{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1290,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.023255813953488372,
      "grad_norm": 2.1102249816437895,
      "learning_rate": 2.0930232558139536e-06,
      "loss": 0.4502,
      "step": 10
    },
    {
      "epoch": 0.046511627906976744,
      "grad_norm": 0.6087936521314388,
      "learning_rate": 4.418604651162791e-06,
      "loss": 0.3222,
      "step": 20
    },
    {
      "epoch": 0.06976744186046512,
      "grad_norm": 0.5276269917240805,
      "learning_rate": 6.744186046511628e-06,
      "loss": 0.2413,
      "step": 30
    },
    {
      "epoch": 0.09302325581395349,
      "grad_norm": 0.33927814937972933,
      "learning_rate": 9.069767441860465e-06,
      "loss": 0.2011,
      "step": 40
    },
    {
      "epoch": 0.11627906976744186,
      "grad_norm": 0.41749972892158266,
      "learning_rate": 1.1395348837209304e-05,
      "loss": 0.1824,
      "step": 50
    },
    {
      "epoch": 0.13953488372093023,
      "grad_norm": 0.3467307268335015,
      "learning_rate": 1.3720930232558139e-05,
      "loss": 0.1777,
      "step": 60
    },
    {
      "epoch": 0.16279069767441862,
      "grad_norm": 0.2909147698009027,
      "learning_rate": 1.6046511627906977e-05,
      "loss": 0.1641,
      "step": 70
    },
    {
      "epoch": 0.18604651162790697,
      "grad_norm": 0.6246257892448243,
      "learning_rate": 1.8372093023255815e-05,
      "loss": 0.1515,
      "step": 80
    },
    {
      "epoch": 0.20930232558139536,
      "grad_norm": 0.35908006141784865,
      "learning_rate": 2.069767441860465e-05,
      "loss": 0.1491,
      "step": 90
    },
    {
      "epoch": 0.23255813953488372,
      "grad_norm": 0.27603388668684564,
      "learning_rate": 2.302325581395349e-05,
      "loss": 0.1481,
      "step": 100
    },
    {
      "epoch": 0.2558139534883721,
      "grad_norm": 0.2527784608738212,
      "learning_rate": 2.5348837209302324e-05,
      "loss": 0.1397,
      "step": 110
    },
    {
      "epoch": 0.27906976744186046,
      "grad_norm": 0.29932144986900905,
      "learning_rate": 2.7674418604651162e-05,
      "loss": 0.1404,
      "step": 120
    },
    {
      "epoch": 0.3023255813953488,
      "grad_norm": 0.3393202145121213,
      "learning_rate": 3e-05,
      "loss": 0.142,
      "step": 130
    },
    {
      "epoch": 0.32558139534883723,
      "grad_norm": 0.24298726842558038,
      "learning_rate": 2.999450876449861e-05,
      "loss": 0.1353,
      "step": 140
    },
    {
      "epoch": 0.3488372093023256,
      "grad_norm": 0.17700673786533003,
      "learning_rate": 2.9978039078483434e-05,
      "loss": 0.1316,
      "step": 150
    },
    {
      "epoch": 0.37209302325581395,
      "grad_norm": 0.23624869484478003,
      "learning_rate": 2.9950603000477722e-05,
      "loss": 0.1296,
      "step": 160
    },
    {
      "epoch": 0.3953488372093023,
      "grad_norm": 0.2450597575472511,
      "learning_rate": 2.991222061821024e-05,
      "loss": 0.1322,
      "step": 170
    },
    {
      "epoch": 0.4186046511627907,
      "grad_norm": 0.2629976539114997,
      "learning_rate": 2.9862920033907654e-05,
      "loss": 0.1272,
      "step": 180
    },
    {
      "epoch": 0.4418604651162791,
      "grad_norm": 0.158909112774518,
      "learning_rate": 2.980273734371914e-05,
      "loss": 0.1265,
      "step": 190
    },
    {
      "epoch": 0.46511627906976744,
      "grad_norm": 0.16963032221578891,
      "learning_rate": 2.9731716611288024e-05,
      "loss": 0.1275,
      "step": 200
    },
    {
      "epoch": 0.4883720930232558,
      "grad_norm": 0.207022304132148,
      "learning_rate": 2.9649909835489935e-05,
      "loss": 0.1262,
      "step": 210
    },
    {
      "epoch": 0.5116279069767442,
