{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.8142589118198873,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01876172607879925,
      "grad_norm": 3.62505841255188,
      "learning_rate": 1.2e-05,
      "loss": 1.0239,
      "step": 10
    },
    {
      "epoch": 0.0375234521575985,
      "grad_norm": 1.075655221939087,
      "learning_rate": 2.5333333333333337e-05,
      "loss": 0.2792,
      "step": 20
    },
    {
      "epoch": 0.05628517823639775,
      "grad_norm": 0.6459623575210571,
      "learning_rate": 3.866666666666667e-05,
      "loss": 0.1393,
      "step": 30
    },
    {
      "epoch": 0.075046904315197,
      "grad_norm": 0.5845428705215454,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.1033,
      "step": 40
    },
    {
      "epoch": 0.09380863039399624,
      "grad_norm": 0.5566798448562622,
      "learning_rate": 6.533333333333334e-05,
      "loss": 0.086,
      "step": 50
    },
    {
      "epoch": 0.1125703564727955,
      "grad_norm": 0.5197991132736206,
      "learning_rate": 7.866666666666666e-05,
      "loss": 0.0709,
      "step": 60
    },
    {
      "epoch": 0.13133208255159476,
      "grad_norm": 0.5038473010063171,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.0693,
      "step": 70
    },
    {
      "epoch": 0.150093808630394,
      "grad_norm": 0.6459262371063232,
      "learning_rate": 9.999805586153205e-05,
      "loss": 0.0631,
      "step": 80
    },
    {
      "epoch": 0.16885553470919323,
      "grad_norm": 0.41942092776298523,
      "learning_rate": 9.997618604001829e-05,
      "loss": 0.0583,
      "step": 90
    },
    {
      "epoch": 0.18761726078799248,
      "grad_norm": 0.5155685544013977,
      "learning_rate": 9.993002688846913e-05,
      "loss": 0.0551,
      "step": 100
    },
    {
      "epoch": 0.20637898686679174,
      "grad_norm": 0.35105767846107483,
      "learning_rate": 9.985960084106682e-05,
      "loss": 0.0477,
      "step": 110
    },
    {
      "epoch": 0.225140712945591,
      "grad_norm": 0.37585997581481934,
      "learning_rate": 9.976494212614377e-05,
      "loss": 0.0469,
      "step": 120
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 0.3624650835990906,
      "learning_rate": 9.964609674954696e-05,
      "loss": 0.0423,
      "step": 130
    },
    {
      "epoch": 0.2626641651031895,
      "grad_norm": 0.41389772295951843,
      "learning_rate": 9.950312247227825e-05,
      "loss": 0.0429,
      "step": 140
    },
    {
      "epoch": 0.28142589118198874,
      "grad_norm": 0.2967846691608429,
      "learning_rate": 9.933608878242153e-05,
      "loss": 0.0432,
      "step": 150
    },
    {
      "epoch": 0.300187617260788,
      "grad_norm": 0.3088073134422302,
      "learning_rate": 9.914507686137019e-05,
      "loss": 0.0392,
      "step": 160
    },
    {
      "epoch": 0.31894934333958724,
      "grad_norm": 0.2572576701641083,
      "learning_rate": 9.893017954437156e-05,
      "loss": 0.0377,
      "step": 170
    },
    {
      "epoch": 0.33771106941838647,
      "grad_norm": 0.40623533725738525,
      "learning_rate": 9.869150127540727e-05,
      "loss": 0.0347,
      "step": 180
    },
    {
      "epoch": 0.35647279549718575,
      "grad_norm": 0.41246405243873596,
      "learning_rate": 9.842915805643155e-05,
