Files
wanna-pilot/checkpoint-46878/trainer_state.json
ModelHub XC 9d63e58716 初始化项目,由ModelHub XC社区提供模型
Model: bbunzeck/wanna-pilot
Source: Original Platform
2026-07-30 01:37:18 +08:00

32892 lines
813 KiB
JSON

{
"best_global_step": 46878,
"best_metric": 1.9871164560317993,
"best_model_checkpoint": "/mnt/myscratch/model/checkpoint-46878",
"epoch": 6.0,
"eval_steps": 500,
"global_step": 46878,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.00128,
"grad_norm": 5.885704040527344,
"learning_rate": 1.3499999999999998e-05,
"loss": 6.605717468261719,
"step": 10
},
{
"epoch": 0.00256,
"grad_norm": 4.186084747314453,
"learning_rate": 2.8499999999999998e-05,
"loss": 5.801454544067383,
"step": 20
},
{
"epoch": 0.00384,
"grad_norm": 2.714238166809082,
"learning_rate": 4.3499999999999993e-05,
"loss": 4.964897155761719,
"step": 30
},
{
"epoch": 0.00512,
"grad_norm": 2.2332475185394287,
"learning_rate": 5.85e-05,
"loss": 4.384710693359375,
"step": 40
},
{
"epoch": 0.0064,
"grad_norm": 2.1372387409210205,
"learning_rate": 7.35e-05,
"loss": 3.9424991607666016,
"step": 50
},
{
"epoch": 0.00768,
"grad_norm": 2.8777883052825928,
"learning_rate": 8.849999999999998e-05,
"loss": 3.6049922943115233,
"step": 60
},
{
"epoch": 0.00896,
"grad_norm": 2.3045167922973633,
"learning_rate": 0.00010349999999999998,
"loss": 3.2790637969970704,
"step": 70
},
{
"epoch": 0.01024,
"grad_norm": 3.263000011444092,
"learning_rate": 0.0001185,
"loss": 3.0744760513305662,
"step": 80
},
{
"epoch": 0.01152,
"grad_norm": 2.0662527084350586,
"learning_rate": 0.0001335,
"loss": 2.8297033309936523,
"step": 90
},
{
"epoch": 0.0128,
"grad_norm": 2.5418498516082764,
"learning_rate": 0.00014849999999999998,
"loss": 2.659013366699219,
"step": 100
},
{
"epoch": 0.01408,
"grad_norm": 1.6310275793075562,
"learning_rate": 0.0001635,
"loss": 2.49525146484375,
"step": 110
},
{
"epoch": 0.01536,
"grad_norm": 2.0985260009765625,
"learning_rate": 0.00017849999999999997,
"loss": 2.42862491607666,
"step": 120
},
{
"epoch": 0.01664,
"grad_norm": 1.5091782808303833,
"learning_rate": 0.0001935,
"loss": 2.3671804428100587,
"step": 130
},
{
"epoch": 0.01792,
"grad_norm": 1.6089136600494385,
"learning_rate": 0.00020849999999999997,
"loss": 2.261024284362793,
"step": 140
},
{
"epoch": 0.0192,
"grad_norm": 1.0941425561904907,
"learning_rate": 0.00022349999999999998,
"loss": 2.221732330322266,
"step": 150
},
{
"epoch": 0.02048,
"grad_norm": 1.2431774139404297,
"learning_rate": 0.0002385,
"loss": 2.2301746368408204,
"step": 160
},
{
"epoch": 0.02176,
"grad_norm": 1.1251002550125122,
"learning_rate": 0.0002535,
"loss": 2.204315948486328,
"step": 170
},
{
"epoch": 0.02304,
"grad_norm": 1.195177435874939,
"learning_rate": 0.00026849999999999997,
"loss": 2.170834541320801,
"step": 180
},
{
"epoch": 0.02432,
"grad_norm": 1.409263014793396,
"learning_rate": 0.00028349999999999995,
"loss": 2.1824182510375976,
"step": 190
},
{
"epoch": 0.0256,
"grad_norm": 1.04000985622406,
"learning_rate": 0.0002985,
"loss": 2.1573196411132813,
"step": 200
},
{
"epoch": 0.02688,
"grad_norm": 1.1753369569778442,
"learning_rate": 0.0002999999901272834,
"loss": 2.1714319229125976,
"step": 210
},
{
"epoch": 0.02816,
"grad_norm": 1.028172492980957,
"learning_rate": 0.0002999999559993758,
"loss": 2.1164165496826173,
"step": 220
},
{
"epoch": 0.02944,
"grad_norm": 1.2896201610565186,
"learning_rate": 0.00029999989749439744,
"loss": 2.112924575805664,
"step": 230
},
{
"epoch": 0.03072,
"grad_norm": 1.0579850673675537,
"learning_rate": 0.00029999981461235783,
"loss": 2.114685821533203,
"step": 240
},
{
"epoch": 0.032,
"grad_norm": 0.8515546321868896,
"learning_rate": 0.00029999970735327037,
"loss": 2.0852130889892577,
"step": 250
},
{
"epoch": 0.03328,
"grad_norm": 1.0059428215026855,
"learning_rate": 0.0002999995757171525,
"loss": 2.0961055755615234,
"step": 260
},
{
"epoch": 0.03456,
"grad_norm": 0.8511396050453186,
"learning_rate": 0.0002999994197040256,
"loss": 2.0771358489990233,
"step": 270
},
{
"epoch": 0.03584,
"grad_norm": 0.9550066590309143,
"learning_rate": 0.0002999992393139151,
"loss": 2.070794677734375,
"step": 280
},
{
"epoch": 0.03712,
"grad_norm": 0.9620731472969055,
"learning_rate": 0.0002999990345468503,
"loss": 2.059033203125,
"step": 290
},
{
"epoch": 0.0384,
"grad_norm": 1.1329026222229004,
"learning_rate": 0.0002999988054028644,
"loss": 2.0862873077392576,
"step": 300
},
{
"epoch": 0.03968,
"grad_norm": 0.8112749457359314,
"learning_rate": 0.00029999855188199474,
"loss": 2.057857704162598,
"step": 310
},
{
"epoch": 0.04096,
"grad_norm": 0.9323590397834778,
"learning_rate": 0.0002999982739842824,
"loss": 2.0480510711669924,
"step": 320
},
{
"epoch": 0.04224,
"grad_norm": 0.8546434640884399,
"learning_rate": 0.0002999979717097727,
"loss": 2.067165756225586,
"step": 330
},
{
"epoch": 0.04352,
"grad_norm": 1.002989411354065,
"learning_rate": 0.00029999764505851456,
"loss": 2.0613639831542967,
"step": 340
},
{
"epoch": 0.0448,
"grad_norm": 0.8182193636894226,
"learning_rate": 0.00029999729403056127,
"loss": 2.0515058517456053,
"step": 350
},
{
"epoch": 0.04608,
"grad_norm": 0.7834963202476501,
"learning_rate": 0.00029999691862596975,
"loss": 2.046108436584473,
"step": 360
},
{
"epoch": 0.04736,
"grad_norm": 0.9317572712898254,
"learning_rate": 0.000299996518844801,
"loss": 2.0235712051391603,
"step": 370
},
{
"epoch": 0.04864,
"grad_norm": 0.8493145108222961,
"learning_rate": 0.0002999960946871201,
"loss": 2.0325557708740236,
"step": 380
},
{
"epoch": 0.04992,
"grad_norm": 0.922864556312561,
"learning_rate": 0.0002999956461529959,
"loss": 2.0653444290161134,
"step": 390
},
{
"epoch": 0.0512,
"grad_norm": 1.0530719757080078,
"learning_rate": 0.0002999951732425013,
"loss": 2.0602413177490235,
"step": 400
},
{
"epoch": 0.05248,
"grad_norm": 0.8999637961387634,
"learning_rate": 0.0002999946759557132,
"loss": 2.0521976470947267,
"step": 410
},
{
"epoch": 0.05376,
"grad_norm": 0.8567571640014648,
"learning_rate": 0.00029999415429271233,
"loss": 2.042643928527832,
"step": 420
},
{
"epoch": 0.05504,
"grad_norm": 0.9853494167327881,
"learning_rate": 0.00029999360825358354,
"loss": 2.043178939819336,
"step": 430
},
{
"epoch": 0.05632,
"grad_norm": 0.9435704946517944,
"learning_rate": 0.00029999303783841553,
"loss": 2.0274379730224608,
"step": 440
},
{
"epoch": 0.0576,
"grad_norm": 0.9081392288208008,
"learning_rate": 0.00029999244304730106,
"loss": 2.0362335205078126,
"step": 450
},
{
"epoch": 0.05888,
"grad_norm": 0.7286208868026733,
"learning_rate": 0.0002999918238803367,
"loss": 2.0380199432373045,
"step": 460
},
{
"epoch": 0.06016,
"grad_norm": 0.8928585648536682,
"learning_rate": 0.00029999118033762314,
"loss": 2.031109428405762,
"step": 470
},
{
"epoch": 0.06144,
"grad_norm": 0.8266701698303223,
"learning_rate": 0.00029999051241926493,
"loss": 2.0246667861938477,
"step": 480
},
{
"epoch": 0.06272,
"grad_norm": 0.7270641922950745,
"learning_rate": 0.00029998982012537063,
"loss": 2.042285346984863,
"step": 490
},
{
"epoch": 0.064,
"grad_norm": 0.7794978022575378,
"learning_rate": 0.0002999891034560528,
"loss": 2.0380592346191406,
"step": 500
},
{
"epoch": 0.06528,
"grad_norm": 0.8687776327133179,
"learning_rate": 0.00029998836241142786,
"loss": 2.0340011596679686,
"step": 510
},
{
"epoch": 0.06656,
"grad_norm": 0.8388617634773254,
"learning_rate": 0.0002999875969916162,
"loss": 2.035272979736328,
"step": 520
},
{
"epoch": 0.06784,
"grad_norm": 0.8601788878440857,
