{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 2385,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.020964360587002098,
      "grad_norm": 3.998500347137451,
      "learning_rate": 9.958071278825997e-05,
      "loss": 0.1821,
      "step": 10
    },
    {
      "epoch": 0.041928721174004195,
      "grad_norm": 4.23317813873291,
      "learning_rate": 9.916142557651992e-05,
      "loss": 0.1617,
      "step": 20
    },
    {
      "epoch": 0.06289308176100629,
      "grad_norm": 4.348649024963379,
      "learning_rate": 9.874213836477988e-05,
      "loss": 0.1799,
      "step": 30
    },
    {
      "epoch": 0.08385744234800839,
      "grad_norm": 4.8091020584106445,
      "learning_rate": 9.832285115303984e-05,
      "loss": 0.2127,
      "step": 40
    },
    {
      "epoch": 0.10482180293501048,
      "grad_norm": 4.523597240447998,
      "learning_rate": 9.790356394129979e-05,
      "loss": 0.2195,
      "step": 50
    },
    {
      "epoch": 0.12578616352201258,
      "grad_norm": 6.083479404449463,
      "learning_rate": 9.748427672955975e-05,
      "loss": 0.2142,
      "step": 60
    },
    {
      "epoch": 0.14675052410901468,
      "grad_norm": 5.268312454223633,
      "learning_rate": 9.706498951781971e-05,
      "loss": 0.2293,
      "step": 70
    },
    {
      "epoch": 0.16771488469601678,
      "grad_norm": 5.908782482147217,
      "learning_rate": 9.664570230607967e-05,
      "loss": 0.2391,
      "step": 80
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 5.333123207092285,
      "learning_rate": 9.622641509433963e-05,
      "loss": 0.2519,
      "step": 90
    },
    {
      "epoch": 0.20964360587002095,
      "grad_norm": 4.541295051574707,
      "learning_rate": 9.58071278825996e-05,
      "loss": 0.241,
      "step": 100
    },
    {
      "epoch": 0.23060796645702306,
      "grad_norm": 5.367836952209473,
      "learning_rate": 9.538784067085954e-05,
      "loss": 0.2516,
      "step": 110
    },
    {
      "epoch": 0.25157232704402516,
      "grad_norm": 4.476315975189209,
      "learning_rate": 9.496855345911951e-05,
      "loss": 0.2581,
      "step": 120
    },
    {
      "epoch": 0.27253668763102723,
      "grad_norm": 4.104060173034668,
      "learning_rate": 9.454926624737947e-05,
      "loss": 0.2612,
      "step": 130
    },
    {
      "epoch": 0.29350104821802936,
      "grad_norm": 5.152798652648926,
      "learning_rate": 9.412997903563942e-05,
      "loss": 0.2736,
      "step": 140
    },
    {
      "epoch": 0.31446540880503143,
      "grad_norm": 4.785223960876465,
      "learning_rate": 9.371069182389938e-05,
      "loss": 0.267,
      "step": 150
    },
    {
      "epoch": 0.33542976939203356,
      "grad_norm": 4.769683837890625,
      "learning_rate": 9.329140461215934e-05,
      "loss": 0.2766,
      "step": 160
    },
    {
      "epoch": 0.35639412997903563,
      "grad_norm": 4.8373942375183105,
      "learning_rate": 9.287211740041929e-05,
      "loss": 0.2878,
      "step": 170
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 4.6672515869140625,
      "learning_rate": 9.245283018867925e-05,
      "loss": 0.2748,
      "step": 180
    },
    {
      "epoch": 0.39832285115303984,
      "grad_norm": 7.549396514892578,
      "learning_rate": 9.203354297693921e-05,
      "loss": 0.264,
      "step": 190
    },
    {
      "epoch": 0.4192872117400419,
      "grad_norm": 6.70684289932251,
      "learning_rate": 9.161425576519916e-05,
      "loss": 0.2843,
      "step": 200
    },
    {
      "epoch": 0.44025157232704404,
      "grad_norm": 5.79678201675415,
      "learning_rate": 9.119496855345912e-05,
      "loss": 0.2907,
      "step": 210
    },
    {
      "epoch": 0.4612159329140461,
      "grad_norm": 6.699160575866699,
      "learning_rate": 9.077568134171907e-05,
      "loss": 0.2678,
      "step": 220
    },
    {
      "epoch": 0.48218029350104824,
      "grad_norm": 4.357872486114502,
      "learning_rate": 9.035639412997903e-05,
      "loss": 0.2965,
      "step": 230
    },
    {
      "epoch": 0.5031446540880503,
      "grad_norm": 5.521259307861328,
      "learning_rate": 8.9937106918239e-05,
      "loss": 0.2876,
      "step": 240
    },
    {
      "epoch": 0.5241090146750524,
      "grad_norm": 6.627674579620361,
      "learning_rate": 8.951781970649896e-05,
      "loss": 0.2825,
      "step": 250
    },
    {
      "epoch": 0.5450733752620545,
      "grad_norm": 5.5945634841918945,
      "learning_rate": 8.909853249475892e-05,
      "loss": 0.2855,
      "step": 260
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 4.596319198608398,
      "learning_rate": 8.867924528301888e-05,
      "loss": 0.2753,
      "step": 270
    },
    {
      "epoch": 0.5870020964360587,
      "grad_norm": 5.92698860168457,
      "learning_rate": 8.825995807127883e-05,
      "loss": 0.3076,
      "step": 280
    },
    {
      "epoch": 0.6079664570230608,
      "grad_norm": 4.971231937408447,
      "learning_rate": 8.784067085953879e-05,
      "loss": 0.2945,
      "step": 290
    },
    {
      "epoch": 0.6289308176100629,
      "grad_norm": 5.803814888000488,
      "learning_rate": 8.742138364779875e-05,
      "loss": 0.3121,
      "step": 300
    },
    {
      "epoch": 0.649895178197065,
      "grad_norm": 4.640927314758301,