      "grad_norm": 0.18726792852194557,
      "learning_rate": 2.955737691236108e-05,
      "loss": 0.1198,
      "step": 220
    },
    {
      "epoch": 0.5348837209302325,
      "grad_norm": 0.1701049540713391,
      "learning_rate": 2.945418559124446e-05,
      "loss": 0.1209,
      "step": 230
    },
    {
      "epoch": 0.5581395348837209,
      "grad_norm": 0.19958209659130646,
      "learning_rate": 2.9340411425186207e-05,
      "loss": 0.122,
      "step": 240
    },
    {
      "epoch": 0.5813953488372093,
      "grad_norm": 0.22288049794980888,
      "learning_rate": 2.921613771561829e-05,
      "loss": 0.1155,
      "step": 250
    },
    {
      "epoch": 0.6046511627906976,
      "grad_norm": 0.17056830281402427,
      "learning_rate": 2.908145545136816e-05,
      "loss": 0.1123,
      "step": 260
    },
    {
      "epoch": 0.627906976744186,
      "grad_norm": 0.1726845936656448,
      "learning_rate": 2.8936463242039924e-05,
      "loss": 0.1139,
      "step": 270
    },
    {
      "epoch": 0.6511627906976745,
      "grad_norm": 0.17997669987525944,
      "learning_rate": 2.8781267245815898e-05,
      "loss": 0.1085,
      "step": 280
    },
    {
      "epoch": 0.6744186046511628,
      "grad_norm": 0.19695001871413667,
      "learning_rate": 2.8615981091731296e-05,
      "loss": 0.113,
      "step": 290
    },
    {
      "epoch": 0.6976744186046512,
      "grad_norm": 0.19008651260372486,
      "learning_rate": 2.8440725796479075e-05,
      "loss": 0.1172,
      "step": 300
    },
    {
      "epoch": 0.7209302325581395,
      "grad_norm": 0.18432725372928307,
      "learning_rate": 2.825562967580579e-05,
      "loss": 0.113,
      "step": 310
    },
    {
      "epoch": 0.7441860465116279,
      "grad_norm": 0.14972827867736163,
      "learning_rate": 2.8060828250563294e-05,
      "loss": 0.1113,
      "step": 320
    },
    {
      "epoch": 0.7674418604651163,
      "grad_norm": 0.1708006253644624,
      "learning_rate": 2.7856464147485202e-05,
      "loss": 0.1118,
      "step": 330
    },
    {
      "epoch": 0.7906976744186046,
      "grad_norm": 0.2545153275078911,
      "learning_rate": 2.764268699476058e-05,
      "loss": 0.1126,
      "step": 340
    },
    {
      "epoch": 0.813953488372093,
      "grad_norm": 0.13582442076029375,
      "learning_rate": 2.7419653312481482e-05,
      "loss": 0.1084,
      "step": 350
    },
    {
      "epoch": 0.8372093023255814,
      "grad_norm": 0.1885021789487428,
      "learning_rate": 2.7187526398044463e-05,
      "loss": 0.1177,
      "step": 360
    },
    {
      "epoch": 0.8604651162790697,
      "grad_norm": 0.18897101004109837,
      "learning_rate": 2.6946476206589972e-05,
      "loss": 0.1044,
      "step": 370
    },
    {
      "epoch": 0.8837209302325582,
      "grad_norm": 0.12704249962452174,
      "learning_rate": 2.6696679226567202e-05,
      "loss": 0.1099,
      "step": 380
    },
    {
      "epoch": 0.9069767441860465,
      "grad_norm": 0.19659069921161837,
      "learning_rate": 2.6438318350515467e-05,
      "loss": 0.1084,
      "step": 390
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 0.17360269942962656,
      "learning_rate": 2.617158274115673e-05,
      "loss": 0.1086,
      "step": 400
    },
    {
      "epoch": 0.9534883720930233,
      "grad_norm": 0.15978057253464667,
      "learning_rate": 2.5896667692897334e-05,
      "loss": 0.098,
      "step": 410
    },
    {
      "epoch": 0.9767441860465116,
      "grad_norm": 0.12225147895714812,
      "learning_rate": 2.5613774488840333e-05,
      "loss": 0.1083,
      "step": 420
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.11136831453413676,
      "learning_rate": 2.532311025341309e-05,
      "loss": 0.1064,
      "step": 430
    },
    {