"learning_rate": 0.00029998680719674225,
"loss": 2.0376710891723633,
"step": 530
},
{
"epoch": 0.06912,
"grad_norm": 0.872636616230011,
"learning_rate": 0.00029998599302693434,
"loss": 2.0254486083984373,
"step": 540
},
{
"epoch": 0.0704,
"grad_norm": 0.7168036699295044,
"learning_rate": 0.00029998515448232484,
"loss": 2.044236755371094,
"step": 550
},
{
"epoch": 0.07168,
"grad_norm": 0.7704122066497803,
"learning_rate": 0.00029998429156304995,
"loss": 2.0272247314453127,
"step": 560
},
{
"epoch": 0.07296,
"grad_norm": 0.7757558822631836,
"learning_rate": 0.00029998340426925,
"loss": 2.0317691802978515,
"step": 570
},
{
"epoch": 0.07424,
"grad_norm": 0.9410697817802429,
"learning_rate": 0.0002999824926010691,
"loss": 2.0405961990356447,
"step": 580
},
{
"epoch": 0.07552,
"grad_norm": 0.7462645173072815,
"learning_rate": 0.0002999815565586555,
"loss": 2.0344850540161135,
"step": 590
},
{
"epoch": 0.0768,
"grad_norm": 0.8859974145889282,
"learning_rate": 0.00029998059614216123,
"loss": 2.0125873565673826,
"step": 600
},
{
"epoch": 0.07808,
"grad_norm": 0.724909782409668,
"learning_rate": 0.00029997961135174236,
"loss": 2.03349609375,
"step": 610
},
{
"epoch": 0.07936,
"grad_norm": 0.7383590936660767,
"learning_rate": 0.000299978602187559,
"loss": 2.033146095275879,
"step": 620
},
{
"epoch": 0.08064,
"grad_norm": 1.3143054246902466,
"learning_rate": 0.00029997756864977515,
"loss": 2.0456708908081054,
"step": 630
},
{
"epoch": 0.08192,
"grad_norm": 0.9198774099349976,
"learning_rate": 0.0002999765107385588,
"loss": 2.029145050048828,
"step": 640
},
{
"epoch": 0.0832,
"grad_norm": 0.7761960625648499,
"learning_rate": 0.00029997542845408174,
"loss": 2.040088653564453,
"step": 650
},
{
"epoch": 0.08448,
"grad_norm": 0.7530298829078674,
"learning_rate": 0.00029997432179652,
"loss": 2.0398796081542967,
"step": 660
},
{
"epoch": 0.08576,
"grad_norm": 0.7231208682060242,
"learning_rate": 0.0002999731907660533,
"loss": 2.029999542236328,
"step": 670
},
{
"epoch": 0.08704,
"grad_norm": 0.7728089094161987,
"learning_rate": 0.00029997203536286565,
"loss": 2.036482810974121,
"step": 680
},
{
"epoch": 0.08832,
"grad_norm": 0.607453465461731,
"learning_rate": 0.0002999708555871446,
"loss": 2.017901611328125,
"step": 690
},
{
"epoch": 0.0896,
"grad_norm": 0.779467761516571,
"learning_rate": 0.00029996965143908196,
"loss": 2.030159759521484,
"step": 700
},
{
"epoch": 0.09088,
"grad_norm": 0.7885509133338928,
"learning_rate": 0.00029996842291887344,
"loss": 2.0309608459472654,
"step": 710
},
{
"epoch": 0.09216,
"grad_norm": 0.7816950678825378,
"learning_rate": 0.0002999671700267187,
"loss": 1.9948287963867188,
"step": 720
},
{
"epoch": 0.09344,
"grad_norm": 0.806575357913971,
"learning_rate": 0.00029996589276282137,
"loss": 2.044095039367676,
"step": 730
},
{
"epoch": 0.09472,
"grad_norm": 0.7333192825317383,
"learning_rate": 0.0002999645911273889,
"loss": 2.038722610473633,
"step": 740
},
{
"epoch": 0.096,
"grad_norm": 0.8800010085105896,
"learning_rate": 0.00029996326512063295,
"loss": 2.035386085510254,
"step": 750
},
{
"epoch": 0.09728,
"grad_norm": 0.7811752557754517,
"learning_rate": 0.000299961914742769,
"loss": 2.004504585266113,
"step": 760
},
{
"epoch": 0.09856,
"grad_norm": 0.7882314920425415,
"learning_rate": 0.00029996053999401646,
"loss": 2.0211507797241213,
"step": 770
},
{
"epoch": 0.09984,
"grad_norm": 0.698653519153595,
"learning_rate": 0.0002999591408745988,
"loss": 2.0169116973876955,
"step": 780
},
{
"epoch": 0.10112,
"grad_norm": 0.6910223960876465,
"learning_rate": 0.00029995771738474335,
"loss": 2.002583694458008,
"step": 790
},
{
"epoch": 0.1024,
"grad_norm": 1.0781288146972656,
"learning_rate": 0.0002999562695246814,
"loss": 2.040352630615234,
"step": 800
},
{
"epoch": 0.10368,
"grad_norm": 0.7604212760925293,
"learning_rate": 0.0002999547972946484,
"loss": 2.012590789794922,
"step": 810
},
{
"epoch": 0.10496,
"grad_norm": 0.8433690071105957,
"learning_rate": 0.0002999533006948835,
"loss": 2.0214279174804686,
"step": 820
},
{
"epoch": 0.10624,
"grad_norm": 0.7420530319213867,
"learning_rate": 0.00029995177972562984,
"loss": 2.01849422454834,
"step": 830
},
{
"epoch": 0.10752,
"grad_norm": 0.8684757351875305,
"learning_rate": 0.00029995023438713475,
"loss": 2.0245651245117187,
"step": 840
},
{
"epoch": 0.1088,
"grad_norm": 0.8758244514465332,
"learning_rate": 0.0002999486646796493,
"loss": 2.0273792266845705,
"step": 850
},
{
"epoch": 0.11008,
"grad_norm": 0.7705071568489075,
"learning_rate": 0.0002999470706034286,
"loss": 2.003208351135254,
"step": 860
},
{
"epoch": 0.11136,
"grad_norm": 0.7473205327987671,
"learning_rate": 0.0002999454521587317,
"loss": 2.0413352966308596,
"step": 870
},
{
"epoch": 0.11264,
"grad_norm": 0.8020169138908386,
"learning_rate": 0.0002999438093458217,
"loss": 2.043915939331055,
"step": 880
},
{
"epoch": 0.11392,
"grad_norm": 0.8231315016746521,
"learning_rate": 0.00029994214216496544,
"loss": 2.0179935455322267,
"step": 890
},
{
"epoch": 0.1152,
"grad_norm": 0.8677017092704773,
"learning_rate": 0.00029994045061643393,
"loss": 2.0080083847045898,
"step": 900
},
{
"epoch": 0.11648,
"grad_norm": 0.7265374660491943,
"learning_rate": 0.0002999387347005021,
"loss": 2.0255220413208006,
"step": 910
},
{
"epoch": 0.11776,
"grad_norm": 0.7804266810417175,
"learning_rate": 0.0002999369944174487,
"loss": 2.0323333740234375,
"step": 920
},
{
"epoch": 0.11904,
"grad_norm": 1.909376859664917,
"learning_rate": 0.00029993522976755674,
"loss": 2.0403774261474608,
"step": 930
},
{
"epoch": 0.12032,
"grad_norm": 0.7820098996162415,
"learning_rate": 0.00029993344075111286,
"loss": 2.0211090087890624,
"step": 940
},
{
"epoch": 0.1216,
"grad_norm": 0.8035023808479309,
"learning_rate": 0.0002999316273684078,
"loss": 2.0083597183227537,
"step": 950
},
{
"epoch": 0.12288,
"grad_norm": 0.8145981431007385,
"learning_rate": 0.0002999297896197363,
"loss": 2.0231433868408204,
"step": 960
},
{
"epoch": 0.12416,
"grad_norm": 0.6948875188827515,
"learning_rate": 0.000299927927505397,
"loss": 2.0229448318481444,
"step": 970
},
{
"epoch": 0.12544,
"grad_norm": 0.7178237438201904,
"learning_rate": 0.00029992604102569256,
"loss": 2.024074363708496,
"step": 980
},
{
"epoch": 0.12672,
"grad_norm": 0.7583261132240295,
"learning_rate": 0.0002999241301809295,
"loss": 2.0232542037963865,
"step": 990
},
{
"epoch": 0.128,
"grad_norm": 0.9775397181510925,
"learning_rate": 0.00029992219497141834,
"loss": 2.0318603515625,
"step": 1000
},
{
"epoch": 0.12928,
"grad_norm": 0.7427451014518738,
"learning_rate": 0.0002999202353974737,
"loss": 2.0150304794311524,
"step": 1010
},
{
"epoch": 0.13056,
"grad_norm": 0.6991038918495178,
"learning_rate": 0.00029991825145941393,
"loss": 2.0047382354736327,
"step": 1020
},
{
"epoch": 0.13184,
"grad_norm": 0.7759991884231567,
"learning_rate": 0.0002999162431575615,
"loss": 2.0170944213867186,
"step": 1030
},
{
"epoch": 0.13312,
"grad_norm": 0.8461285829544067,
"learning_rate": 0.0002999142104922427,
"loss": 2.0140987396240235,
"step": 1040
},
{
"epoch": 0.1344,
"grad_norm": 0.872622549533844,
"learning_rate": 0.000299912153463788,
"loss": 2.0137298583984373,
"step": 1050
},
{
"epoch": 0.13568,
"grad_norm": 0.6421040892601013,
"learning_rate": 0.0002999100720725316,
"loss": 2.0034528732299806,
"step": 1060
},
{
"epoch": 0.13696,
"grad_norm": 0.7797923684120178,
"learning_rate": 0.0002999079663188118,
"loss": 2.0397003173828123,
"step": 1070
},