      "learning_rate": 8.70020964360587e-05,
      "loss": 0.2965,
      "step": 310
    },
    {
      "epoch": 0.6708595387840671,
      "grad_norm": 4.943049430847168,
      "learning_rate": 8.662473794549267e-05,
      "loss": 0.2814,
      "step": 320
    },
    {
      "epoch": 0.6918238993710691,
      "grad_norm": 18.59223175048828,
      "learning_rate": 8.620545073375263e-05,
      "loss": 0.2998,
      "step": 330
    },
    {
      "epoch": 0.7127882599580713,
      "grad_norm": 4.966686725616455,
      "learning_rate": 8.578616352201259e-05,
      "loss": 0.282,
      "step": 340
    },
    {
      "epoch": 0.7337526205450734,
      "grad_norm": 6.274560451507568,
      "learning_rate": 8.536687631027254e-05,
      "loss": 0.3123,
      "step": 350
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 5.680440902709961,
      "learning_rate": 8.49475890985325e-05,
      "loss": 0.2915,
      "step": 360
    },
    {
      "epoch": 0.7756813417190775,
      "grad_norm": 5.426858425140381,
      "learning_rate": 8.452830188679246e-05,
      "loss": 0.3192,
      "step": 370
    },
    {
      "epoch": 0.7966457023060797,
      "grad_norm": 5.293428421020508,
      "learning_rate": 8.410901467505241e-05,
      "loss": 0.3207,
      "step": 380
    },
    {
      "epoch": 0.8176100628930818,
      "grad_norm": 5.778203964233398,
      "learning_rate": 8.368972746331237e-05,
      "loss": 0.3202,
      "step": 390
    },
    {
      "epoch": 0.8385744234800838,
      "grad_norm": 5.326625347137451,
      "learning_rate": 8.327044025157233e-05,
      "loss": 0.3088,
      "step": 400
    },
    {
      "epoch": 0.859538784067086,
      "grad_norm": 5.26389217376709,
      "learning_rate": 8.285115303983228e-05,
      "loss": 0.3084,
      "step": 410
    },
    {
      "epoch": 0.8805031446540881,
      "grad_norm": 5.999328136444092,
      "learning_rate": 8.243186582809224e-05,
      "loss": 0.2822,
      "step": 420
    },
    {
      "epoch": 0.9014675052410901,
      "grad_norm": 5.0098958015441895,
      "learning_rate": 8.20125786163522e-05,
      "loss": 0.307,
      "step": 430
    },
    {
      "epoch": 0.9224318658280922,
      "grad_norm": 6.384605407714844,
      "learning_rate": 8.159329140461215e-05,
      "loss": 0.2976,
      "step": 440
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 9.32125186920166,
      "learning_rate": 8.117400419287212e-05,
      "loss": 0.3198,
      "step": 450
    },
    {
      "epoch": 0.9643605870020965,
      "grad_norm": 4.630229949951172,
      "learning_rate": 8.075471698113208e-05,
      "loss": 0.3031,
      "step": 460
    },
    {
      "epoch": 0.9853249475890985,
      "grad_norm": 4.775624752044678,
      "learning_rate": 8.033542976939204e-05,
      "loss": 0.2676,
      "step": 470
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.6731473207473755,
      "eval_runtime": 20.7096,
      "eval_samples_per_second": 72.43,
      "eval_steps_per_second": 9.078,
      "step": 477
    },
    {
      "epoch": 1.0062893081761006,
      "grad_norm": 1.9550973176956177,
      "learning_rate": 7.9916142557652e-05,
      "loss": 0.2451,
      "step": 480
    },
    {
      "epoch": 1.0272536687631026,
      "grad_norm": 3.6781623363494873,
      "learning_rate": 7.949685534591196e-05,
      "loss": 0.1127,
      "step": 490
    },
    {
      "epoch": 1.0482180293501049,
      "grad_norm": 3.6007821559906006,
      "learning_rate": 7.907756813417191e-05,
      "loss": 0.1089,
      "step": 500
    },
    {
      "epoch": 1.069182389937107,
      "grad_norm": 3.0092623233795166,
      "learning_rate": 7.865828092243187e-05,
      "loss": 0.1203,
      "step": 510
    },
    {
      "epoch": 1.090146750524109,
      "grad_norm": 2.852006196975708,
      "learning_rate": 7.823899371069184e-05,
      "loss": 0.1126,
      "step": 520
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 3.5420737266540527,
      "learning_rate": 7.781970649895178e-05,
      "loss": 0.1254,
      "step": 530
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 3.2206077575683594,
      "learning_rate": 7.740041928721175e-05,
      "loss": 0.1328,
      "step": 540
    },
    {
      "epoch": 1.1530398322851152,
      "grad_norm": 3.895273447036743,
      "learning_rate": 7.698113207547171e-05,
      "loss": 0.125,
      "step": 550
    },
    {
      "epoch": 1.1740041928721174,
      "grad_norm": 3.2703685760498047,
      "learning_rate": 7.656184486373166e-05,
      "loss": 0.1306,
      "step": 560
    },
    {
      "epoch": 1.1949685534591195,
      "grad_norm": 2.977438449859619,
      "learning_rate": 7.614255765199162e-05,
      "loss": 0.1261,
      "step": 570
    },
    {
      "epoch": 1.2159329140461215,
      "grad_norm": 3.917140007019043,
      "learning_rate": 7.572327044025158e-05,
      "loss": 0.1232,
      "step": 580
    },
    {
      "epoch": 1.2368972746331237,
      "grad_norm": 4.090723514556885,
      "learning_rate": 7.530398322851153e-05,
      "loss": 0.1374,
      "step": 590
    },
    {
      "epoch": 1.2578616352201257,
      "grad_norm": 3.666203737258911,
      "learning_rate": 7.488469601677149e-05,
      "loss": 0.1317,
      "step": 600
    },
    {
      "epoch": 1.2788259958071277,