      "epoch": 1.0232558139534884,
      "grad_norm": 0.11500268275827884,
      "learning_rate": 2.502488780071807e-05,
      "loss": 0.1053,
      "step": 440
    },
    {
      "epoch": 1.0465116279069768,
      "grad_norm": 0.13683341482554096,
      "learning_rate": 2.4719325478717893e-05,
      "loss": 0.103,
      "step": 450
    },
    {
      "epoch": 1.069767441860465,
      "grad_norm": 0.1391528532587655,
      "learning_rate": 2.440664700936861e-05,
      "loss": 0.102,
      "step": 460
    },
    {
      "epoch": 1.0930232558139534,
      "grad_norm": 0.1591189829381333,
      "learning_rate": 2.408708132481842e-05,
      "loss": 0.1017,
      "step": 470
    },
    {
      "epoch": 1.1162790697674418,
      "grad_norm": 0.10083202954080314,
      "learning_rate": 2.376086239979158e-05,
      "loss": 0.1047,
      "step": 480
    },
    {
      "epoch": 1.1395348837209303,
      "grad_norm": 0.2036430998599195,
      "learning_rate": 2.3428229080280407e-05,
      "loss": 0.1018,
      "step": 490
    },
    {
      "epoch": 1.1627906976744187,
      "grad_norm": 0.15460019789887855,
      "learning_rate": 2.3089424908670642e-05,
      "loss": 0.1006,
      "step": 500
    },
    {
      "epoch": 1.1860465116279069,
      "grad_norm": 0.1171523300082374,
      "learning_rate": 2.2744697945428307e-05,
      "loss": 0.1007,
      "step": 510
    },
    {
      "epoch": 1.2093023255813953,
      "grad_norm": 0.19738915842381824,
      "learning_rate": 2.2394300587478566e-05,
      "loss": 0.0972,
      "step": 520
    },
    {
      "epoch": 1.2325581395348837,
      "grad_norm": 0.16778096884891686,
      "learning_rate": 2.2038489383409652e-05,
      "loss": 0.1027,
      "step": 530
    },
    {
      "epoch": 1.255813953488372,
      "grad_norm": 0.13067156491026763,
      "learning_rate": 2.167752484563696e-05,
      "loss": 0.0997,
      "step": 540
    },
    {
      "epoch": 1.2790697674418605,
      "grad_norm": 0.16129671511714355,
      "learning_rate": 2.13116712596651e-05,
      "loss": 0.1025,
      "step": 550
    },
    {
      "epoch": 1.302325581395349,
      "grad_norm": 0.15785271361590572,
      "learning_rate": 2.0941196490587352e-05,
      "loss": 0.1002,
      "step": 560
    },
    {
      "epoch": 1.3255813953488373,
      "grad_norm": 0.13841764309679408,
      "learning_rate": 2.05663717869643e-05,
      "loss": 0.0994,
      "step": 570
    },
    {
      "epoch": 1.3488372093023255,
      "grad_norm": 0.17509131465232344,
      "learning_rate": 2.0187471582225173e-05,
      "loss": 0.0993,
      "step": 580
    },
    {
      "epoch": 1.372093023255814,
      "grad_norm": 0.1221451212745489,
      "learning_rate": 1.9804773293737416e-05,
      "loss": 0.0998,
      "step": 590
    },
    {
      "epoch": 1.3953488372093024,
      "grad_norm": 0.14688306712971377,
      "learning_rate": 1.9418557119691434e-05,
      "loss": 0.0988,
      "step": 600
    },
    {
      "epoch": 1.4186046511627908,
      "grad_norm": 0.10658674461240546,
      "learning_rate": 1.902910583394938e-05,
      "loss": 0.0994,
      "step": 610
    },
    {
      "epoch": 1.441860465116279,
      "grad_norm": 0.11935538384825148,
      "learning_rate": 1.8636704579008096e-05,
      "loss": 0.0985,
      "step": 620
    },
    {
      "epoch": 1.4651162790697674,
      "grad_norm": 0.12721218184322167,
      "learning_rate": 1.824164065722783e-05,
      "loss": 0.0994,
      "step": 630
    },
    {
      "epoch": 1.4883720930232558,
      "grad_norm": 0.09137892917036859,
      "learning_rate": 1.7844203320479614e-05,
      "loss": 0.1047,
      "step": 640
    },
    {
      "epoch": 1.5116279069767442,
      "grad_norm": 0.23254378397439393,
      "learning_rate": 1.7444683558365182e-05,