{
"epoch": 0.13824,
"grad_norm": 0.6871171593666077,
"learning_rate": 0.0002999058362029708,
"loss": 2.0207529067993164,
"step": 1080
},
{
"epoch": 0.13952,
"grad_norm": 0.725261390209198,
"learning_rate": 0.00029990368172535473,
"loss": 2.0161842346191405,
"step": 1090
},
{
"epoch": 0.1408,
"grad_norm": 0.766487717628479,
"learning_rate": 0.0002999015028863138,
"loss": 2.0156953811645506,
"step": 1100
},
{
"epoch": 0.14208,
"grad_norm": 0.6803558468818665,
"learning_rate": 0.000299899299686202,
"loss": 2.029312515258789,
"step": 1110
},
{
"epoch": 0.14336,
"grad_norm": 0.8679984211921692,
"learning_rate": 0.0002998970721253776,
"loss": 2.003293991088867,
"step": 1120
},
{
"epoch": 0.14464,
"grad_norm": 0.8779763579368591,
"learning_rate": 0.0002998948202042023,
"loss": 2.0375221252441404,
"step": 1130
},
{
"epoch": 0.14592,
"grad_norm": 0.7275505065917969,
"learning_rate": 0.00029989254392304225,
"loss": 1.998856544494629,
"step": 1140
},
{
"epoch": 0.1472,
"grad_norm": 0.7902148962020874,
"learning_rate": 0.0002998902432822674,
"loss": 2.0269426345825194,
"step": 1150
},
{
"epoch": 0.14848,
"grad_norm": 0.7712279558181763,
"learning_rate": 0.00029988791828225155,
"loss": 2.010341835021973,
"step": 1160
},
{
"epoch": 0.14976,
"grad_norm": 0.7030239701271057,
"learning_rate": 0.0002998855689233726,
"loss": 2.005113983154297,
"step": 1170
},
{
"epoch": 0.15104,
"grad_norm": 0.6882237195968628,
"learning_rate": 0.00029988319520601235,
"loss": 2.0123071670532227,
"step": 1180
},
{
"epoch": 0.15232,
"grad_norm": 0.711328387260437,
"learning_rate": 0.0002998807971305565,
"loss": 2.019595718383789,
"step": 1190
},
{
"epoch": 0.1536,
"grad_norm": 0.8122184872627258,
"learning_rate": 0.0002998783746973948,
"loss": 2.018142509460449,
"step": 1200
},
{
"epoch": 0.15488,
"grad_norm": 0.8645702004432678,
"learning_rate": 0.000299875927906921,
"loss": 2.0178430557250975,
"step": 1210
},
{
"epoch": 0.15616,
"grad_norm": 0.84963458776474,
"learning_rate": 0.0002998734567595327,
"loss": 2.0299404144287108,
"step": 1220
},
{
"epoch": 0.15744,
"grad_norm": 0.7000954151153564,
"learning_rate": 0.0002998709612556314,
"loss": 2.008827781677246,
"step": 1230
},
{
"epoch": 0.15872,
"grad_norm": 0.7619990110397339,
"learning_rate": 0.0002998684413956228,
"loss": 2.0060449600219727,
"step": 1240
},
{
"epoch": 0.16,
"grad_norm": 0.7732600569725037,
"learning_rate": 0.0002998658971799163,
"loss": 2.0067914962768554,
"step": 1250
},
{
"epoch": 0.16128,
"grad_norm": 0.7497115135192871,
"learning_rate": 0.0002998633286089255,
"loss": 2.006074142456055,
"step": 1260
},
{
"epoch": 0.16256,
"grad_norm": 0.8291988372802734,
"learning_rate": 0.0002998607356830677,
"loss": 2.0243513107299806,
"step": 1270
},
{
"epoch": 0.16384,
"grad_norm": 0.7676268219947815,
"learning_rate": 0.0002998581184027642,
"loss": 2.009158706665039,
"step": 1280
},
{
"epoch": 0.16512,
"grad_norm": 0.6954494714736938,
"learning_rate": 0.0002998554767684406,
"loss": 2.0111303329467773,
"step": 1290
},
{
"epoch": 0.1664,
"grad_norm": 0.8298436999320984,
"learning_rate": 0.0002998528107805261,
"loss": 2.024542236328125,
"step": 1300
},
{
"epoch": 0.16768,
"grad_norm": 0.783103883266449,
"learning_rate": 0.00029985012043945387,
"loss": 2.0101518630981445,
"step": 1310
},
{
"epoch": 0.16896,
"grad_norm": 0.8619032502174377,
"learning_rate": 0.00029984740574566125,
"loss": 2.018630027770996,
"step": 1320
},
{
"epoch": 0.17024,
"grad_norm": 0.7219138145446777,
"learning_rate": 0.0002998446666995893,
"loss": 1.9975326538085938,
"step": 1330
},
{
"epoch": 0.17152,
"grad_norm": 0.7052103281021118,
"learning_rate": 0.00029984190330168325,
"loss": 2.0083797454833983,
"step": 1340
},
{
"epoch": 0.1728,
"grad_norm": 1.1350038051605225,
"learning_rate": 0.00029983911555239213,
"loss": 2.011388397216797,
"step": 1350
},
{
"epoch": 0.17408,
"grad_norm": 0.78801429271698,
"learning_rate": 0.00029983630345216904,
"loss": 2.008464050292969,
"step": 1360
},
{
"epoch": 0.17536,
"grad_norm": 0.7578643560409546,
"learning_rate": 0.0002998334670014709,
"loss": 2.025455284118652,
"step": 1370
},
{
"epoch": 0.17664,
"grad_norm": 0.9882360100746155,
"learning_rate": 0.00029983060620075876,
"loss": 2.0122611999511717,
"step": 1380
},
{
"epoch": 0.17792,
"grad_norm": 0.7348957061767578,
"learning_rate": 0.00029982772105049754,
"loss": 2.018396759033203,
"step": 1390
},
{
"epoch": 0.1792,
"grad_norm": 0.7000976800918579,
"learning_rate": 0.00029982481155115605,
"loss": 2.0135793685913086,
"step": 1400
},
{
"epoch": 0.18048,
"grad_norm": 0.8569319248199463,
"learning_rate": 0.0002998218777032072,
"loss": 2.0004940032958984,
"step": 1410
},
{
"epoch": 0.18176,
"grad_norm": 0.7729265689849854,
"learning_rate": 0.0002998189195071277,
"loss": 1.9858884811401367,
"step": 1420
},
{
"epoch": 0.18304,
"grad_norm": 0.6754656434059143,
"learning_rate": 0.00029981593696339835,
"loss": 2.004491424560547,
"step": 1430
},
{
"epoch": 0.18432,
"grad_norm": 0.7580174803733826,
"learning_rate": 0.0002998129300725038,
"loss": 2.0258563995361327,
"step": 1440
},
{
"epoch": 0.1856,
"grad_norm": 0.8150157928466797,
"learning_rate": 0.0002998098988349328,
"loss": 2.0141185760498046,
"step": 1450
},
{
"epoch": 0.18688,
"grad_norm": 0.8222833275794983,
"learning_rate": 0.00029980684325117793,
"loss": 1.9962928771972657,
"step": 1460
},
{
"epoch": 0.18816,
"grad_norm": 0.7298377156257629,
"learning_rate": 0.00029980376332173575,
"loss": 2.0095237731933593,
"step": 1470
},
{
"epoch": 0.18944,
"grad_norm": 0.7222643494606018,
"learning_rate": 0.00029980065904710673,
"loss": 2.016450119018555,
"step": 1480
},
{
"epoch": 0.19072,
"grad_norm": 0.7348253726959229,
"learning_rate": 0.0002997975304277955,
"loss": 1.9989503860473632,
"step": 1490
},
{
"epoch": 0.192,
"grad_norm": 0.7901546359062195,
"learning_rate": 0.00029979437746431044,
"loss": 2.017726516723633,
"step": 1500
},
{
"epoch": 0.19328,
"grad_norm": 0.7217196822166443,
"learning_rate": 0.0002997912001571639,
"loss": 2.002857208251953,
"step": 1510
},
{
"epoch": 0.19456,
"grad_norm": 0.7372826337814331,
"learning_rate": 0.00029978799850687227,
"loss": 2.018671989440918,
"step": 1520
},
{
"epoch": 0.19584,
"grad_norm": 0.8103359937667847,
"learning_rate": 0.0002997847725139559,
"loss": 2.0113971710205076,
"step": 1530
},
{
"epoch": 0.19712,
"grad_norm": 0.7941066026687622,
"learning_rate": 0.000299781522178939,
"loss": 1.9965641021728515,
"step": 1540
},
{
"epoch": 0.1984,
"grad_norm": 0.7437220215797424,
"learning_rate": 0.00029977824750234983,
"loss": 2.0001338958740233,
"step": 1550
},
{
"epoch": 0.19968,
"grad_norm": 0.8794814348220825,
"learning_rate": 0.0002997749484847206,
"loss": 2.023735237121582,
"step": 1560
},
{
"epoch": 0.20096,
"grad_norm": 0.7125862836837769,
"learning_rate": 0.0002997716251265873,
"loss": 2.0026830673217773,
"step": 1570
},
{
"epoch": 0.20224,
"grad_norm": 0.729263961315155,
"learning_rate": 0.00029976827742849024,
"loss": 2.005397415161133,
"step": 1580
},
{
"epoch": 0.20352,
"grad_norm": 0.7843178510665894,
"learning_rate": 0.00029976490539097335,
"loss": 2.027159881591797,
"step": 1590
},
{
"epoch": 0.2048,
"grad_norm": 0.7313951253890991,
"learning_rate": 0.00029976150901458456,
"loss": 2.0337509155273437,
"step": 1600
},
{
"epoch": 0.20608,
"grad_norm": 0.7452402114868164,
"learning_rate": 0.00029975808829987594,
"loss": 1.9958429336547852,
"step": 1610
},
{
"epoch": 0.20736,
"grad_norm": 0.7813226580619812,