      "grad_norm": 4.610381126403809,
      "learning_rate": 7.446540880503144e-05,
      "loss": 0.1276,
      "step": 610
    },
    {
      "epoch": 1.29979035639413,
      "grad_norm": 5.928828716278076,
      "learning_rate": 7.40461215932914e-05,
      "loss": 0.1258,
      "step": 620
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 4.312713623046875,
      "learning_rate": 7.362683438155136e-05,
      "loss": 0.1318,
      "step": 630
    },
    {
      "epoch": 1.3417190775681342,
      "grad_norm": 3.831658363342285,
      "learning_rate": 7.320754716981132e-05,
      "loss": 0.1217,
      "step": 640
    },
    {
      "epoch": 1.3626834381551363,
      "grad_norm": 2.7465438842773438,
      "learning_rate": 7.278825995807129e-05,
      "loss": 0.1157,
      "step": 650
    },
    {
      "epoch": 1.3836477987421385,
      "grad_norm": 2.915738582611084,
      "learning_rate": 7.236897274633125e-05,
      "loss": 0.1414,
      "step": 660
    },
    {
      "epoch": 1.4046121593291405,
      "grad_norm": 3.2107832431793213,
      "learning_rate": 7.19496855345912e-05,
      "loss": 0.1253,
      "step": 670
    },
    {
      "epoch": 1.4255765199161425,
      "grad_norm": 3.7944188117980957,
      "learning_rate": 7.153039832285116e-05,
      "loss": 0.1249,
      "step": 680
    },
    {
      "epoch": 1.4465408805031448,
      "grad_norm": 3.0400636196136475,
      "learning_rate": 7.111111111111112e-05,
      "loss": 0.1323,
      "step": 690
    },
    {
      "epoch": 1.4675052410901468,
      "grad_norm": 3.911773443222046,
      "learning_rate": 7.069182389937107e-05,
      "loss": 0.1265,
      "step": 700
    },
    {
      "epoch": 1.4884696016771488,
      "grad_norm": 4.3404765129089355,
      "learning_rate": 7.027253668763103e-05,
      "loss": 0.1328,
      "step": 710
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 3.5862231254577637,
      "learning_rate": 6.985324947589099e-05,
      "loss": 0.1336,
      "step": 720
    },
    {
      "epoch": 1.530398322851153,
      "grad_norm": 4.099681377410889,
      "learning_rate": 6.943396226415094e-05,
      "loss": 0.1345,
      "step": 730
    },
    {
      "epoch": 1.551362683438155,
      "grad_norm": 4.485490322113037,
      "learning_rate": 6.90146750524109e-05,
      "loss": 0.1368,
      "step": 740
    },
    {
      "epoch": 1.5723270440251573,
      "grad_norm": 3.4771294593811035,
      "learning_rate": 6.859538784067086e-05,
      "loss": 0.1197,
      "step": 750
    },
    {
      "epoch": 1.5932914046121593,
      "grad_norm": 3.602762460708618,
      "learning_rate": 6.817610062893081e-05,
      "loss": 0.125,
      "step": 760
    },
    {
      "epoch": 1.6142557651991614,
      "grad_norm": 3.6131410598754883,
      "learning_rate": 6.775681341719077e-05,
      "loss": 0.1363,
      "step": 770
    },
    {
      "epoch": 1.6352201257861636,
      "grad_norm": 4.117647647857666,
      "learning_rate": 6.733752620545074e-05,
      "loss": 0.1289,
      "step": 780
    },
    {
      "epoch": 1.6561844863731656,
      "grad_norm": 3.723130941390991,
      "learning_rate": 6.691823899371068e-05,
      "loss": 0.1319,
      "step": 790
    },
    {
      "epoch": 1.6771488469601676,
      "grad_norm": 3.42946195602417,
      "learning_rate": 6.649895178197065e-05,
      "loss": 0.1449,
      "step": 800
    },
    {
      "epoch": 1.6981132075471699,
      "grad_norm": 3.705895185470581,
      "learning_rate": 6.607966457023061e-05,
      "loss": 0.1446,
      "step": 810
    },
    {
      "epoch": 1.719077568134172,
      "grad_norm": 4.407285213470459,
      "learning_rate": 6.566037735849057e-05,
      "loss": 0.1417,
      "step": 820
    },
    {
      "epoch": 1.740041928721174,
      "grad_norm": 3.557760000228882,
      "learning_rate": 6.524109014675053e-05,
      "loss": 0.1338,
      "step": 830
    },
    {
      "epoch": 1.7610062893081762,
      "grad_norm": 3.342160224914551,
      "learning_rate": 6.48218029350105e-05,
      "loss": 0.1439,
      "step": 840
    },
    {
      "epoch": 1.7819706498951782,
      "grad_norm": 3.4098939895629883,
      "learning_rate": 6.440251572327044e-05,
      "loss": 0.1444,
      "step": 850
    },
    {
      "epoch": 1.8029350104821802,
      "grad_norm": 3.757795572280884,
      "learning_rate": 6.39832285115304e-05,
      "loss": 0.1327,
      "step": 860
    },
    {
      "epoch": 1.8238993710691824,
      "grad_norm": 3.3094663619995117,
      "learning_rate": 6.356394129979037e-05,
      "loss": 0.1338,
      "step": 870
    },
    {
      "epoch": 1.8448637316561844,
      "grad_norm": 3.925997495651245,
      "learning_rate": 6.314465408805031e-05,
      "loss": 0.1391,
      "step": 880
    },
    {
      "epoch": 1.8658280922431865,
      "grad_norm": 3.38191819190979,
      "learning_rate": 6.272536687631028e-05,
      "loss": 0.1396,
      "step": 890
    },
    {
      "epoch": 1.8867924528301887,
      "grad_norm": 3.194735288619995,
      "learning_rate": 6.230607966457024e-05,
      "loss": 0.1304,
      "step": 900
    },
    {
      "epoch": 1.9077568134171907,
      "grad_norm": 3.6266653537750244,
      "learning_rate": 6.188679245283019e-05,
      "loss": 0.1259,
      "step": 910
    },
    {