      "loss": 0.0349,
      "step": 190
    },
    {
      "epoch": 0.37523452157598497,
      "grad_norm": 0.34366941452026367,
      "learning_rate": 9.81432773909923e-05,
      "loss": 0.0348,
      "step": 200
    },
    {
      "epoch": 0.39399624765478425,
      "grad_norm": 0.41731664538383484,
      "learning_rate": 9.783399822226189e-05,
      "loss": 0.0313,
      "step": 210
    },
    {
      "epoch": 0.41275797373358347,
      "grad_norm": 0.39282381534576416,
      "learning_rate": 9.750147086550844e-05,
      "loss": 0.0361,
      "step": 220
    },
    {
      "epoch": 0.43151969981238275,
      "grad_norm": 0.2867172956466675,
      "learning_rate": 9.714585693503974e-05,
      "loss": 0.0347,
      "step": 230
    },
    {
      "epoch": 0.450281425891182,
      "grad_norm": 0.2782723903656006,
      "learning_rate": 9.676732926565585e-05,
      "loss": 0.0302,
      "step": 240
    },
    {
      "epoch": 0.46904315196998125,
      "grad_norm": 0.34300780296325684,
      "learning_rate": 9.636607182864827e-05,
      "loss": 0.0286,
      "step": 250
    },
    {
      "epoch": 0.4878048780487805,
      "grad_norm": 0.31700846552848816,
      "learning_rate": 9.594227964238653e-05,
      "loss": 0.032,
      "step": 260
    },
    {
      "epoch": 0.5065666041275797,
      "grad_norm": 0.48065921664237976,
      "learning_rate": 9.549615867753573e-05,
      "loss": 0.0303,
      "step": 270
    },
    {
      "epoch": 0.525328330206379,
      "grad_norm": 0.3679749369621277,
      "learning_rate": 9.502792575695112e-05,
      "loss": 0.0313,
      "step": 280
    },
    {
      "epoch": 0.5440900562851783,
      "grad_norm": 0.3818820118904114,
      "learning_rate": 9.453780845029821e-05,
      "loss": 0.0287,
      "step": 290
    },
    {
      "epoch": 0.5628517823639775,
      "grad_norm": 0.2918836176395416,
      "learning_rate": 9.402604496344984e-05,
      "loss": 0.0281,
      "step": 300
    },
    {
      "epoch": 0.5816135084427767,
      "grad_norm": 0.35366421937942505,
      "learning_rate": 9.349288402271388e-05,
      "loss": 0.0286,
      "step": 310
    },
    {
      "epoch": 0.600375234521576,
      "grad_norm": 0.3901827931404114,
      "learning_rate": 9.293858475394754e-05,
      "loss": 0.0286,
      "step": 320
    },
    {
      "epoch": 0.6191369606003753,
      "grad_norm": 0.4397772252559662,
      "learning_rate": 9.236341655661778e-05,
      "loss": 0.0267,
      "step": 330
    },
    {
      "epoch": 0.6378986866791745,
      "grad_norm": 0.3393397033214569,
      "learning_rate": 9.176765897286813e-05,
      "loss": 0.0284,
      "step": 340
    },
    {
      "epoch": 0.6566604127579737,
      "grad_norm": 0.2746170163154602,
      "learning_rate": 9.115160155165614e-05,
      "loss": 0.0251,
      "step": 350
    },
    {
      "epoch": 0.6754221388367729,
      "grad_norm": 0.36951932311058044,
      "learning_rate": 9.05155437080275e-05,
      "loss": 0.0263,
      "step": 360
    },
    {
      "epoch": 0.6941838649155723,
      "grad_norm": 0.32861998677253723,
      "learning_rate": 8.98597945775948e-05,
      "loss": 0.0254,
      "step": 370
    },
    {
      "epoch": 0.7129455909943715,
      "grad_norm": 0.2612588703632355,
      "learning_rate": 8.9184672866292e-05,
      "loss": 0.0224,