      "loss": 0.0959,
      "step": 650
    },
    {
      "epoch": 1.5348837209302326,
      "grad_norm": 0.12936296282790546,
      "learning_rate": 1.7043373885164703e-05,
      "loss": 0.0968,
      "step": 660
    },
    {
      "epoch": 1.558139534883721,
      "grad_norm": 0.15526942233091734,
      "learning_rate": 1.664056812566812e-05,
      "loss": 0.1004,
      "step": 670
    },
    {
      "epoch": 1.5813953488372094,
      "grad_norm": 0.12963665517042525,
      "learning_rate": 1.623656120004698e-05,
      "loss": 0.0921,
      "step": 680
    },
    {
      "epoch": 1.6046511627906976,
      "grad_norm": 0.11191785293454326,
      "learning_rate": 1.5831648907924337e-05,
      "loss": 0.0969,
      "step": 690
    },
    {
      "epoch": 1.627906976744186,
      "grad_norm": 0.13773389105774028,
      "learning_rate": 1.5426127711800636e-05,
      "loss": 0.0994,
      "step": 700
    },
    {
      "epoch": 1.6511627906976745,
      "grad_norm": 0.10576569528021193,
      "learning_rate": 1.5020294519994381e-05,
      "loss": 0.0957,
      "step": 710
    },
    {
      "epoch": 1.6744186046511627,
      "grad_norm": 0.1437216064703046,
      "learning_rate": 1.4614446469256305e-05,
      "loss": 0.0945,
      "step": 720
    },
    {
      "epoch": 1.697674418604651,
      "grad_norm": 0.10803257779082216,
      "learning_rate": 1.4208880707216323e-05,
      "loss": 0.095,
      "step": 730
    },
    {
      "epoch": 1.7209302325581395,
      "grad_norm": 0.12123283799800175,
      "learning_rate": 1.3803894174822518e-05,
      "loss": 0.0944,
      "step": 740
    },
    {
      "epoch": 1.744186046511628,
      "grad_norm": 0.100414996108917,
      "learning_rate": 1.3399783388931468e-05,
      "loss": 0.0954,
      "step": 750
    },
    {
      "epoch": 1.7674418604651163,
      "grad_norm": 0.1369011425168964,
      "learning_rate": 1.2996844225209033e-05,
      "loss": 0.0984,
      "step": 760
    },
    {
      "epoch": 1.7906976744186047,
      "grad_norm": 0.1359084312900855,
      "learning_rate": 1.2595371701500639e-05,
      "loss": 0.0962,
      "step": 770
    },
    {
      "epoch": 1.8139534883720931,
      "grad_norm": 0.13125777727884955,
      "learning_rate": 1.219565976182963e-05,
      "loss": 0.0972,
      "step": 780
    },
    {
      "epoch": 1.8372093023255816,
      "grad_norm": 0.13341223023541443,
      "learning_rate": 1.1798001061181799e-05,
      "loss": 0.0952,
      "step": 790
    },
    {
      "epoch": 1.8604651162790697,
      "grad_norm": 0.08587984083521312,
      "learning_rate": 1.1402686751233723e-05,
      "loss": 0.0969,
      "step": 800
    },
    {
      "epoch": 1.8837209302325582,
      "grad_norm": 0.11630108959701958,
      "learning_rate": 1.101000626718182e-05,
      "loss": 0.0915,
      "step": 810
    },
    {
      "epoch": 1.9069767441860463,
      "grad_norm": 0.0951067714002941,
      "learning_rate": 1.0620247115828044e-05,
      "loss": 0.0966,
      "step": 820
    },
    {
      "epoch": 1.9302325581395348,
      "grad_norm": 0.10789836511240755,
      "learning_rate": 1.0233694665077584e-05,
      "loss": 0.0946,
      "step": 830
    },
    {
      "epoch": 1.9534883720930232,
      "grad_norm": 0.10824454265797709,
      "learning_rate": 9.850631935002531e-06,
      "loss": 0.0956,
      "step": 840
    },
    {
      "epoch": 1.9767441860465116,
      "grad_norm": 0.10283011375161531,
      "learning_rate": 9.471339390624574e-06,
      "loss": 0.0949,
      "step": 850
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.15809185521039087,
      "learning_rate": 9.0960947365684e-06,
      "loss": 0.0934,
      "step": 860
    },
    {
      "epoch": 2.0232558139534884,
      "grad_norm": 0.15384352891044772,