"learning_rate": 0.0002997546432474034,
"loss": 1.9967842102050781,
"step": 1620
},
{
"epoch": 0.20864,
"grad_norm": 0.6847978830337524,
"learning_rate": 0.0002997511738577268,
"loss": 2.0223621368408202,
"step": 1630
},
{
"epoch": 0.20992,
"grad_norm": 0.710649311542511,
"learning_rate": 0.00029974768013140985,
"loss": 2.0229820251464843,
"step": 1640
},
{
"epoch": 0.2112,
"grad_norm": 0.7057698965072632,
"learning_rate": 0.0002997441620690205,
"loss": 2.0189645767211912,
"step": 1650
},
{
"epoch": 0.21248,
"grad_norm": 1.0434540510177612,
"learning_rate": 0.00029974061967113043,
"loss": 2.0161455154418944,
"step": 1660
},
{
"epoch": 0.21376,
"grad_norm": 0.7728395462036133,
"learning_rate": 0.00029973705293831525,
"loss": 2.0083614349365235,
"step": 1670
},
{
"epoch": 0.21504,
"grad_norm": 0.7454256415367126,
"learning_rate": 0.00029973346187115466,
"loss": 2.000332069396973,
"step": 1680
},
{
"epoch": 0.21632,
"grad_norm": 0.6826542019844055,
"learning_rate": 0.0002997298464702323,
"loss": 2.0105188369750975,
"step": 1690
},
{
"epoch": 0.2176,
"grad_norm": 0.8427369594573975,
"learning_rate": 0.00029972620673613564,
"loss": 1.9911138534545898,
"step": 1700
},
{
"epoch": 0.21888,
"grad_norm": 0.8828198313713074,
"learning_rate": 0.00029972254266945623,
"loss": 2.017778778076172,
"step": 1710
},
{
"epoch": 0.22016,
"grad_norm": 0.8785969614982605,
"learning_rate": 0.0002997188542707895,
"loss": 2.0029521942138673,
"step": 1720
},
{
"epoch": 0.22144,
"grad_norm": 0.7189239859580994,
"learning_rate": 0.00029971514154073496,
"loss": 1.9962823867797852,
"step": 1730
},
{
"epoch": 0.22272,
"grad_norm": 0.65921550989151,
"learning_rate": 0.00029971140447989584,
"loss": 2.0048191070556642,
"step": 1740
},
{
"epoch": 0.224,
"grad_norm": 0.8463709950447083,
"learning_rate": 0.0002997076430888796,
"loss": 2.013206672668457,
"step": 1750
},
{
"epoch": 0.22528,
"grad_norm": 0.6320969462394714,
"learning_rate": 0.00029970385736829747,
"loss": 2.0032022476196287,
"step": 1760
},
{
"epoch": 0.22656,
"grad_norm": 0.777824342250824,
"learning_rate": 0.00029970004731876465,
"loss": 1.9946500778198242,
"step": 1770
},
{
"epoch": 0.22784,
"grad_norm": 0.7572101354598999,
"learning_rate": 0.0002996962129409003,
"loss": 2.011198043823242,
"step": 1780
},
{
"epoch": 0.22912,
"grad_norm": 0.6622071266174316,
"learning_rate": 0.00029969235423532765,
"loss": 1.9974157333374023,
"step": 1790
},
{
"epoch": 0.2304,
"grad_norm": 0.6875664591789246,
"learning_rate": 0.0002996884712026737,
"loss": 1.9945417404174806,
"step": 1800
},
{
"epoch": 0.23168,
"grad_norm": 0.7916296124458313,
"learning_rate": 0.0002996845638435696,
"loss": 2.035945510864258,
"step": 1810
},
{
"epoch": 0.23296,
"grad_norm": 0.7541037797927856,
"learning_rate": 0.00029968063215865033,
"loss": 1.9989112854003905,
"step": 1820
},
{
"epoch": 0.23424,
"grad_norm": 0.599982738494873,
"learning_rate": 0.0002996766761485547,
"loss": 1.9939571380615235,
"step": 1830
},
{
"epoch": 0.23552,
"grad_norm": 0.739617645740509,
"learning_rate": 0.00029967269581392575,
"loss": 1.9879533767700195,
"step": 1840
},
{
"epoch": 0.2368,
"grad_norm": 0.7203322649002075,
"learning_rate": 0.00029966869115541036,
"loss": 1.9934770584106445,
"step": 1850
},
{
"epoch": 0.23808,
"grad_norm": 0.9270490407943726,
"learning_rate": 0.00029966466217365927,
"loss": 1.9913738250732422,
"step": 1860
},
{
"epoch": 0.23936,
"grad_norm": 0.8059845566749573,
"learning_rate": 0.0002996606088693273,
"loss": 2.032894325256348,
"step": 1870
},
{
"epoch": 0.24064,
"grad_norm": 0.7291269898414612,
"learning_rate": 0.0002996565312430731,
"loss": 2.0083896636962892,
"step": 1880
},
{
"epoch": 0.24192,
"grad_norm": 0.7227823734283447,
"learning_rate": 0.0002996524292955594,
"loss": 1.998744010925293,
"step": 1890
},
{
"epoch": 0.2432,
"grad_norm": 0.7445799708366394,
"learning_rate": 0.0002996483030274528,
"loss": 1.9974910736083984,
"step": 1900
},
{
"epoch": 0.24448,
"grad_norm": 0.7004072070121765,
"learning_rate": 0.0002996441524394239,
"loss": 2.00911865234375,
"step": 1910
},
{
"epoch": 0.24576,
"grad_norm": 0.9402841329574585,
"learning_rate": 0.00029963997753214717,
"loss": 2.0093734741210936,
"step": 1920
},
{
"epoch": 0.24704,
"grad_norm": 0.6809439659118652,
"learning_rate": 0.0002996357783063011,
"loss": 2.0103471755981444,
"step": 1930
},
{
"epoch": 0.24832,
"grad_norm": 0.7677909731864929,
"learning_rate": 0.00029963155476256826,
"loss": 1.9892677307128905,
"step": 1940
},
{
"epoch": 0.2496,
"grad_norm": 0.8318672180175781,
"learning_rate": 0.0002996273069016348,
"loss": 1.9922115325927734,
"step": 1950
},
{
"epoch": 0.25088,
"grad_norm": 0.6714010834693909,
"learning_rate": 0.00029962303472419124,
"loss": 1.993598747253418,
"step": 1960
},
{
"epoch": 0.25216,
"grad_norm": 0.7420302629470825,
"learning_rate": 0.00029961873823093176,
"loss": 2.010415267944336,
"step": 1970
},
{
"epoch": 0.25344,
"grad_norm": 0.8029727339744568,
"learning_rate": 0.00029961441742255477,
"loss": 2.0113759994506837,
"step": 1980
},
{
"epoch": 0.25472,
"grad_norm": 0.7108703255653381,
"learning_rate": 0.0002996100722997622,
"loss": 2.0091278076171877,
"step": 1990
},
{
"epoch": 0.256,
"grad_norm": 0.6735569834709167,
"learning_rate": 0.0002996057028632604,
"loss": 2.0023717880249023,
"step": 2000
},
{
"epoch": 0.25728,
"grad_norm": 0.8867200016975403,
"learning_rate": 0.0002996013091137594,
"loss": 1.997395896911621,
"step": 2010
},
{
"epoch": 0.25856,
"grad_norm": 0.7353429794311523,
"learning_rate": 0.0002995968910519732,
"loss": 2.015260124206543,
"step": 2020
},
{
"epoch": 0.25984,
"grad_norm": 0.8277012705802917,
"learning_rate": 0.00029959244867861987,
"loss": 1.979180145263672,
"step": 2030
},
{
"epoch": 0.26112,
"grad_norm": 0.6707190871238708,
"learning_rate": 0.00029958798199442136,
"loss": 2.0032121658325197,
"step": 2040
},
{
"epoch": 0.2624,
"grad_norm": 0.734474241733551,
"learning_rate": 0.0002995834910001035,
"loss": 2.0063905715942383,
"step": 2050
},
{
"epoch": 0.26368,
"grad_norm": 0.6844320297241211,
"learning_rate": 0.0002995789756963961,
"loss": 1.9935205459594727,
"step": 2060
},
{
"epoch": 0.26496,
"grad_norm": 0.7805824875831604,
"learning_rate": 0.00029957443608403313,
"loss": 1.9970199584960937,
"step": 2070
},
{
"epoch": 0.26624,
"grad_norm": 0.7733249664306641,
"learning_rate": 0.0002995698721637522,
"loss": 2.0413721084594725,
"step": 2080
},
{
"epoch": 0.26752,
"grad_norm": 1.118902564048767,
"learning_rate": 0.00029956528393629506,
"loss": 2.0041427612304688,
"step": 2090
},
{
"epoch": 0.2688,
"grad_norm": 0.7652356624603271,
"learning_rate": 0.00029956067140240734,
"loss": 2.009052276611328,
"step": 2100
},
{
"epoch": 0.27008,
"grad_norm": 0.7870017290115356,
"learning_rate": 0.0002995560345628387,
"loss": 2.0207101821899416,
"step": 2110
},
{
"epoch": 0.27136,
"grad_norm": 0.5984968543052673,
"learning_rate": 0.0002995513734183426,
"loss": 2.008289337158203,
"step": 2120
},
{
"epoch": 0.27264,
"grad_norm": 0.7700141072273254,
"learning_rate": 0.00029954668796967654,
"loss": 2.0115030288696287,
"step": 2130
},
{
"epoch": 0.27392,
"grad_norm": 1.2647989988327026,
"learning_rate": 0.0002995419782176021,
"loss": 2.023280143737793,
"step": 2140
},
{
"epoch": 0.2752,
"grad_norm": 0.8249433040618896,
"learning_rate": 0.0002995372441628845,
"loss": 1.9807783126831056,
"step": 2150
},
{
"epoch": 0.27648,
"grad_norm": 0.7661439776420593,
