082611415e-05,
      "loss": 4.0611,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7854049135100895e-05,
      "loss": 4.0514,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845663187590375e-05,
      "loss": 4.0493,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837277240079855e-05,
      "loss": 4.0486,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7828891292569335e-05,
      "loss": 4.0494,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7820505345058815e-05,
      "loss": 4.0466,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7812119397548295e-05,
      "loss": 4.0576,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7803733450037775e-05,
      "loss": 4.0434,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7795363881330984e-05,
      "loss": 4.0529,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7786977933820464e-05,
      "loss": 4.0626,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7778591986309944e-05,
      "loss": 4.0558,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7770206038799423e-05,
      "loss": 4.0535,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776183647009263e-05,
      "loss": 4.0486,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775345052258212e-05,
      "loss": 4.0447,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77450645750716e-05,
      "loss": 4.0469,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773667862756108e-05,
      "loss": 4.0384,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772830905885429e-05,
      "loss": 4.0395,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771992311134377e-05,
      "loss": 4.0396,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771153716383325e-05,
      "loss": 4.045,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.770315121632272e-05,
      "loss": 4.0415,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.76947652688122e-05,
      "loss": 4.0417,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768637932130168e-05,
      "loss": 4.0424,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767799337379116e-05,
      "loss": 4.0324,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766962380508437e-05,
      "loss": 4.0323,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766123785757386e-05,
      "loss": 4.0522,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765285191006334e-05,
      "loss": 4.0438,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764446596255282e-05,
      "loss": 4.0392,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636096393846026e-05,
      "loss": 4.0335,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7627710446335506e-05,
      "loss": 4.0396,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7619324498824986e-05,
      "loss": 4.0358,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7610938551314466e-05,
      "loss": 4.0433,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7602568982607675e-05,
      "loss": 4.0352,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7594183035097155e-05,
      "loss": 4.0313,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7585797087586635e-05,
      "loss": 4.032,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7577411140076115e-05,
      "loss": 4.03,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7569025192565595e-05,
      "loss": 4.0221,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7560639245055075e-05,
      "loss": 4.0282,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7552253297544555e-05,
      "loss": 4.0327,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754388372883777e-05,
      "loss": 4.0172,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753549778132725e-05,
      "loss": 4.0351,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752711183381673e-05,
      "loss": 4.0276,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751872588630621e-05,
      "loss": 4.0239,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751035631759942e-05,
      "loss": 4.0298,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75019703700889e-05,
      "loss": 4.0167,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.023011207580566,
      "eval_runtime": 312.5224,
      "eval_samples_per_second": 1221.004,
      "eval_steps_per_second": 38.157,
      "step": 152640
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 1.0623358555226726e+17,
  "trial_name": null,
  "trial_params": null
}