      "epoch": 1.9287211740041927,
      "grad_norm": 4.087035655975342,
      "learning_rate": 6.146750524109015e-05,
      "loss": 0.145,
      "step": 920
    },
    {
      "epoch": 1.949685534591195,
      "grad_norm": 3.725505828857422,
      "learning_rate": 6.104821802935011e-05,
      "loss": 0.1291,
      "step": 930
    },
    {
      "epoch": 1.9706498951781972,
      "grad_norm": 3.440652847290039,
      "learning_rate": 6.0628930817610065e-05,
      "loss": 0.1348,
      "step": 940
    },
    {
      "epoch": 1.991614255765199,
      "grad_norm": 3.5756237506866455,
      "learning_rate": 6.020964360587003e-05,
      "loss": 0.1535,
      "step": 950
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.6702887415885925,
      "eval_runtime": 20.8625,
      "eval_samples_per_second": 71.9,
      "eval_steps_per_second": 9.011,
      "step": 954
    },
    {
      "epoch": 2.0125786163522013,
      "grad_norm": 1.9692338705062866,
      "learning_rate": 5.979035639412999e-05,
      "loss": 0.0812,
      "step": 960
    },
    {
      "epoch": 2.0335429769392035,
      "grad_norm": 2.1989524364471436,
      "learning_rate": 5.937106918238994e-05,
      "loss": 0.0489,
      "step": 970
    },
    {
      "epoch": 2.0545073375262053,
      "grad_norm": 1.9158607721328735,
      "learning_rate": 5.89517819706499e-05,
      "loss": 0.0465,
      "step": 980
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 1.7822378873825073,
      "learning_rate": 5.853249475890986e-05,
      "loss": 0.049,
      "step": 990
    },
    {
      "epoch": 2.0964360587002098,
      "grad_norm": 2.644869089126587,
      "learning_rate": 5.811320754716981e-05,
      "loss": 0.0545,
      "step": 1000
    },
    {
      "epoch": 2.1174004192872116,
      "grad_norm": 2.888030529022217,
      "learning_rate": 5.769392033542977e-05,
      "loss": 0.0563,
      "step": 1010
    },
    {
      "epoch": 2.138364779874214,
      "grad_norm": 2.034641742706299,
      "learning_rate": 5.727463312368973e-05,
      "loss": 0.0527,
      "step": 1020
    },
    {
      "epoch": 2.159329140461216,
      "grad_norm": 1.8326114416122437,
      "learning_rate": 5.685534591194969e-05,
      "loss": 0.0509,
      "step": 1030
    },
    {
      "epoch": 2.180293501048218,
      "grad_norm": 2.1253228187561035,
      "learning_rate": 5.643605870020965e-05,
      "loss": 0.0533,
      "step": 1040
    },
    {
      "epoch": 2.20125786163522,
      "grad_norm": 2.369814872741699,
      "learning_rate": 5.60167714884696e-05,
      "loss": 0.0535,
      "step": 1050
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 2.15043568611145,
      "learning_rate": 5.559748427672956e-05,
      "loss": 0.0525,
      "step": 1060
    },
    {
      "epoch": 2.243186582809224,
      "grad_norm": 2.138932943344116,
      "learning_rate": 5.517819706498952e-05,
      "loss": 0.052,
      "step": 1070
    },
    {
      "epoch": 2.2641509433962264,
      "grad_norm": 2.2840254306793213,
      "learning_rate": 5.475890985324947e-05,
      "loss": 0.0582,
      "step": 1080
    },
    {
      "epoch": 2.2851153039832286,
      "grad_norm": 1.516945481300354,
      "learning_rate": 5.433962264150943e-05,
      "loss": 0.0557,
      "step": 1090
    },
    {
      "epoch": 2.3060796645702304,
      "grad_norm": 2.2309811115264893,
      "learning_rate": 5.3920335429769395e-05,
      "loss": 0.0563,
      "step": 1100
    },
    {
      "epoch": 2.3270440251572326,
      "grad_norm": 2.9257094860076904,
      "learning_rate": 5.350104821802935e-05,
      "loss": 0.056,
      "step": 1110
    },
    {
      "epoch": 2.348008385744235,
      "grad_norm": 2.112194538116455,
      "learning_rate": 5.308176100628931e-05,
      "loss": 0.0567,
      "step": 1120
    },
    {
      "epoch": 2.368972746331237,
      "grad_norm": 2.411298990249634,
      "learning_rate": 5.2662473794549274e-05,
      "loss": 0.0558,
      "step": 1130
    },
    {
      "epoch": 2.389937106918239,
      "grad_norm": 2.410545825958252,
      "learning_rate": 5.224318658280922e-05,
      "loss": 0.0557,
      "step": 1140
    },
    {
      "epoch": 2.410901467505241,
      "grad_norm": 2.1554999351501465,
      "learning_rate": 5.1823899371069184e-05,
      "loss": 0.0589,
      "step": 1150
    },
    {
      "epoch": 2.431865828092243,
      "grad_norm": 2.1737091541290283,
      "learning_rate": 5.1404612159329146e-05,
      "loss": 0.0673,
      "step": 1160
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 3.798069477081299,
      "learning_rate": 5.0985324947589094e-05,
      "loss": 0.0546,
      "step": 1170
    },
    {
      "epoch": 2.4737945492662474,
      "grad_norm": 2.5850603580474854,
      "learning_rate": 5.0566037735849056e-05,
      "loss": 0.0546,
      "step": 1180
    },
    {
      "epoch": 2.4947589098532497,
      "grad_norm": 2.899369239807129,
      "learning_rate": 5.014675052410902e-05,
      "loss": 0.0576,
      "step": 1190
    },
    {
      "epoch": 2.5157232704402515,
      "grad_norm": 2.534893035888672,
      "learning_rate": 4.972746331236898e-05,
      "loss": 0.0529,
      "step": 1200
    },
    {
      "epoch": 2.5366876310272537,
      "grad_norm": 4.3690924644470215,
      "learning_rate": 4.9308176100628935e-05,