      "learning_rate": 8.725172713736136e-06,
      "loss": 0.0889,
      "step": 870
    },
    {
      "epoch": 2.046511627906977,
      "grad_norm": 0.11467341711361749,
      "learning_rate": 8.358844898151791e-06,
      "loss": 0.0882,
      "step": 880
    },
    {
      "epoch": 2.0697674418604652,
      "grad_norm": 0.10965841387662369,
      "learning_rate": 7.997379502122849e-06,
      "loss": 0.0944,
      "step": 890
    },
    {
      "epoch": 2.0930232558139537,
      "grad_norm": 0.1024892604780862,
      "learning_rate": 7.641041177864661e-06,
      "loss": 0.0892,
      "step": 900
    },
    {
      "epoch": 2.116279069767442,
      "grad_norm": 0.14766897087136052,
      "learning_rate": 7.290090823731452e-06,
      "loss": 0.0904,
      "step": 910
    },
    {
      "epoch": 2.13953488372093,
      "grad_norm": 0.11551061066180639,
      "learning_rate": 6.944785393195742e-06,
      "loss": 0.0933,
      "step": 920
    },
    {
      "epoch": 2.1627906976744184,
      "grad_norm": 0.1307266761701476,
      "learning_rate": 6.605377706716049e-06,
      "loss": 0.0866,
      "step": 930
    },
    {
      "epoch": 2.186046511627907,
      "grad_norm": 0.10681432279549767,
      "learning_rate": 6.27211626663071e-06,
      "loss": 0.0869,
      "step": 940
    },
    {
      "epoch": 2.2093023255813953,
      "grad_norm": 0.1471123103304021,
      "learning_rate": 5.945245075213187e-06,
      "loss": 0.0858,
      "step": 950
    },
    {
      "epoch": 2.2325581395348837,
      "grad_norm": 0.1518492318767705,
      "learning_rate": 5.625003456022247e-06,
      "loss": 0.0854,
      "step": 960
    },
    {
      "epoch": 2.255813953488372,
      "grad_norm": 0.12598219156122908,
      "learning_rate": 5.311625878677658e-06,
      "loss": 0.0853,
      "step": 970
    },
    {
      "epoch": 2.2790697674418605,
      "grad_norm": 0.15991189489983731,
      "learning_rate": 5.005341787189832e-06,
      "loss": 0.0914,
      "step": 980
    },
    {
      "epoch": 2.302325581395349,
      "grad_norm": 0.10711098788498662,
      "learning_rate": 4.706375431968998e-06,
      "loss": 0.0833,
      "step": 990
    },
    {
      "epoch": 2.3255813953488373,
      "grad_norm": 0.15238996613139555,
      "learning_rate": 4.414945705636949e-06,
      "loss": 0.088,
      "step": 1000
    },
    {
      "epoch": 2.3488372093023258,
      "grad_norm": 0.13302573278619106,
      "learning_rate": 4.131265982761614e-06,
      "loss": 0.0872,
      "step": 1010
    },
    {
      "epoch": 2.3720930232558137,
      "grad_norm": 0.09653650821534612,
      "learning_rate": 3.855543963631685e-06,
      "loss": 0.0895,
      "step": 1020
    },
    {
      "epoch": 2.395348837209302,
      "grad_norm": 0.12648429239194942,
      "learning_rate": 3.587981522185829e-06,
      "loss": 0.0854,
      "step": 1030
    },
    {
      "epoch": 2.4186046511627906,
      "grad_norm": 0.1545950486769342,
      "learning_rate": 3.328774558207692e-06,
      "loss": 0.0909,
      "step": 1040
    },
    {
      "epoch": 2.441860465116279,
      "grad_norm": 0.10549590353114048,
      "learning_rate": 3.0781128538949714e-06,
      "loss": 0.0878,
      "step": 1050
    },
    {
      "epoch": 2.4651162790697674,
      "grad_norm": 0.18177777322706803,
      "learning_rate": 2.8361799349076143e-06,
      "loss": 0.0852,
      "step": 1060
    },
    {
      "epoch": 2.488372093023256,
      "grad_norm": 0.11868713958151146,
      "learning_rate": 2.6031529359967833e-06,
      "loss": 0.0891,
      "step": 1070
    },
    {
      "epoch": 2.511627906976744,
      "grad_norm": 0.1284488259520864,
      "learning_rate": 2.3792024713130284e-06,
      "loss": 0.089,
      "step": 1080
    },
    {
      "epoch": 2.5348837209302326,