"learning_rate": 0.00029953248580629323,
"loss": 2.0067214965820312,
"step": 2160
},
{
"epoch": 0.27776,
"grad_norm": 0.6710767149925232,
"learning_rate": 0.0002995277031486015,
"loss": 2.005890464782715,
"step": 2170
},
{
"epoch": 0.27904,
"grad_norm": 0.8246187567710876,
"learning_rate": 0.00029952289619058673,
"loss": 2.016609764099121,
"step": 2180
},
{
"epoch": 0.28032,
"grad_norm": 0.6944814324378967,
"learning_rate": 0.0002995180649330299,
"loss": 2.0185358047485353,
"step": 2190
},
{
"epoch": 0.2816,
"grad_norm": 1.15223228931427,
"learning_rate": 0.0002995132093767162,
"loss": 1.9930595397949218,
"step": 2200
},
{
"epoch": 0.28288,
"grad_norm": 1.1536227464675903,
"learning_rate": 0.00029950832952243487,
"loss": 2.0209888458251952,
"step": 2210
},
{
"epoch": 0.28416,
"grad_norm": 0.8789283037185669,
"learning_rate": 0.00029950342537097874,
"loss": 2.014572334289551,
"step": 2220
},
{
"epoch": 0.28544,
"grad_norm": 0.8173903226852417,
"learning_rate": 0.00029949849692314503,
"loss": 2.0143598556518554,
"step": 2230
},
{
"epoch": 0.28672,
"grad_norm": 0.7095955014228821,
"learning_rate": 0.0002994935441797345,
"loss": 2.004468536376953,
"step": 2240
},
{
"epoch": 0.288,
"grad_norm": 0.8468304872512817,
"learning_rate": 0.0002994885671415521,
"loss": 2.037894630432129,
"step": 2250
},
{
"epoch": 0.28928,
"grad_norm": 0.8071026802062988,
"learning_rate": 0.0002994835658094067,
"loss": 2.023789405822754,
"step": 2260
},
{
"epoch": 0.29056,
"grad_norm": 0.7277827262878418,
"learning_rate": 0.0002994785401841111,
"loss": 2.010818672180176,
"step": 2270
},
{
"epoch": 0.29184,
"grad_norm": 0.7613065838813782,
"learning_rate": 0.00029947349026648184,
"loss": 1.9951431274414062,
"step": 2280
},
{
"epoch": 0.29312,
"grad_norm": 0.7575923800468445,
"learning_rate": 0.0002994684160573399,
"loss": 2.0147769927978514,
"step": 2290
},
{
"epoch": 0.2944,
"grad_norm": 1.050876259803772,
"learning_rate": 0.0002994633175575097,
"loss": 2.0217056274414062,
"step": 2300
},
{
"epoch": 0.29568,
"grad_norm": 0.6943807601928711,
"learning_rate": 0.0002994581947678199,
"loss": 2.0191987991333007,
"step": 2310
},
{
"epoch": 0.29696,
"grad_norm": 0.8438522219657898,
"learning_rate": 0.00029945304768910297,
"loss": 1.9989118576049805,
"step": 2320
},
{
"epoch": 0.29824,
"grad_norm": 0.7962123155593872,
"learning_rate": 0.00029944787632219547,
"loss": 2.0173967361450194,
"step": 2330
},
{
"epoch": 0.29952,
"grad_norm": 0.7668672204017639,
"learning_rate": 0.0002994426806679378,
"loss": 2.0132951736450195,
"step": 2340
},
{
"epoch": 0.3008,
"grad_norm": 0.6691519618034363,
"learning_rate": 0.0002994374607271742,
"loss": 2.0168371200561523,
"step": 2350
},
{
"epoch": 0.30208,
"grad_norm": 0.6511347889900208,
"learning_rate": 0.0002994322165007531,
"loss": 2.0063940048217774,
"step": 2360
},
{
"epoch": 0.30336,
"grad_norm": 0.8013675212860107,
"learning_rate": 0.00029942694798952677,
"loss": 2.0087839126586915,
"step": 2370
},
{
"epoch": 0.30464,
"grad_norm": 0.899410605430603,
"learning_rate": 0.0002994216551943513,
"loss": 2.0143436431884765,
"step": 2380
},
{
"epoch": 0.30592,
"grad_norm": 0.6686886548995972,
"learning_rate": 0.000299416338116087,
"loss": 1.9976930618286133,
"step": 2390
},
{
"epoch": 0.3072,
"grad_norm": 0.8113393187522888,
"learning_rate": 0.0002994109967555978,
"loss": 2.0220170974731446,
"step": 2400
},
{
"epoch": 0.30848,
"grad_norm": 0.7538885474205017,
"learning_rate": 0.00029940563111375195,
"loss": 2.004685974121094,
"step": 2410
},
{
"epoch": 0.30976,
"grad_norm": 0.8250473141670227,
"learning_rate": 0.0002994002411914213,
"loss": 2.0201698303222657,
"step": 2420
},
{
"epoch": 0.31104,
"grad_norm": 0.7206689119338989,
"learning_rate": 0.00029939482698948175,
"loss": 1.9911544799804688,
"step": 2430
},
{
"epoch": 0.31232,
"grad_norm": 0.7519621253013611,
"learning_rate": 0.0002993893885088133,
"loss": 2.0177082061767577,
"step": 2440
},
{
"epoch": 0.3136,
"grad_norm": 0.777955174446106,
"learning_rate": 0.0002993839257502997,
"loss": 1.9929460525512694,
"step": 2450
},
{
"epoch": 0.31488,
"grad_norm": 0.7577837705612183,
"learning_rate": 0.0002993784387148288,
"loss": 2.0109020233154298,
"step": 2460
},
{
"epoch": 0.31616,
"grad_norm": 0.7473259568214417,
"learning_rate": 0.0002993729274032922,
"loss": 2.0051565170288086,
"step": 2470
},
{
"epoch": 0.31744,
"grad_norm": 0.7150516510009766,
"learning_rate": 0.0002993673918165856,
"loss": 1.988702392578125,
"step": 2480
},
{
"epoch": 0.31872,
"grad_norm": 0.7589870095252991,
"learning_rate": 0.00029936183195560873,
"loss": 2.0132009506225588,
"step": 2490
},
{
"epoch": 0.32,
"grad_norm": 0.8448571562767029,
"learning_rate": 0.00029935624782126505,
"loss": 2.00791072845459,
"step": 2500
},
{
"epoch": 0.32128,
"grad_norm": 0.9435201287269592,
"learning_rate": 0.000299350639414462,
"loss": 2.0194942474365236,
"step": 2510
},
{
"epoch": 0.32256,
"grad_norm": 0.65596604347229,
"learning_rate": 0.00029934500673611114,
"loss": 2.0138187408447266,
"step": 2520
},
{
"epoch": 0.32384,
"grad_norm": 0.7170972228050232,
"learning_rate": 0.00029933934978712784,
"loss": 2.015252876281738,
"step": 2530
},
{
"epoch": 0.32512,
"grad_norm": 0.7384889125823975,
"learning_rate": 0.00029933366856843135,
"loss": 2.0059532165527343,
"step": 2540
},
{
"epoch": 0.3264,
"grad_norm": 0.7757416367530823,
"learning_rate": 0.000299327963080945,
"loss": 2.024062728881836,
"step": 2550
},
{
"epoch": 0.32768,
"grad_norm": 0.8023697733879089,
"learning_rate": 0.000299322233325596,
"loss": 1.9900062561035157,
"step": 2560
},
{
"epoch": 0.32896,
"grad_norm": 0.7611852288246155,
"learning_rate": 0.00029931647930331556,
"loss": 2.0226030349731445,
"step": 2570
},
{
"epoch": 0.33024,
"grad_norm": 0.8219276070594788,
"learning_rate": 0.0002993107010150387,
"loss": 1.9970281600952149,
"step": 2580
},
{
"epoch": 0.33152,
"grad_norm": 0.7846634984016418,
"learning_rate": 0.0002993048984617046,
"loss": 1.9928812026977538,
"step": 2590
},
{
"epoch": 0.3328,
"grad_norm": 0.7521805167198181,
"learning_rate": 0.0002992990716442561,
"loss": 1.99877986907959,
"step": 2600
},
{
"epoch": 0.33408,
"grad_norm": 0.660119354724884,
"learning_rate": 0.0002992932205636403,
"loss": 2.007124900817871,
"step": 2610
},
{
"epoch": 0.33536,
"grad_norm": 0.7946996688842773,
"learning_rate": 0.0002992873452208079,
"loss": 1.9985849380493164,
"step": 2620
},
{
"epoch": 0.33664,
"grad_norm": 0.8377641439437866,
"learning_rate": 0.00029928144561671387,
"loss": 2.0259756088256835,
"step": 2630
},
{
"epoch": 0.33792,
"grad_norm": 0.6834446787834167,
"learning_rate": 0.0002992755217523169,
"loss": 1.9996057510375977,
"step": 2640
},
{
"epoch": 0.3392,
"grad_norm": 0.8117668628692627,
"learning_rate": 0.00029926957362857976,
"loss": 1.9947677612304688,
"step": 2650
},
{
"epoch": 0.34048,
"grad_norm": 0.6758447289466858,
"learning_rate": 0.0002992636012464691,
"loss": 1.984456443786621,
"step": 2660
},
{
"epoch": 0.34176,
"grad_norm": 0.6741154789924622,
"learning_rate": 0.00029925760460695547,
"loss": 1.9969703674316406,
"step": 2670
},
{
"epoch": 0.34304,
"grad_norm": 0.7465670108795166,
"learning_rate": 0.0002992515837110134,
"loss": 2.0042219161987305,
"step": 2680
},
{
"epoch": 0.34432,
"grad_norm": 0.8404413461685181,
"learning_rate": 0.0002992455385596214,
"loss": 1.9949943542480468,
"step": 2690
},
{
"epoch": 0.3456,
"grad_norm": 0.7381366491317749,
"learning_rate": 0.00029923946915376187,