      "loss": 0.0494,
      "step": 1210
    },
    {
      "epoch": 2.5576519916142555,
      "grad_norm": 1.8138058185577393,
      "learning_rate": 4.888888888888889e-05,
      "loss": 0.0559,
      "step": 1220
    },
    {
      "epoch": 2.5786163522012577,
      "grad_norm": 1.9707310199737549,
      "learning_rate": 4.846960167714885e-05,
      "loss": 0.0544,
      "step": 1230
    },
    {
      "epoch": 2.59958071278826,
      "grad_norm": 2.480036973953247,
      "learning_rate": 4.805031446540881e-05,
      "loss": 0.0591,
      "step": 1240
    },
    {
      "epoch": 2.620545073375262,
      "grad_norm": 3.3198599815368652,
      "learning_rate": 4.763102725366876e-05,
      "loss": 0.0519,
      "step": 1250
    },
    {
      "epoch": 2.641509433962264,
      "grad_norm": 4.302845478057861,
      "learning_rate": 4.7211740041928724e-05,
      "loss": 0.0588,
      "step": 1260
    },
    {
      "epoch": 2.6624737945492662,
      "grad_norm": 2.769380807876587,
      "learning_rate": 4.679245283018868e-05,
      "loss": 0.0561,
      "step": 1270
    },
    {
      "epoch": 2.6834381551362685,
      "grad_norm": 2.0273640155792236,
      "learning_rate": 4.637316561844864e-05,
      "loss": 0.0531,
      "step": 1280
    },
    {
      "epoch": 2.7044025157232703,
      "grad_norm": 3.1066033840179443,
      "learning_rate": 4.59538784067086e-05,
      "loss": 0.0547,
      "step": 1290
    },
    {
      "epoch": 2.7253668763102725,
      "grad_norm": 3.7857868671417236,
      "learning_rate": 4.553459119496856e-05,
      "loss": 0.0552,
      "step": 1300
    },
    {
      "epoch": 2.7463312368972748,
      "grad_norm": 2.8125133514404297,
      "learning_rate": 4.511530398322851e-05,
      "loss": 0.0592,
      "step": 1310
    },
    {
      "epoch": 2.767295597484277,
      "grad_norm": 1.8335483074188232,
      "learning_rate": 4.469601677148847e-05,
      "loss": 0.0576,
      "step": 1320
    },
    {
      "epoch": 2.788259958071279,
      "grad_norm": 2.8290367126464844,
      "learning_rate": 4.427672955974843e-05,
      "loss": 0.0566,
      "step": 1330
    },
    {
      "epoch": 2.809224318658281,
      "grad_norm": 2.31840443611145,
      "learning_rate": 4.3857442348008385e-05,
      "loss": 0.0504,
      "step": 1340
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 2.372063159942627,
      "learning_rate": 4.343815513626835e-05,
      "loss": 0.0591,
      "step": 1350
    },
    {
      "epoch": 2.851153039832285,
      "grad_norm": 2.2945821285247803,
      "learning_rate": 4.301886792452831e-05,
      "loss": 0.0571,
      "step": 1360
    },
    {
      "epoch": 2.8721174004192873,
      "grad_norm": 1.8437625169754028,
      "learning_rate": 4.2599580712788264e-05,
      "loss": 0.0544,
      "step": 1370
    },
    {
      "epoch": 2.8930817610062896,
      "grad_norm": 2.612316608428955,
      "learning_rate": 4.218029350104822e-05,
      "loss": 0.0517,
      "step": 1380
    },
    {
      "epoch": 2.9140461215932913,
      "grad_norm": 2.7713685035705566,
      "learning_rate": 4.176100628930818e-05,
      "loss": 0.0565,
      "step": 1390
    },
    {
      "epoch": 2.9350104821802936,
      "grad_norm": 2.0892138481140137,
      "learning_rate": 4.1341719077568136e-05,
      "loss": 0.0591,
      "step": 1400
    },
    {
      "epoch": 2.9559748427672954,
      "grad_norm": 1.9455903768539429,
      "learning_rate": 4.092243186582809e-05,
      "loss": 0.0494,
      "step": 1410
    },
    {
      "epoch": 2.9769392033542976,
      "grad_norm": 3.191281318664551,
      "learning_rate": 4.050314465408805e-05,
      "loss": 0.0562,
      "step": 1420
    },
    {
      "epoch": 2.9979035639413,
      "grad_norm": 2.252624034881592,
      "learning_rate": 4.008385744234801e-05,
      "loss": 0.053,
      "step": 1430
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.667631983757019,
      "eval_runtime": 22.0522,
      "eval_samples_per_second": 68.02,
      "eval_steps_per_second": 8.525,
      "step": 1431
    },
    {
      "epoch": 3.018867924528302,
      "grad_norm": 1.2633923292160034,
      "learning_rate": 3.966457023060797e-05,
      "loss": 0.025,
      "step": 1440
    },
    {
      "epoch": 3.039832285115304,
      "grad_norm": 1.2608442306518555,
      "learning_rate": 3.924528301886793e-05,
      "loss": 0.0191,
      "step": 1450
    },
    {
      "epoch": 3.060796645702306,
      "grad_norm": 3.670048475265503,
      "learning_rate": 3.882599580712789e-05,
      "loss": 0.0231,
      "step": 1460
    },
    {
      "epoch": 3.0817610062893084,
      "grad_norm": 3.2211225032806396,
      "learning_rate": 3.840670859538784e-05,
      "loss": 0.0225,
      "step": 1470
    },
    {
      "epoch": 3.10272536687631,
      "grad_norm": 1.0529983043670654,
      "learning_rate": 3.7987421383647804e-05,
      "loss": 0.0242,
      "step": 1480
    },
    {
      "epoch": 3.1236897274633124,
      "grad_norm": 1.5843929052352905,
      "learning_rate": 3.756813417190776e-05,
      "loss": 0.0199,
      "step": 1490
    },
    {
      "epoch": 3.1446540880503147,
      "grad_norm": 1.28718900680542,
      "learning_rate": 3.7148846960167714e-05,
      "loss": 0.022,
      "step": 1500
    },
    {
      "epoch": 3.1656184486373165,