      "step": 380
    },
    {
      "epoch": 0.7317073170731707,
      "grad_norm": 0.27696406841278076,
      "learning_rate": 8.849050669547768e-05,
      "loss": 0.0222,
      "step": 390
    },
    {
      "epoch": 0.7504690431519699,
      "grad_norm": 0.32792651653289795,
      "learning_rate": 8.77776334424621e-05,
      "loss": 0.0235,
      "step": 400
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.30844274163246155,
      "learning_rate": 8.704639957653567e-05,
      "loss": 0.0235,
      "step": 410
    },
    {
      "epoch": 0.7879924953095685,
      "grad_norm": 0.1702653169631958,
      "learning_rate": 8.629716049057872e-05,
      "loss": 0.0245,
      "step": 420
    },
    {
      "epoch": 0.8067542213883677,
      "grad_norm": 0.2871593236923218,
      "learning_rate": 8.553028032833397e-05,
      "loss": 0.0219,
      "step": 430
    },
    {
      "epoch": 0.8255159474671669,
      "grad_norm": 0.27894458174705505,
      "learning_rate": 8.474613180742628e-05,
      "loss": 0.0187,
      "step": 440
    },
    {
      "epoch": 0.8442776735459663,
      "grad_norm": 0.2544475197792053,
      "learning_rate": 8.394509603821499e-05,
      "loss": 0.0192,
      "step": 450
    },
    {
      "epoch": 0.8630393996247655,
      "grad_norm": 0.22808410227298737,
      "learning_rate": 8.31275623385675e-05,
      "loss": 0.0217,
      "step": 460
    },
    {
      "epoch": 0.8818011257035647,
      "grad_norm": 0.23319919407367706,
      "learning_rate": 8.229392804464362e-05,
      "loss": 0.0194,
      "step": 470
    },
    {
      "epoch": 0.900562851782364,
      "grad_norm": 0.3450873792171478,
      "learning_rate": 8.14445983177832e-05,
      "loss": 0.0219,
      "step": 480
    },
    {
      "epoch": 0.9193245778611632,
      "grad_norm": 0.32911404967308044,
      "learning_rate": 8.057998594759022e-05,
      "loss": 0.0204,
      "step": 490
    },
    {
      "epoch": 0.9380863039399625,
      "grad_norm": 0.31988227367401123,
      "learning_rate": 7.970051115130966e-05,
      "loss": 0.0214,
      "step": 500
    },
    {
      "epoch": 0.9568480300187617,
      "grad_norm": 0.3178263306617737,
      "learning_rate": 7.880660136959428e-05,
      "loss": 0.0187,
      "step": 510
    },
    {
      "epoch": 0.975609756097561,
      "grad_norm": 0.27726197242736816,
      "learning_rate": 7.789869105876083e-05,
      "loss": 0.0176,
      "step": 520
    },
    {
      "epoch": 0.9943714821763602,
      "grad_norm": 0.2947522699832916,
      "learning_rate": 7.697722147963626e-05,
      "loss": 0.0184,
      "step": 530
    },
    {
      "epoch": 1.0131332082551594,
      "grad_norm": 0.3290204405784607,
      "learning_rate": 7.604264048309717e-05,
      "loss": 0.0186,
      "step": 540
    },
    {
      "epoch": 1.0318949343339587,
      "grad_norm": 0.30177435278892517,
      "learning_rate": 7.509540229240601e-05,
      "loss": 0.0186,
      "step": 550
    },
    {
      "epoch": 1.050656660412758,
      "grad_norm": 0.2899879813194275,
      "learning_rate": 7.413596728245054e-05,
      "loss": 0.0186,
      "step": 560
    },
    {
      "epoch": 1.0694183864915572,
      "grad_norm": 0.29170823097229004,
      "learning_rate": 7.316480175599309e-05,
      "loss": 0.019,
      "step": 570
    },
    {