"loss": 2.0078285217285154,
"step": 2700
},
{
"epoch": 0.34688,
"grad_norm": 0.7588344216346741,
"learning_rate": 0.00029923337549442124,
"loss": 2.008309173583984,
"step": 2710
},
{
"epoch": 0.34816,
"grad_norm": 0.7521999478340149,
"learning_rate": 0.0002992272575825897,
"loss": 2.0139354705810546,
"step": 2720
},
{
"epoch": 0.34944,
"grad_norm": 0.6759751439094543,
"learning_rate": 0.0002992211154192616,
"loss": 2.001015853881836,
"step": 2730
},
{
"epoch": 0.35072,
"grad_norm": 0.6119874119758606,
"learning_rate": 0.00029921494900543507,
"loss": 2.0114923477172852,
"step": 2740
},
{
"epoch": 0.352,
"grad_norm": 0.7517712116241455,
"learning_rate": 0.0002992087583421123,
"loss": 1.9976957321166993,
"step": 2750
},
{
"epoch": 0.35328,
"grad_norm": 0.769814133644104,
"learning_rate": 0.0002992025434302992,
"loss": 1.9972566604614257,
"step": 2760
},
{
"epoch": 0.35456,
"grad_norm": 0.6553206443786621,
"learning_rate": 0.00029919630427100596,
"loss": 2.007891845703125,
"step": 2770
},
{
"epoch": 0.35584,
"grad_norm": 0.6894077658653259,
"learning_rate": 0.0002991900408652465,
"loss": 2.0165834426879883,
"step": 2780
},
{
"epoch": 0.35712,
"grad_norm": 0.669863224029541,
"learning_rate": 0.00029918375321403857,
"loss": 2.010799789428711,
"step": 2790
},
{
"epoch": 0.3584,
"grad_norm": 0.7974678874015808,
"learning_rate": 0.00029917744131840416,
"loss": 2.0101280212402344,
"step": 2800
},
{
"epoch": 0.35968,
"grad_norm": 0.7307773232460022,
"learning_rate": 0.00029917110517936894,
"loss": 2.024929428100586,
"step": 2810
},
{
"epoch": 0.36096,
"grad_norm": 0.7141740322113037,
"learning_rate": 0.00029916474479796265,
"loss": 2.0083641052246093,
"step": 2820
},
{
"epoch": 0.36224,
"grad_norm": 0.6874279379844666,
"learning_rate": 0.000299158360175219,
"loss": 2.0239688873291017,
"step": 2830
},
{
"epoch": 0.36352,
"grad_norm": 0.6707704663276672,
"learning_rate": 0.0002991519513121755,
"loss": 2.005034828186035,
"step": 2840
},
{
"epoch": 0.3648,
"grad_norm": 0.8018107414245605,
"learning_rate": 0.00029914551820987373,
"loss": 1.997845458984375,
"step": 2850
},
{
"epoch": 0.36608,
"grad_norm": 0.7515741586685181,
"learning_rate": 0.00029913906086935916,
"loss": 2.0154483795166014,
"step": 2860
},
{
"epoch": 0.36736,
"grad_norm": 0.7631517648696899,
"learning_rate": 0.0002991325792916811,
"loss": 1.977918815612793,
"step": 2870
},
{
"epoch": 0.36864,
"grad_norm": 0.8842434287071228,
"learning_rate": 0.000299126073477893,
"loss": 1.975722885131836,
"step": 2880
},
{
"epoch": 0.36992,
"grad_norm": 0.7274171113967896,
"learning_rate": 0.00029911954342905217,
"loss": 1.9984859466552733,
"step": 2890
},
{
"epoch": 0.3712,
"grad_norm": 0.7220802307128906,
"learning_rate": 0.00029911298914621977,
"loss": 2.0087894439697265,
"step": 2900
},
{
"epoch": 0.37248,
"grad_norm": 0.6422837376594543,
"learning_rate": 0.0002991064106304609,
"loss": 2.002351760864258,
"step": 2910
},
{
"epoch": 0.37376,
"grad_norm": 0.753294825553894,
"learning_rate": 0.00029909980788284476,
"loss": 1.986056137084961,
"step": 2920
},
{
"epoch": 0.37504,
"grad_norm": 0.7144089937210083,
"learning_rate": 0.00029909318090444437,
"loss": 1.9991958618164063,
"step": 2930
},
{
"epoch": 0.37632,
"grad_norm": 0.704556405544281,
"learning_rate": 0.0002990865296963367,
"loss": 1.9974645614624023,
"step": 2940
},
{
"epoch": 0.3776,
"grad_norm": 0.7290600538253784,
"learning_rate": 0.0002990798542596026,
"loss": 2.015464019775391,
"step": 2950
},
{
"epoch": 0.37888,
"grad_norm": 0.7425021529197693,
"learning_rate": 0.00029907315459532706,
"loss": 2.0130130767822267,
"step": 2960
},
{
"epoch": 0.38016,
"grad_norm": 0.7028226852416992,
"learning_rate": 0.0002990664307045988,
"loss": 1.999576950073242,
"step": 2970
},
{
"epoch": 0.38144,
"grad_norm": 0.6917806267738342,
"learning_rate": 0.00029905968258851047,
"loss": 2.00009765625,
"step": 2980
},
{
"epoch": 0.38272,
"grad_norm": 0.7476458549499512,
"learning_rate": 0.00029905291024815876,
"loss": 1.9876760482788085,
"step": 2990
},
{
"epoch": 0.384,
"grad_norm": 0.8785274028778076,
"learning_rate": 0.00029904611368464435,
"loss": 2.018217086791992,
"step": 3000
},
{
"epoch": 0.38528,
"grad_norm": 0.7297069430351257,
"learning_rate": 0.0002990392928990717,
"loss": 1.9931718826293945,
"step": 3010
},
{
"epoch": 0.38656,
"grad_norm": 0.6672376990318298,
"learning_rate": 0.0002990324478925494,
"loss": 2.0107067108154295,
"step": 3020
},
{
"epoch": 0.38784,
"grad_norm": 0.7302089929580688,
"learning_rate": 0.0002990255786661897,
"loss": 1.994314956665039,
"step": 3030
},
{
"epoch": 0.38912,
"grad_norm": 0.7202204465866089,
"learning_rate": 0.00029901868522110904,
"loss": 2.0174509048461915,
"step": 3040
},
{
"epoch": 0.3904,
"grad_norm": 0.8425076603889465,
"learning_rate": 0.00029901176755842764,
"loss": 2.0013553619384767,
"step": 3050
},
{
"epoch": 0.39168,
"grad_norm": 0.8686426877975464,
"learning_rate": 0.00029900482567926976,
"loss": 2.007709503173828,
"step": 3060
},
{
"epoch": 0.39296,
"grad_norm": 0.6840008497238159,
"learning_rate": 0.00029899785958476357,
"loss": 2.004677963256836,
"step": 3070
},
{
"epoch": 0.39424,
"grad_norm": 0.6260886192321777,
"learning_rate": 0.0002989908692760411,
"loss": 1.9879840850830077,
"step": 3080
},
{
"epoch": 0.39552,
"grad_norm": 0.6933361887931824,
"learning_rate": 0.0002989838547542384,
"loss": 2.011644744873047,
"step": 3090
},
{
"epoch": 0.3968,
"grad_norm": 0.7048612833023071,
"learning_rate": 0.0002989768160204955,
"loss": 2.019415092468262,
"step": 3100
},
{
"epoch": 0.39808,
"grad_norm": 0.7651655673980713,
"learning_rate": 0.00029896975307595614,
"loss": 2.0010753631591798,
"step": 3110
},
{
"epoch": 0.39936,
"grad_norm": 0.6941404938697815,
"learning_rate": 0.00029896266592176825,
"loss": 1.984878921508789,
"step": 3120
},
{
"epoch": 0.40064,
"grad_norm": 0.8553006649017334,
"learning_rate": 0.00029895555455908355,
"loss": 1.9996231079101563,
"step": 3130
},
{
"epoch": 0.40192,
"grad_norm": 0.7001233100891113,
"learning_rate": 0.00029894841898905775,
"loss": 1.9921398162841797,
"step": 3140
},
{
"epoch": 0.4032,
"grad_norm": 0.8051826357841492,
"learning_rate": 0.0002989412592128505,
"loss": 1.999976921081543,
"step": 3150
},
{
"epoch": 0.40448,
"grad_norm": 0.7095255255699158,
"learning_rate": 0.00029893407523162534,
"loss": 2.0126005172729493,
"step": 3160
},
{
"epoch": 0.40576,
"grad_norm": 0.9202665686607361,
"learning_rate": 0.00029892686704654977,
"loss": 1.9911052703857421,
"step": 3170
},
{
"epoch": 0.40704,
"grad_norm": 0.7405928373336792,
"learning_rate": 0.0002989196346587952,
"loss": 1.9965196609497071,
"step": 3180
},
{
"epoch": 0.40832,
"grad_norm": 1.4823166131973267,
"learning_rate": 0.0002989123780695371,
"loss": 2.0123992919921876,
"step": 3190
},
{
"epoch": 0.4096,
"grad_norm": 0.7466282248497009,
"learning_rate": 0.00029890509727995457,
"loss": 2.0086360931396485,
"step": 3200
},
{
"epoch": 0.41088,
"grad_norm": 0.7171423435211182,
"learning_rate": 0.00029889779229123095,
"loss": 2.0146774291992187,
"step": 3210
},
{
"epoch": 0.41216,
"grad_norm": 0.8924009203910828,
"learning_rate": 0.00029889046310455346,
"loss": 2.006013107299805,
"step": 3220
},
{
"epoch": 0.41344,
"grad_norm": 0.6809982657432556,
"learning_rate": 0.0002988831097211131,
"loss": 2.007125473022461,
"step": 3230
},
{
"epoch": 0.41472,
"grad_norm": 0.6054995656013489,
"learning_rate": 0.000298875732142105,
"loss": 1.9968708038330079,