      "grad_norm": 1.1010375022888184,
      "learning_rate": 3.672955974842767e-05,
      "loss": 0.0185,
      "step": 1510
    },
    {
      "epoch": 3.1865828092243187,
      "grad_norm": 1.0108301639556885,
      "learning_rate": 3.631027253668763e-05,
      "loss": 0.0203,
      "step": 1520
    },
    {
      "epoch": 3.207547169811321,
      "grad_norm": 1.856557846069336,
      "learning_rate": 3.589098532494759e-05,
      "loss": 0.0188,
      "step": 1530
    },
    {
      "epoch": 3.2285115303983227,
      "grad_norm": 1.2554458379745483,
      "learning_rate": 3.547169811320755e-05,
      "loss": 0.0217,
      "step": 1540
    },
    {
      "epoch": 3.249475890985325,
      "grad_norm": 1.5458911657333374,
      "learning_rate": 3.505241090146751e-05,
      "loss": 0.0217,
      "step": 1550
    },
    {
      "epoch": 3.270440251572327,
      "grad_norm": 1.8719966411590576,
      "learning_rate": 3.4633123689727465e-05,
      "loss": 0.0239,
      "step": 1560
    },
    {
      "epoch": 3.291404612159329,
      "grad_norm": 0.8294363617897034,
      "learning_rate": 3.421383647798742e-05,
      "loss": 0.0174,
      "step": 1570
    },
    {
      "epoch": 3.3123689727463312,
      "grad_norm": 1.013922095298767,
      "learning_rate": 3.379454926624738e-05,
      "loss": 0.0191,
      "step": 1580
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 2.7998366355895996,
      "learning_rate": 3.337526205450734e-05,
      "loss": 0.0211,
      "step": 1590
    },
    {
      "epoch": 3.3542976939203353,
      "grad_norm": 1.3313286304473877,
      "learning_rate": 3.295597484276729e-05,
      "loss": 0.0214,
      "step": 1600
    },
    {
      "epoch": 3.3752620545073375,
      "grad_norm": 1.3300907611846924,
      "learning_rate": 3.2536687631027254e-05,
      "loss": 0.0229,
      "step": 1610
    },
    {
      "epoch": 3.3962264150943398,
      "grad_norm": 1.3926533460617065,
      "learning_rate": 3.2117400419287216e-05,
      "loss": 0.0192,
      "step": 1620
    },
    {
      "epoch": 3.4171907756813416,
      "grad_norm": 1.8201309442520142,
      "learning_rate": 3.169811320754717e-05,
      "loss": 0.0194,
      "step": 1630
    },
    {
      "epoch": 3.438155136268344,
      "grad_norm": 1.1363880634307861,
      "learning_rate": 3.127882599580713e-05,
      "loss": 0.0196,
      "step": 1640
    },
    {
      "epoch": 3.459119496855346,
      "grad_norm": 1.1997497081756592,
      "learning_rate": 3.085953878406709e-05,
      "loss": 0.0192,
      "step": 1650
    },
    {
      "epoch": 3.480083857442348,
      "grad_norm": 1.1632133722305298,
      "learning_rate": 3.0440251572327043e-05,
      "loss": 0.0165,
      "step": 1660
    },
    {
      "epoch": 3.50104821802935,
      "grad_norm": 1.141566514968872,
      "learning_rate": 3.0020964360587005e-05,
      "loss": 0.0182,
      "step": 1670
    },
    {
      "epoch": 3.5220125786163523,
      "grad_norm": 1.4071044921875,
      "learning_rate": 2.9601677148846964e-05,
      "loss": 0.0177,
      "step": 1680
    },
    {
      "epoch": 3.5429769392033545,
      "grad_norm": 0.6424669623374939,
      "learning_rate": 2.918238993710692e-05,
      "loss": 0.018,
      "step": 1690
    },
    {
      "epoch": 3.5639412997903563,
      "grad_norm": 1.240339756011963,
      "learning_rate": 2.8763102725366874e-05,
      "loss": 0.0181,
      "step": 1700
    },
    {
      "epoch": 3.5849056603773586,
      "grad_norm": 1.8148901462554932,
      "learning_rate": 2.8343815513626836e-05,
      "loss": 0.0203,
      "step": 1710
    },
    {
      "epoch": 3.6058700209643604,
      "grad_norm": 2.1308140754699707,
      "learning_rate": 2.7924528301886794e-05,
      "loss": 0.0208,
      "step": 1720
    },
    {
      "epoch": 3.6268343815513626,
      "grad_norm": 1.0319384336471558,
      "learning_rate": 2.750524109014675e-05,
      "loss": 0.0171,
      "step": 1730
    },
    {
      "epoch": 3.647798742138365,
      "grad_norm": 1.056301236152649,
      "learning_rate": 2.708595387840671e-05,
      "loss": 0.0193,
      "step": 1740
    },
    {
      "epoch": 3.668763102725367,
      "grad_norm": 1.0502632856369019,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.0181,
      "step": 1750
    },
    {
      "epoch": 3.689727463312369,
      "grad_norm": 1.5024515390396118,
      "learning_rate": 2.6247379454926625e-05,
      "loss": 0.0194,
      "step": 1760
    },
    {
      "epoch": 3.710691823899371,
      "grad_norm": 1.1332629919052124,
      "learning_rate": 2.5828092243186587e-05,
      "loss": 0.0154,
      "step": 1770
    },
    {
      "epoch": 3.731656184486373,
      "grad_norm": 1.3538895845413208,
      "learning_rate": 2.5408805031446542e-05,
      "loss": 0.0174,
      "step": 1780
    },
    {
      "epoch": 3.752620545073375,
      "grad_norm": 3.9446728229522705,
      "learning_rate": 2.49895178197065e-05,
      "loss": 0.0174,
      "step": 1790
    },
    {
      "epoch": 3.7735849056603774,
      "grad_norm": 1.6350576877593994,
      "learning_rate": 2.4570230607966456e-05,
      "loss": 0.0211,
      "step": 1800
    },
    {
      "epoch": 3.7945492662473796,
      "grad_norm": 0.998974621295929,
      "learning_rate": 2.4150943396226418e-05,