      "epoch": 1.0881801125703565,
      "grad_norm": 0.2978293299674988,
      "learning_rate": 7.218237771703921e-05,
      "loss": 0.0189,
      "step": 580
    },
    {
      "epoch": 1.1069418386491556,
      "grad_norm": 0.2766323685646057,
      "learning_rate": 7.118917264143501e-05,
      "loss": 0.0171,
      "step": 590
    },
    {
      "epoch": 1.125703564727955,
      "grad_norm": 0.2202291339635849,
      "learning_rate": 7.018566924480543e-05,
      "loss": 0.0185,
      "step": 600
    },
    {
      "epoch": 1.1444652908067543,
      "grad_norm": 0.27024173736572266,
      "learning_rate": 6.917235524794558e-05,
      "loss": 0.0176,
      "step": 610
    },
    {
      "epoch": 1.1632270168855534,
      "grad_norm": 0.3646942675113678,
      "learning_rate": 6.814972313977967e-05,
      "loss": 0.018,
      "step": 620
    },
    {
      "epoch": 1.1819887429643527,
      "grad_norm": 0.28385287523269653,
      "learning_rate": 6.711826993800248e-05,
      "loss": 0.0175,
      "step": 630
    },
    {
      "epoch": 1.200750469043152,
      "grad_norm": 0.21348656713962555,
      "learning_rate": 6.607849694751977e-05,
      "loss": 0.0174,
      "step": 640
    },
    {
      "epoch": 1.2195121951219512,
      "grad_norm": 0.23341545462608337,
      "learning_rate": 6.503090951680512e-05,
      "loss": 0.0163,
      "step": 650
    },
    {
      "epoch": 1.2382739212007505,
      "grad_norm": 0.21429835259914398,
      "learning_rate": 6.397601679229126e-05,
      "loss": 0.0173,
      "step": 660
    },
    {
      "epoch": 1.2570356472795496,
      "grad_norm": 0.22369572520256042,
      "learning_rate": 6.291433147091583e-05,
      "loss": 0.0138,
      "step": 670
    },
    {
      "epoch": 1.275797373358349,
      "grad_norm": 0.2960052788257599,
      "learning_rate": 6.184636955094138e-05,
      "loss": 0.0147,
      "step": 680
    },
    {
      "epoch": 1.2945590994371483,
      "grad_norm": 0.2096583992242813,
      "learning_rate": 6.077265008117081e-05,
      "loss": 0.0164,
      "step": 690
    },
    {
      "epoch": 1.3133208255159474,
      "grad_norm": 0.16649533808231354,
      "learning_rate": 5.969369490868042e-05,
      "loss": 0.0154,
      "step": 700
    },
    {
      "epoch": 1.3320825515947468,
      "grad_norm": 0.21355998516082764,
      "learning_rate": 5.861002842519259e-05,
      "loss": 0.0135,
      "step": 710
    },
    {
      "epoch": 1.3508442776735459,
      "grad_norm": 0.203876331448555,
      "learning_rate": 5.75221773122121e-05,
      "loss": 0.0149,
      "step": 720
    },
    {
      "epoch": 1.3696060037523452,
      "grad_norm": 0.20695096254348755,
      "learning_rate": 5.6430670285049314e-05,
      "loss": 0.0154,
      "step": 730
    },
    {
      "epoch": 1.3883677298311445,
      "grad_norm": 0.2057274878025055,
      "learning_rate": 5.533603783585496e-05,
      "loss": 0.0159,
      "step": 740
    },
    {
      "epoch": 1.4071294559099436,
      "grad_norm": 0.20767205953598022,
      "learning_rate": 5.423881197579144e-05,
      "loss": 0.0168,
      "step": 750
    },
    {
      "epoch": 1.425891181988743,
      "grad_norm": 0.16897110641002655,
      "learning_rate": 5.313952597646568e-05,
      "loss": 0.016,
      "step": 760
    },
    {