"step": 3240
},
{
"epoch": 0.416,
"grad_norm": 0.6415665149688721,
"learning_rate": 0.000298868330368728,
"loss": 2.004538345336914,
"step": 3250
},
{
"epoch": 0.41728,
"grad_norm": 0.738610029220581,
"learning_rate": 0.00029886090440218505,
"loss": 1.9990394592285157,
"step": 3260
},
{
"epoch": 0.41856,
"grad_norm": 0.7058972120285034,
"learning_rate": 0.00029885345424368294,
"loss": 2.0169488906860353,
"step": 3270
},
{
"epoch": 0.41984,
"grad_norm": 0.6690015196800232,
"learning_rate": 0.0002988459798944324,
"loss": 1.990290069580078,
"step": 3280
},
{
"epoch": 0.42112,
"grad_norm": 0.627063512802124,
"learning_rate": 0.00029883848135564827,
"loss": 1.992024040222168,
"step": 3290
},
{
"epoch": 0.4224,
"grad_norm": 0.6873062252998352,
"learning_rate": 0.000298830958628549,
"loss": 2.018503189086914,
"step": 3300
},
{
"epoch": 0.42368,
"grad_norm": 0.7181523442268372,
"learning_rate": 0.0002988234117143572,
"loss": 2.0042495727539062,
"step": 3310
},
{
"epoch": 0.42496,
"grad_norm": 0.7824463248252869,
"learning_rate": 0.0002988158406142994,
"loss": 1.9997383117675782,
"step": 3320
},
{
"epoch": 0.42624,
"grad_norm": 0.7802339196205139,
"learning_rate": 0.0002988082453296059,
"loss": 1.996378517150879,
"step": 3330
},
{
"epoch": 0.42752,
"grad_norm": 0.7132096886634827,
"learning_rate": 0.00029880062586151105,
"loss": 2.0132312774658203,
"step": 3340
},
{
"epoch": 0.4288,
"grad_norm": 0.8338831067085266,
"learning_rate": 0.00029879298221125323,
"loss": 2.010465621948242,
"step": 3350
},
{
"epoch": 0.43008,
"grad_norm": 0.7584986090660095,
"learning_rate": 0.0002987853143800745,
"loss": 1.9958574295043945,
"step": 3360
},
{
"epoch": 0.43136,
"grad_norm": 0.6547184586524963,
"learning_rate": 0.00029877762236922117,
"loss": 1.9954259872436524,
"step": 3370
},
{
"epoch": 0.43264,
"grad_norm": 0.7016308903694153,
"learning_rate": 0.0002987699061799431,
"loss": 1.9987804412841796,
"step": 3380
},
{
"epoch": 0.43392,
"grad_norm": 0.7248575687408447,
"learning_rate": 0.0002987621658134944,
"loss": 2.0054561614990236,
"step": 3390
},
{
"epoch": 0.4352,
"grad_norm": 1.0197134017944336,
"learning_rate": 0.00029875440127113295,
"loss": 2.0060552597045898,
"step": 3400
},
{
"epoch": 0.43648,
"grad_norm": 0.7604638338088989,
"learning_rate": 0.00029874661255412056,
"loss": 1.9959848403930665,
"step": 3410
},
{
"epoch": 0.43776,
"grad_norm": 0.7341379523277283,
"learning_rate": 0.0002987387996637231,
"loss": 2.0023780822753907,
"step": 3420
},
{
"epoch": 0.43904,
"grad_norm": 0.6850576400756836,
"learning_rate": 0.0002987309626012102,
"loss": 2.020222282409668,
"step": 3430
},
{
"epoch": 0.44032,
"grad_norm": 0.7283312082290649,
"learning_rate": 0.00029872310136785544,
"loss": 1.994460105895996,
"step": 3440
},
{
"epoch": 0.4416,
"grad_norm": 0.7015148997306824,
"learning_rate": 0.00029871521596493654,
"loss": 2.015683174133301,
"step": 3450
},
{
"epoch": 0.44288,
"grad_norm": 0.6451849341392517,
"learning_rate": 0.0002987073063937348,
"loss": 1.9941566467285157,
"step": 3460
},
{
"epoch": 0.44416,
"grad_norm": 0.6728188991546631,
"learning_rate": 0.0002986993726555358,
"loss": 2.0009782791137694,
"step": 3470
},
{
"epoch": 0.44544,
"grad_norm": 0.7595881223678589,
"learning_rate": 0.0002986914147516288,
"loss": 1.994197463989258,
"step": 3480
},
{
"epoch": 0.44672,
"grad_norm": 0.7372469902038574,
"learning_rate": 0.00029868343268330704,
"loss": 1.9918861389160156,
"step": 3490
},
{
"epoch": 0.448,
"grad_norm": 0.8467145562171936,
"learning_rate": 0.0002986754264518678,
"loss": 1.970982551574707,
"step": 3500
},
{
"epoch": 0.44928,
"grad_norm": 0.6036944389343262,
"learning_rate": 0.00029866739605861214,
"loss": 1.970067024230957,
"step": 3510
},
{
"epoch": 0.45056,
"grad_norm": 0.7512437701225281,
"learning_rate": 0.00029865934150484513,
"loss": 1.9916757583618163,
"step": 3520
},
{
"epoch": 0.45184,
"grad_norm": 0.7820079326629639,
"learning_rate": 0.00029865126279187575,
"loss": 2.015567970275879,
"step": 3530
},
{
"epoch": 0.45312,
"grad_norm": 0.753780722618103,
"learning_rate": 0.0002986431599210169,
"loss": 1.9925758361816406,
"step": 3540
},
{
"epoch": 0.4544,
"grad_norm": 0.583357036113739,
"learning_rate": 0.00029863503289358543,
"loss": 2.0011922836303713,
"step": 3550
},
{
"epoch": 0.45568,
"grad_norm": 0.7955908179283142,
"learning_rate": 0.00029862688171090207,
"loss": 1.9962081909179688,
"step": 3560
},
{
"epoch": 0.45696,
"grad_norm": 0.6503195762634277,
"learning_rate": 0.0002986187063742914,
"loss": 2.0152355194091798,
"step": 3570
},
{
"epoch": 0.45824,
"grad_norm": 0.702178955078125,
"learning_rate": 0.00029861050688508224,
"loss": 2.010386848449707,
"step": 3580
},
{
"epoch": 0.45952,
"grad_norm": 0.6806907057762146,
"learning_rate": 0.000298602283244607,
"loss": 2.002973747253418,
"step": 3590
},
{
"epoch": 0.4608,
"grad_norm": 0.6157857775688171,
"learning_rate": 0.00029859403545420214,
"loss": 1.995046615600586,
"step": 3600
},
{
"epoch": 0.46208,
"grad_norm": 0.6844906806945801,
"learning_rate": 0.000298585763515208,
"loss": 2.016707992553711,
"step": 3610
},
{
"epoch": 0.46336,
"grad_norm": 0.7760967016220093,
"learning_rate": 0.00029857746742896897,
"loss": 1.9860107421875,
"step": 3620
},
{
"epoch": 0.46464,
"grad_norm": 0.8712556958198547,
"learning_rate": 0.0002985691471968332,
"loss": 2.001064491271973,
"step": 3630
},
{
"epoch": 0.46592,
"grad_norm": 0.7285525798797607,
"learning_rate": 0.00029856080282015293,
"loss": 2.0140453338623048,
"step": 3640
},
{
"epoch": 0.4672,
"grad_norm": 0.7722259759902954,
"learning_rate": 0.00029855243430028415,
"loss": 2.0120145797729494,
"step": 3650
},
{
"epoch": 0.46848,
"grad_norm": 0.7177460193634033,
"learning_rate": 0.0002985440416385869,
"loss": 1.9976810455322265,
"step": 3660
},
{
"epoch": 0.46976,
"grad_norm": 0.8264991044998169,
"learning_rate": 0.0002985356248364251,
"loss": 2.005207633972168,
"step": 3670
},
{
"epoch": 0.47104,
"grad_norm": 0.8237878084182739,
"learning_rate": 0.00029852718389516666,
"loss": 1.988731575012207,
"step": 3680
},
{
"epoch": 0.47232,
"grad_norm": 1.3742607831954956,
"learning_rate": 0.00029851871881618323,
"loss": 2.0098743438720703,
"step": 3690
},
{
"epoch": 0.4736,
"grad_norm": 0.7714928388595581,
"learning_rate": 0.0002985102296008505,
"loss": 1.9839420318603516,
"step": 3700
},
{
"epoch": 0.47488,
"grad_norm": 0.7396987080574036,
"learning_rate": 0.0002985017162505483,
"loss": 1.9785545349121094,
"step": 3710
},
{
"epoch": 0.47616,
"grad_norm": 0.7527230978012085,
"learning_rate": 0.0002984931787666599,
"loss": 2.0031604766845703,
"step": 3720
},
{
"epoch": 0.47744,
"grad_norm": 0.7075295448303223,
"learning_rate": 0.0002984846171505729,
"loss": 1.9907520294189454,
"step": 3730
},
{
"epoch": 0.47872,
"grad_norm": 0.6408486366271973,
"learning_rate": 0.0002984760314036787,
"loss": 2.024155044555664,
"step": 3740
},
{
"epoch": 0.48,
"grad_norm": 0.7029430866241455,
"learning_rate": 0.0002984674215273725,
"loss": 2.00035400390625,
"step": 3750
},
{
"epoch": 0.48128,
"grad_norm": 0.7707000970840454,
"learning_rate": 0.0002984587875230536,
"loss": 2.0184585571289064,
"step": 3760
},
{
"epoch": 0.48256,
"grad_norm": 0.6695795655250549,
"learning_rate": 0.00029845012939212507,
"loss": 1.9930442810058593,
"step": 3770
},
{
"epoch": 0.48384,
"grad_norm": 0.6928637623786926,
"learning_rate": 0.0002984414471359941,
"loss": 1.9891483306884765,
"step": 3780
},
{
"epoch": 0.48512,