      "loss": 0.0185,
      "step": 1810
    },
    {
      "epoch": 3.8155136268343814,
      "grad_norm": 1.1701918840408325,
      "learning_rate": 2.3731656184486376e-05,
      "loss": 0.019,
      "step": 1820
    },
    {
      "epoch": 3.8364779874213837,
      "grad_norm": 1.005288004875183,
      "learning_rate": 2.331236897274633e-05,
      "loss": 0.0169,
      "step": 1830
    },
    {
      "epoch": 3.8574423480083855,
      "grad_norm": 1.4519301652908325,
      "learning_rate": 2.289308176100629e-05,
      "loss": 0.0191,
      "step": 1840
    },
    {
      "epoch": 3.8784067085953877,
      "grad_norm": 0.8834218382835388,
      "learning_rate": 2.2473794549266248e-05,
      "loss": 0.0171,
      "step": 1850
    },
    {
      "epoch": 3.89937106918239,
      "grad_norm": 1.3449293375015259,
      "learning_rate": 2.2054507337526207e-05,
      "loss": 0.0149,
      "step": 1860
    },
    {
      "epoch": 3.920335429769392,
      "grad_norm": 1.1032202243804932,
      "learning_rate": 2.1635220125786165e-05,
      "loss": 0.0175,
      "step": 1870
    },
    {
      "epoch": 3.941299790356394,
      "grad_norm": 0.9004182815551758,
      "learning_rate": 2.121593291404612e-05,
      "loss": 0.0152,
      "step": 1880
    },
    {
      "epoch": 3.9622641509433962,
      "grad_norm": 1.1716896295547485,
      "learning_rate": 2.0796645702306082e-05,
      "loss": 0.0189,
      "step": 1890
    },
    {
      "epoch": 3.9832285115303985,
      "grad_norm": 1.6147270202636719,
      "learning_rate": 2.037735849056604e-05,
      "loss": 0.0183,
      "step": 1900
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.6632634401321411,
      "eval_runtime": 22.7298,
      "eval_samples_per_second": 65.993,
      "eval_steps_per_second": 8.271,
      "step": 1908
    },
    {
      "epoch": 4.0041928721174,
      "grad_norm": 0.37951168417930603,
      "learning_rate": 1.9958071278825996e-05,
      "loss": 0.0163,
      "step": 1910
    },
    {
      "epoch": 4.0251572327044025,
      "grad_norm": 0.3050394654273987,
      "learning_rate": 1.9538784067085954e-05,
      "loss": 0.007,
      "step": 1920
    },
    {
      "epoch": 4.046121593291405,
      "grad_norm": 0.34913668036460876,
      "learning_rate": 1.9119496855345913e-05,
      "loss": 0.0064,
      "step": 1930
    },
    {
      "epoch": 4.067085953878407,
      "grad_norm": 0.36414623260498047,
      "learning_rate": 1.870020964360587e-05,
      "loss": 0.0067,
      "step": 1940
    },
    {
      "epoch": 4.088050314465409,
      "grad_norm": 0.5772935152053833,
      "learning_rate": 1.828092243186583e-05,
      "loss": 0.0061,
      "step": 1950
    },
    {
      "epoch": 4.109014675052411,
      "grad_norm": 0.2230810970067978,
      "learning_rate": 1.7861635220125785e-05,
      "loss": 0.0059,
      "step": 1960
    },
    {
      "epoch": 4.129979035639413,
      "grad_norm": 0.46709367632865906,
      "learning_rate": 1.7442348008385743e-05,
      "loss": 0.0063,
      "step": 1970
    },
    {
      "epoch": 4.150943396226415,
      "grad_norm": 0.40340206027030945,
      "learning_rate": 1.7023060796645705e-05,
      "loss": 0.006,
      "step": 1980
    },
    {
      "epoch": 4.171907756813417,
      "grad_norm": 0.22066733241081238,
      "learning_rate": 1.660377358490566e-05,
      "loss": 0.0055,
      "step": 1990
    },
    {
      "epoch": 4.1928721174004195,
      "grad_norm": 0.22712306678295135,
      "learning_rate": 1.618448637316562e-05,
      "loss": 0.0075,
      "step": 2000
    },
    {
      "epoch": 4.213836477987422,
      "grad_norm": 0.3246201276779175,
      "learning_rate": 1.5765199161425577e-05,
      "loss": 0.0059,
      "step": 2010
    },
    {
      "epoch": 4.234800838574423,
      "grad_norm": 0.6034521460533142,
      "learning_rate": 1.5345911949685536e-05,
      "loss": 0.0058,
      "step": 2020
    },
    {
      "epoch": 4.255765199161425,
      "grad_norm": 0.3003758490085602,
      "learning_rate": 1.4926624737945494e-05,
      "loss": 0.0059,
      "step": 2030
    },
    {
      "epoch": 4.276729559748428,
      "grad_norm": 0.38291364908218384,
      "learning_rate": 1.450733752620545e-05,
      "loss": 0.0053,
      "step": 2040
    },
    {
      "epoch": 4.29769392033543,
      "grad_norm": 0.24918225407600403,
      "learning_rate": 1.408805031446541e-05,
      "loss": 0.0055,
      "step": 2050
    },
    {
      "epoch": 4.318658280922432,
      "grad_norm": 0.29115813970565796,
      "learning_rate": 1.3668763102725368e-05,
      "loss": 0.0057,
      "step": 2060
    },
    {
      "epoch": 4.339622641509434,
      "grad_norm": 0.6018015146255493,
      "learning_rate": 1.3249475890985325e-05,
      "loss": 0.0056,
      "step": 2070
    },
    {
      "epoch": 4.360587002096436,
      "grad_norm": 0.4561573565006256,
      "learning_rate": 1.2830188679245283e-05,
      "loss": 0.0053,
      "step": 2080
    },
    {
      "epoch": 4.381551362683438,
      "grad_norm": 0.2961859703063965,
      "learning_rate": 1.2410901467505242e-05,
      "loss": 0.0058,
      "step": 2090
    },
    {
      "epoch": 4.40251572327044,
      "grad_norm": 0.7806200385093689,
      "learning_rate": 1.19916142557652e-05,
      "loss": 0.0051,
      "step": 2100