      "epoch": 1.444652908067542,
      "grad_norm": 0.2598036527633667,
      "learning_rate": 5.203871411074954e-05,
      "loss": 0.015,
      "step": 770
    },
    {
      "epoch": 1.4634146341463414,
      "grad_norm": 0.2077157348394394,
      "learning_rate": 5.093691139311356e-05,
      "loss": 0.0138,
      "step": 780
    },
    {
      "epoch": 1.4821763602251408,
      "grad_norm": 0.24033433198928833,
      "learning_rate": 4.9834653319600246e-05,
      "loss": 0.0135,
      "step": 790
    },
    {
      "epoch": 1.50093808630394,
      "grad_norm": 0.29492080211639404,
      "learning_rate": 4.873247560756324e-05,
      "loss": 0.0129,
      "step": 800
    },
    {
      "epoch": 1.5196998123827392,
      "grad_norm": 0.11821957677602768,
      "learning_rate": 4.7630913935299066e-05,
      "loss": 0.013,
      "step": 810
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.14121685922145844,
      "learning_rate": 4.65305036816978e-05,
      "loss": 0.0148,
      "step": 820
    },
    {
      "epoch": 1.5572232645403377,
      "grad_norm": 0.17172347009181976,
      "learning_rate": 4.543177966603925e-05,
      "loss": 0.014,
      "step": 830
    },
    {
      "epoch": 1.575984990619137,
      "grad_norm": 0.17106831073760986,
      "learning_rate": 4.433527588806103e-05,
      "loss": 0.0119,
      "step": 840
    },
    {
      "epoch": 1.5947467166979363,
      "grad_norm": 0.20430134236812592,
      "learning_rate": 4.324152526842517e-05,
      "loss": 0.0145,
      "step": 850
    },
    {
      "epoch": 1.6135084427767354,
      "grad_norm": 0.19569315016269684,
      "learning_rate": 4.215105938970889e-05,
      "loss": 0.0137,
      "step": 860
    },
    {
      "epoch": 1.6322701688555346,
      "grad_norm": 0.16233587265014648,
      "learning_rate": 4.1064408238045994e-05,
      "loss": 0.0134,
      "step": 870
    },
    {
      "epoch": 1.6510318949343339,
      "grad_norm": 0.16989213228225708,
      "learning_rate": 3.9982099945543945e-05,
      "loss": 0.0126,
      "step": 880
    },
    {
      "epoch": 1.6697936210131332,
      "grad_norm": 0.18682365119457245,
      "learning_rate": 3.890466053360211e-05,
      "loss": 0.0115,
      "step": 890
    },
    {
      "epoch": 1.6885553470919326,
      "grad_norm": 0.20883384346961975,
      "learning_rate": 3.783261365725592e-05,
      "loss": 0.0124,
      "step": 900
    },
    {
      "epoch": 1.7073170731707317,
      "grad_norm": 0.21428252756595612,
      "learning_rate": 3.676648035067093e-05,
      "loss": 0.0122,
      "step": 910
    },
    {
      "epoch": 1.726078799249531,
      "grad_norm": 0.20880082249641418,
      "learning_rate": 3.570677877391092e-05,
      "loss": 0.0125,
      "step": 920
    },
    {
      "epoch": 1.7448405253283301,
      "grad_norm": 0.16143251955509186,
      "learning_rate": 3.465402396110269e-05,
      "loss": 0.012,
      "step": 930
    },
    {
      "epoch": 1.7636022514071295,
      "grad_norm": 0.1445058137178421,
      "learning_rate": 3.360872757012011e-05,
      "loss": 0.0123,
      "step": 940
    },
    {
      "epoch": 1.7823639774859288,
      "grad_norm": 0.21280893683433533,
      "learning_rate": 3.257139763390925e-05,
      "loss": 0.0125,
      "step": 950
    },
    {
      "epoch": 1.8011257035647281,