      "grad_norm": 0.14486344360735834,
      "learning_rate": 2.164492509488657e-06,
      "loss": 0.0803,
      "step": 1090
    },
    {
      "epoch": 2.558139534883721,
      "grad_norm": 0.12768774247632075,
      "learning_rate": 1.9591802535856433e-06,
      "loss": 0.0852,
      "step": 1100
    },
    {
      "epoch": 2.5813953488372094,
      "grad_norm": 0.12486343748182972,
      "learning_rate": 1.763416025997126e-06,
      "loss": 0.0854,
      "step": 1110
    },
    {
      "epoch": 2.604651162790698,
      "grad_norm": 0.1325469478337174,
      "learning_rate": 1.5773431583866227e-06,
      "loss": 0.09,
      "step": 1120
    },
    {
      "epoch": 2.6279069767441863,
      "grad_norm": 0.1483837067276731,
      "learning_rate": 1.4010978867456664e-06,
      "loss": 0.0864,
      "step": 1130
    },
    {
      "epoch": 2.6511627906976747,
      "grad_norm": 0.10867631647084727,
      "learning_rate": 1.2348092516466032e-06,
      "loss": 0.0859,
      "step": 1140
    },
    {
      "epoch": 2.6744186046511627,
      "grad_norm": 0.10578223359112342,
      "learning_rate": 1.0785990037636335e-06,
      "loss": 0.0852,
      "step": 1150
    },
    {
      "epoch": 2.697674418604651,
      "grad_norm": 0.11026052474123807,
      "learning_rate": 9.325815147312739e-07,
      "loss": 0.0858,
      "step": 1160
    },
    {
      "epoch": 2.7209302325581395,
      "grad_norm": 0.1433636551446035,
      "learning_rate": 7.968636934054741e-07,
      "loss": 0.0833,
      "step": 1170
    },
    {
      "epoch": 2.744186046511628,
      "grad_norm": 0.12573674370411025,
      "learning_rate": 6.71544907588712e-07,
      "loss": 0.0783,
      "step": 1180
    },
    {
      "epoch": 2.7674418604651163,
      "grad_norm": 0.13603777866417854,
      "learning_rate": 5.567169112764109e-07,
      "loss": 0.0856,
      "step": 1190
    },
    {
      "epoch": 2.7906976744186047,
      "grad_norm": 0.12748468973009236,
      "learning_rate": 4.524637774778984e-07,
      "loss": 0.0832,
      "step": 1200
    },
    {
      "epoch": 2.813953488372093,
      "grad_norm": 0.10740073013695721,
      "learning_rate": 3.588618366610941e-07,
      "loss": 0.0859,
      "step": 1210
    },
    {
      "epoch": 2.8372093023255816,
      "grad_norm": 0.16114245090981888,
      "learning_rate": 2.7597962086605255e-07,
      "loss": 0.0842,
      "step": 1220
    },
    {
      "epoch": 2.8604651162790695,
      "grad_norm": 0.1597114067458733,
      "learning_rate": 2.038778135282171e-07,
      "loss": 0.0891,
      "step": 1230
    },
    {
      "epoch": 2.883720930232558,
      "grad_norm": 0.15931507855411944,
      "learning_rate": 1.4260920504814366e-07,
      "loss": 0.086,
      "step": 1240
    },
    {
      "epoch": 2.9069767441860463,
      "grad_norm": 0.15611356894003056,
      "learning_rate": 9.221865414023201e-08,
      "loss": 0.0808,
      "step": 1250
    },
    {
      "epoch": 2.9302325581395348,
      "grad_norm": 0.12919745509778163,
      "learning_rate": 5.2743054988758085e-08,
      "loss": 0.0813,
      "step": 1260
    },
    {
      "epoch": 2.953488372093023,
      "grad_norm": 0.11402034227218683,
      "learning_rate": 2.4211310235258687e-08,
      "loss": 0.0787,
      "step": 1270
    },
    {
      "epoch": 2.9767441860465116,
      "grad_norm": 0.14204729077366124,
      "learning_rate": 6.6443098170271276e-09,
      "loss": 0.0846,
      "step": 1280
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.1646401717020459,
      "learning_rate": 5.4915672239586045e-11,
      "loss": 0.088,
      "step": 1290
    }
  ],
  "logging_steps": 10,
  "max_steps": 1290,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 173923848683520.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}