    },
    {
      "epoch": 4.423480083857442,
      "grad_norm": 0.2288259118795395,
      "learning_rate": 1.1572327044025157e-05,
      "loss": 0.0062,
      "step": 2110
    },
    {
      "epoch": 4.444444444444445,
      "grad_norm": 0.5480152368545532,
      "learning_rate": 1.1153039832285116e-05,
      "loss": 0.0054,
      "step": 2120
    },
    {
      "epoch": 4.465408805031447,
      "grad_norm": 0.3903510272502899,
      "learning_rate": 1.0733752620545073e-05,
      "loss": 0.0068,
      "step": 2130
    },
    {
      "epoch": 4.486373165618448,
      "grad_norm": 0.26396581530570984,
      "learning_rate": 1.0314465408805033e-05,
      "loss": 0.0051,
      "step": 2140
    },
    {
      "epoch": 4.5073375262054505,
      "grad_norm": 0.3533041477203369,
      "learning_rate": 9.89517819706499e-06,
      "loss": 0.0062,
      "step": 2150
    },
    {
      "epoch": 4.528301886792453,
      "grad_norm": 0.23736946284770966,
      "learning_rate": 9.475890985324948e-06,
      "loss": 0.0056,
      "step": 2160
    },
    {
      "epoch": 4.549266247379455,
      "grad_norm": 0.7073323726654053,
      "learning_rate": 9.056603773584905e-06,
      "loss": 0.0054,
      "step": 2170
    },
    {
      "epoch": 4.570230607966457,
      "grad_norm": 0.22164012491703033,
      "learning_rate": 8.637316561844865e-06,
      "loss": 0.0055,
      "step": 2180
    },
    {
      "epoch": 4.591194968553459,
      "grad_norm": 0.27965155243873596,
      "learning_rate": 8.218029350104822e-06,
      "loss": 0.0048,
      "step": 2190
    },
    {
      "epoch": 4.612159329140461,
      "grad_norm": 0.7329670786857605,
      "learning_rate": 7.79874213836478e-06,
      "loss": 0.0057,
      "step": 2200
    },
    {
      "epoch": 4.633123689727463,
      "grad_norm": 0.310923308134079,
      "learning_rate": 7.379454926624739e-06,
      "loss": 0.0054,
      "step": 2210
    },
    {
      "epoch": 4.654088050314465,
      "grad_norm": 0.19455011188983917,
      "learning_rate": 6.9601677148846965e-06,
      "loss": 0.0072,
      "step": 2220
    },
    {
      "epoch": 4.6750524109014675,
      "grad_norm": 0.257545530796051,
      "learning_rate": 6.540880503144654e-06,
      "loss": 0.0057,
      "step": 2230
    },
    {
      "epoch": 4.69601677148847,
      "grad_norm": 1.0093029737472534,
      "learning_rate": 6.121593291404613e-06,
      "loss": 0.006,
      "step": 2240
    },
    {
      "epoch": 4.716981132075472,
      "grad_norm": 0.4301735758781433,
      "learning_rate": 5.70230607966457e-06,
      "loss": 0.0056,
      "step": 2250
    },
    {
      "epoch": 4.737945492662474,
      "grad_norm": 0.18497976660728455,
      "learning_rate": 5.283018867924529e-06,
      "loss": 0.0052,
      "step": 2260
    },
    {
      "epoch": 4.758909853249476,
      "grad_norm": 0.21521350741386414,
      "learning_rate": 4.8637316561844865e-06,
      "loss": 0.0055,
      "step": 2270
    },
    {
      "epoch": 4.779874213836478,
      "grad_norm": 0.20319782197475433,
      "learning_rate": 4.444444444444445e-06,
      "loss": 0.0058,
      "step": 2280
    },
    {
      "epoch": 4.80083857442348,
      "grad_norm": 0.2288380116224289,
      "learning_rate": 4.025157232704403e-06,
      "loss": 0.005,
      "step": 2290
    },
    {
      "epoch": 4.821802935010482,
      "grad_norm": 1.407477855682373,
      "learning_rate": 3.6058700209643607e-06,
      "loss": 0.0073,
      "step": 2300
    },
    {
      "epoch": 4.8427672955974845,
      "grad_norm": 0.26325657963752747,
      "learning_rate": 3.1865828092243184e-06,
      "loss": 0.0057,
      "step": 2310
    },
    {
      "epoch": 4.863731656184486,
      "grad_norm": 0.4354378283023834,
      "learning_rate": 2.767295597484277e-06,
      "loss": 0.006,
      "step": 2320
    },
    {
      "epoch": 4.884696016771488,
      "grad_norm": 0.7161938548088074,
      "learning_rate": 2.348008385744235e-06,
      "loss": 0.0055,
      "step": 2330
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 1.1639270782470703,
      "learning_rate": 1.928721174004193e-06,
      "loss": 0.0056,
      "step": 2340
    },
    {
      "epoch": 4.926624737945493,
      "grad_norm": 0.3333396017551422,
      "learning_rate": 1.509433962264151e-06,
      "loss": 0.0061,
      "step": 2350
    },
    {
      "epoch": 4.947589098532495,
      "grad_norm": 0.3744851052761078,
      "learning_rate": 1.090146750524109e-06,
      "loss": 0.0056,
      "step": 2360
    },
    {
      "epoch": 4.968553459119497,
      "grad_norm": 0.1569022685289383,
      "learning_rate": 6.70859538784067e-07,
      "loss": 0.0052,
      "step": 2370
    },
    {
      "epoch": 4.989517819706499,
      "grad_norm": 0.30756279826164246,
      "learning_rate": 2.5157232704402517e-07,
      "loss": 0.0056,
      "step": 2380
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.6631835699081421,
      "eval_runtime": 22.601,
      "eval_samples_per_second": 66.369,
      "eval_steps_per_second": 8.318,
      "step": 2385
    }
  ],
  "logging_steps": 10,
  "max_steps": 2385,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "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": 1.8785440290816e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}