      "grad_norm": 0.16495130956172943,
      "learning_rate": 3.1542538313575035e-05,
      "loss": 0.0116,
      "step": 960
    },
    {
      "epoch": 1.8198874296435272,
      "grad_norm": 0.19219523668289185,
      "learning_rate": 3.052264965335e-05,
      "loss": 0.0133,
      "step": 970
    },
    {
      "epoch": 1.8386491557223263,
      "grad_norm": 0.173895001411438,
      "learning_rate": 2.9512227337563604e-05,
      "loss": 0.0116,
      "step": 980
    },
    {
      "epoch": 1.8574108818011257,
      "grad_norm": 0.16958042979240417,
      "learning_rate": 2.8511762449730795e-05,
      "loss": 0.0125,
      "step": 990
    },
    {
      "epoch": 1.876172607879925,
      "grad_norm": 0.1639089733362198,
      "learning_rate": 2.7521741233876496e-05,
      "loss": 0.0122,
      "step": 1000
    },
    {
      "epoch": 1.8949343339587243,
      "grad_norm": 0.16494053602218628,
      "learning_rate": 2.654264485821214e-05,
      "loss": 0.011,
      "step": 1010
    },
    {
      "epoch": 1.9136960600375235,
      "grad_norm": 0.17180362343788147,
      "learning_rate": 2.55749491812794e-05,
      "loss": 0.0113,
      "step": 1020
    },
    {
      "epoch": 1.9324577861163226,
      "grad_norm": 0.19289036095142365,
      "learning_rate": 2.4619124520674146e-05,
      "loss": 0.0131,
      "step": 1030
    },
    {
      "epoch": 1.951219512195122,
      "grad_norm": 0.1478220820426941,
      "learning_rate": 2.3675635424463754e-05,
      "loss": 0.0097,
      "step": 1040
    },
    {
      "epoch": 1.9699812382739212,
      "grad_norm": 0.18471956253051758,
      "learning_rate": 2.2744940445408202e-05,
      "loss": 0.0118,
      "step": 1050
    },
    {
      "epoch": 1.9887429643527206,
      "grad_norm": 0.16719312965869904,
      "learning_rate": 2.182749191809518e-05,
      "loss": 0.0104,
      "step": 1060
    },
    {
      "epoch": 2.00750469043152,
      "grad_norm": 0.15944059193134308,
      "learning_rate": 2.09237357390974e-05,
      "loss": 0.0109,
      "step": 1070
    },
    {
      "epoch": 2.026266416510319,
      "grad_norm": 0.20150484144687653,
      "learning_rate": 2.0034111150258666e-05,
      "loss": 0.0129,
      "step": 1080
    },
    {
      "epoch": 2.045028142589118,
      "grad_norm": 0.15957467257976532,
      "learning_rate": 1.9159050525214452e-05,
      "loss": 0.011,
      "step": 1090
    },
    {
      "epoch": 2.0637898686679175,
      "grad_norm": 0.133390411734581,
      "learning_rate": 1.8298979159250557e-05,
      "loss": 0.0103,
      "step": 1100
    },
    {
      "epoch": 2.082551594746717,
      "grad_norm": 0.13808146119117737,
      "learning_rate": 1.745431506260173e-05,
      "loss": 0.0115,
      "step": 1110
    },
    {
      "epoch": 2.101313320825516,
      "grad_norm": 0.16296911239624023,
      "learning_rate": 1.662546875729138e-05,
      "loss": 0.0112,
      "step": 1120
    },
    {
      "epoch": 2.120075046904315,
      "grad_norm": 0.13522660732269287,
      "learning_rate": 1.581284307761024e-05,
      "loss": 0.0106,
      "step": 1130
    },
    {
      "epoch": 2.1388367729831144,
      "grad_norm": 0.19323870539665222,
      "learning_rate": 1.5016832974331724e-05,
      "loss": 0.0123,
      "step": 1140
    },
    {
      "epoch": 2.1575984990619137,
      "grad_norm": 0.1635001301765442,
      "learning_rate": 1.4237825322758736e-05,
      "loss": 0.013,
      "step": 1150
    },
    {
      "epoch": 2.176360225140713,
      "grad_norm": 0.17513315379619598,
      "learning_rate": 1.3476198734695122e-05,
      "loss": 0.0104,
      "step": 1160
    },
    {
      "epoch": 2.1951219512195124,
      "grad_norm": 0.10277465730905533,
      "learning_rate": 1.2732323374433707e-05,
      "loss": 0.0105,
      "step": 1170
    },
    {
      "epoch": 2.2138836772983113,
      "grad_norm": 0.1144745722413063,
      "learning_rate": 1.2006560778849578e-05,
      "loss": 0.0103,
      "step": 1180
    },
    {
      "epoch": 2.2326454033771106,
      "grad_norm": 0.1264655739068985,
      "learning_rate": 1.1299263681686706e-05,
      "loss": 0.0119,
      "step": 1190
    },
    {
      "epoch": 2.25140712945591,
      "grad_norm": 0.10604596883058548,
      "learning_rate": 1.0610775842122972e-05,
      "loss": 0.0109,
      "step": 1200
    },
    {
      "epoch": 2.2701688555347093,
      "grad_norm": 0.14508432149887085,
      "learning_rate": 9.941431877696955e-06,
      "loss": 0.0109,
      "step": 1210
    },
    {
      "epoch": 2.2889305816135086,
      "grad_norm": 0.12284737080335617,
      "learning_rate": 9.291557101677784e-06,
      "loss": 0.0093,
      "step": 1220
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 0.1272381991147995,
      "learning_rate": 8.661467364956993e-06,
      "loss": 0.0116,
      "step": 1230
    },
    {
      "epoch": 2.326454033771107,
      "grad_norm": 0.14646074175834656,
      "learning_rate": 8.051468902539272e-06,
      "loss": 0.0102,
      "step": 1240
    },
    {
      "epoch": 2.345215759849906,
      "grad_norm": 0.11431489139795303,
      "learning_rate": 7.461858184706777e-06,
      "loss": 0.0091,
      "step": 1250
    },
    {
      "epoch": 2.3639774859287055,
      "grad_norm": 0.1557149738073349,
      "learning_rate": 6.892921772929112e-06,
      "loss": 0.0097,
      "step": 1260
    },
    {
      "epoch": 2.382739212007505,
      "grad_norm": 0.14029847085475922,
      "learning_rate": 6.344936180589351e-06,
      "loss": 0.0108,
      "step": 1270
    },
    {
      "epoch": 2.401500938086304,
      "grad_norm": 0.11800851672887802,
      "learning_rate": 5.818167738593505e-06,
      "loss": 0.0098,
      "step": 1280
    },
    {
      "epoch": 2.420262664165103,
      "grad_norm": 0.11177805066108704,
      "learning_rate": 5.312872465928881e-06,
      "loss": 0.0093,
      "step": 1290
    },
    {
      "epoch": 2.4390243902439024,
      "grad_norm": 0.12152400612831116,
      "learning_rate": 4.829295945234258e-06,
      "loss": 0.0099,
      "step": 1300
    },
    {
      "epoch": 2.4577861163227017,
      "grad_norm": 0.10167639702558517,
      "learning_rate": 4.367673203442241e-06,
      "loss": 0.0096,
      "step": 1310
    },
    {
      "epoch": 2.476547842401501,
      "grad_norm": 0.13529738783836365,
      "learning_rate": 3.928228597551947e-06,
      "loss": 0.0099,
      "step": 1320
    },
    {
      "epoch": 2.4953095684803,
      "grad_norm": 0.10373085737228394,
      "learning_rate": 3.511175705587433e-06,
      "loss": 0.0088,
      "step": 1330
    },
    {
      "epoch": 2.5140712945590993,
      "grad_norm": 0.13128240406513214,
      "learning_rate": 3.1167172227949347e-06,
      "loss": 0.0113,
      "step": 1340
    },
    {
      "epoch": 2.5328330206378986,
      "grad_norm": 0.11484728753566742,
      "learning_rate": 2.7450448631293036e-06,
      "loss": 0.0088,
      "step": 1350
    },
    {
      "epoch": 2.551594746716698,
      "grad_norm": 0.1032821536064148,
      "learning_rate": 2.3963392660775575e-06,
      "loss": 0.0088,
      "step": 1360
    },
    {
      "epoch": 2.5703564727954973,
      "grad_norm": 0.10594484210014343,
      "learning_rate": 2.0707699088647836e-06,
      "loss": 0.0099,
      "step": 1370
    },
    {
      "epoch": 2.5891181988742966,
      "grad_norm": 0.0838296115398407,
      "learning_rate": 1.7684950240852372e-06,
      "loss": 0.0088,
      "step": 1380
    },
    {
      "epoch": 2.607879924953096,
      "grad_norm": 0.09543336927890778,
      "learning_rate": 1.4896615227983468e-06,
      "loss": 0.0091,
      "step": 1390
    },
    {
      "epoch": 2.626641651031895,
      "grad_norm": 0.1242263913154602,
      "learning_rate": 1.2344049231273302e-06,
      "loss": 0.0092,
      "step": 1400
    },
    {
      "epoch": 2.645403377110694,
      "grad_norm": 0.1851985901594162,
      "learning_rate": 1.00284928439493e-06,
      "loss": 0.0106,
      "step": 1410
    },
    {
      "epoch": 2.6641651031894935,
      "grad_norm": 0.08041252940893173,
      "learning_rate": 7.951071468283167e-07,
      "loss": 0.0098,
      "step": 1420
    },
    {
      "epoch": 2.682926829268293,
      "grad_norm": 0.11750051379203796,
      "learning_rate": 6.11279476862553e-07,
      "loss": 0.0096,
      "step": 1430
    },
    {
      "epoch": 2.7016885553470917,
      "grad_norm": 0.08895982056856155,
      "learning_rate": 4.514556180690188e-07,
      "loss": 0.0093,
      "step": 1440
    },
    {
      "epoch": 2.720450281425891,
      "grad_norm": 0.11088193953037262,
      "learning_rate": 3.157132477328628e-07,
      "loss": 0.0108,
      "step": 1450
    },
    {
      "epoch": 2.7392120075046904,
      "grad_norm": 0.08702103048563004,
      "learning_rate": 2.041183391004453e-07,
      "loss": 0.0083,
      "step": 1460
    },
    {
      "epoch": 2.7579737335834897,
      "grad_norm": 0.10216435045003891,
      "learning_rate": 1.1672512931509283e-07,
      "loss": 0.0098,
      "step": 1470
    },
    {
      "epoch": 2.776735459662289,
      "grad_norm": 0.11117573082447052,
      "learning_rate": 5.3576093056922906e-08,
      "loss": 0.0097,
      "step": 1480
    },
    {
      "epoch": 2.7954971857410884,
      "grad_norm": 0.1127699688076973,
      "learning_rate": 1.4701921899362703e-08,
      "loss": 0.01,
      "step": 1490
    },
    {
      "epoch": 2.8142589118198873,
      "grad_norm": 0.15439969301223755,
      "learning_rate": 1.2150939247002058e-10,
      "loss": 0.0112,
      "step": 1500
    },
    {
      "epoch": 2.8142589118198873,
      "step": 1500,
      "total_flos": 0.0,
      "train_loss": 0.0293558024764061,
      "train_runtime": 1931.3019,
      "train_samples_per_second": 77.668,
      "train_steps_per_second": 0.777
    }
  ],
  "logging_steps": 10,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 100,
  "trial_name": null,
  "trial_params": null
}