"grad_norm": 0.7137778997421265,
"learning_rate": 0.00029843274075607167,
"loss": 2.0126466751098633,
"step": 3790
},
{
"epoch": 0.4864,
"grad_norm": 0.7941726446151733,
"learning_rate": 0.00029842401025377254,
"loss": 1.9977178573608398,
"step": 3800
},
{
"epoch": 0.48768,
"grad_norm": 0.7112129926681519,
"learning_rate": 0.0002984152556305156,
"loss": 2.0004758834838867,
"step": 3810
},
{
"epoch": 0.48896,
"grad_norm": 0.6955267190933228,
"learning_rate": 0.0002984064768877237,
"loss": 2.0171720504760744,
"step": 3820
},
{
"epoch": 0.49024,
"grad_norm": 0.7090415358543396,
"learning_rate": 0.00029839767402682334,
"loss": 2.005842399597168,
"step": 3830
},
{
"epoch": 0.49152,
"grad_norm": 0.8107616305351257,
"learning_rate": 0.00029838884704924527,
"loss": 2.0197629928588867,
"step": 3840
},
{
"epoch": 0.4928,
"grad_norm": 0.7303240299224854,
"learning_rate": 0.0002983799959564239,
"loss": 2.0161752700805664,
"step": 3850
},
{
"epoch": 0.49408,
"grad_norm": 0.7189793586730957,
"learning_rate": 0.0002983711207497977,
"loss": 2.0196367263793946,
"step": 3860
},
{
"epoch": 0.49536,
"grad_norm": 0.7447757124900818,
"learning_rate": 0.000298362221430809,
"loss": 1.9797452926635741,
"step": 3870
},
{
"epoch": 0.49664,
"grad_norm": 0.7726782560348511,
"learning_rate": 0.000298353298000904,
"loss": 1.994917869567871,
"step": 3880
},
{
"epoch": 0.49792,
"grad_norm": 0.7866300344467163,
"learning_rate": 0.000298344350461533,
"loss": 1.994272232055664,
"step": 3890
},
{
"epoch": 0.4992,
"grad_norm": 0.7009814381599426,
"learning_rate": 0.00029833537881414997,
"loss": 1.9947071075439453,
"step": 3900
},
{
"epoch": 0.50048,
"grad_norm": 0.786959171295166,
"learning_rate": 0.000298326383060213,
"loss": 1.9846187591552735,
"step": 3910
},
{
"epoch": 0.50176,
"grad_norm": 0.798842191696167,
"learning_rate": 0.000298317363201184,
"loss": 2.0009033203125,
"step": 3920
},
{
"epoch": 0.50304,
"grad_norm": 0.6344483494758606,
"learning_rate": 0.0002983083192385289,
"loss": 2.0058782577514647,
"step": 3930
},
{
"epoch": 0.50432,
"grad_norm": 0.7020363211631775,
"learning_rate": 0.0002982992511737174,
"loss": 2.0104549407958983,
"step": 3940
},
{
"epoch": 0.5056,
"grad_norm": 0.7292479872703552,
"learning_rate": 0.00029829015900822314,
"loss": 2.0061105728149413,
"step": 3950
},
{
"epoch": 0.50688,
"grad_norm": 0.7374621629714966,
"learning_rate": 0.00029828104274352385,
"loss": 1.991245651245117,
"step": 3960
},
{
"epoch": 0.50816,
"grad_norm": 0.7191388010978699,
"learning_rate": 0.0002982719023811009,
"loss": 2.005914878845215,
"step": 3970
},
{
"epoch": 0.50944,
"grad_norm": 0.7209792137145996,
"learning_rate": 0.0002982627379224399,
"loss": 1.9991714477539062,
"step": 3980
},
{
"epoch": 0.51072,
"grad_norm": 0.7501109838485718,
"learning_rate": 0.00029825354936903,
"loss": 2.0138620376586913,
"step": 3990
},
{
"epoch": 0.512,
"grad_norm": 0.7204382419586182,
"learning_rate": 0.0002982443367223646,
"loss": 2.002537155151367,
"step": 4000
},
{
"epoch": 0.51328,
"grad_norm": 0.6654132008552551,
"learning_rate": 0.0002982350999839409,
"loss": 2.0058666229248048,
"step": 4010
},
{
"epoch": 0.51456,
"grad_norm": 0.6496480107307434,
"learning_rate": 0.00029822583915525987,
"loss": 2.0121301651000976,
"step": 4020
},
{
"epoch": 0.51584,
"grad_norm": 0.7528657913208008,
"learning_rate": 0.00029821655423782666,
"loss": 1.997304916381836,
"step": 4030
},
{
"epoch": 0.51712,
"grad_norm": 0.7558349967002869,
"learning_rate": 0.0002982072452331501,
"loss": 2.0115285873413087,
"step": 4040
},
{
"epoch": 0.5184,
"grad_norm": 0.649427056312561,
"learning_rate": 0.0002981979121427431,
"loss": 1.9857589721679687,
"step": 4050
},
{
"epoch": 0.51968,
"grad_norm": 0.8884936571121216,
"learning_rate": 0.0002981885549681224,
"loss": 1.9965480804443358,
"step": 4060
},
{
"epoch": 0.52096,
"grad_norm": 0.6860265135765076,
"learning_rate": 0.00029817917371080863,
"loss": 1.9892051696777344,
"step": 4070
},
{
"epoch": 0.52224,
"grad_norm": 0.6648100018501282,
"learning_rate": 0.00029816976837232647,
"loss": 1.9910903930664063,
"step": 4080
},
{
"epoch": 0.52352,
"grad_norm": 0.669529139995575,
"learning_rate": 0.0002981603389542043,
"loss": 1.9876266479492188,
"step": 4090
},
{
"epoch": 0.5248,
"grad_norm": 0.7481969594955444,
"learning_rate": 0.0002981508854579746,
"loss": 2.0230886459350588,
"step": 4100
},
{
"epoch": 0.52608,
"grad_norm": 0.7897581458091736,
"learning_rate": 0.00029814140788517364,
"loss": 1.9916162490844727,
"step": 4110
},
{
"epoch": 0.52736,
"grad_norm": 0.6596736907958984,
"learning_rate": 0.0002981319062373417,
"loss": 1.9848915100097657,
"step": 4120
},
{
"epoch": 0.52864,
"grad_norm": 0.6131610870361328,
"learning_rate": 0.00029812238051602294,
"loss": 2.003184700012207,
"step": 4130
},
{
"epoch": 0.52992,
"grad_norm": 0.775517463684082,
"learning_rate": 0.00029811283072276543,
"loss": 2.0039646148681642,
"step": 4140
},
{
"epoch": 0.5312,
"grad_norm": 0.6710156798362732,
"learning_rate": 0.00029810325685912113,
"loss": 2.0029441833496096,
"step": 4150
},
{
"epoch": 0.53248,
"grad_norm": 0.6885665655136108,
"learning_rate": 0.00029809365892664587,
"loss": 1.9940208435058593,
"step": 4160
},
{
"epoch": 0.53376,
"grad_norm": 0.7027019262313843,
"learning_rate": 0.00029808403692689953,
"loss": 2.003041076660156,
"step": 4170
},
{
"epoch": 0.53504,
"grad_norm": 0.7704663276672363,
"learning_rate": 0.00029807439086144575,
"loss": 2.008600044250488,
"step": 4180
},
{
"epoch": 0.53632,
"grad_norm": 0.7134334444999695,
"learning_rate": 0.0002980647207318522,
"loss": 2.0230785369873048,
"step": 4190
},
{
"epoch": 0.5376,
"grad_norm": 0.6899425983428955,
"learning_rate": 0.00029805502653969035,
"loss": 2.014421272277832,
"step": 4200
},
{
"epoch": 0.53888,
"grad_norm": 0.7387205362319946,
"learning_rate": 0.0002980453082865357,
"loss": 1.9908321380615235,
"step": 4210
},
{
"epoch": 0.54016,
"grad_norm": 0.6980050802230835,
"learning_rate": 0.0002980355659739676,
"loss": 1.9989233016967773,
"step": 4220
},
{
"epoch": 0.54144,
"grad_norm": 0.6020520925521851,
"learning_rate": 0.00029802579960356933,
"loss": 2.00181884765625,
"step": 4230
},
{
"epoch": 0.54272,
"grad_norm": 0.6786367893218994,
"learning_rate": 0.0002980160091769279,
"loss": 1.9844453811645508,
"step": 4240
},
{
"epoch": 0.544,
"grad_norm": 0.6527010202407837,
"learning_rate": 0.00029800619469563463,
"loss": 1.9895397186279298,
"step": 4250
},
{
"epoch": 0.54528,
"grad_norm": 0.6559035181999207,
"learning_rate": 0.00029799635616128435,
"loss": 2.0083845138549803,
"step": 4260
},
{
"epoch": 0.54656,
"grad_norm": 0.7348132729530334,
"learning_rate": 0.000297986493575476,
"loss": 2.0096080780029295,
"step": 4270
},
{
"epoch": 0.54784,
"grad_norm": 0.6425591111183167,
"learning_rate": 0.00029797660693981235,
"loss": 1.9981698989868164,
"step": 4280
},
{
"epoch": 0.54912,
"grad_norm": 0.6125735640525818,
"learning_rate": 0.0002979666962559002,
"loss": 2.0015987396240233,
"step": 4290
},
{
"epoch": 0.5504,
"grad_norm": 0.6617387533187866,
"learning_rate": 0.0002979567615253501,
"loss": 2.0117170333862306,
"step": 4300
},
{
"epoch": 0.55168,
"grad_norm": 0.8098113536834717,
"learning_rate": 0.00029794680274977664,
"loss": 1.976462173461914,
"step": 4310
},
{
"epoch": 0.55296,
"grad_norm": 0.7238470911979675,
"learning_rate": 0.00029793681993079817,
"loss": 1.9960739135742187,
"step": 4320
},
{
"epoch": 0.55424,
"grad_norm": 0.8663386702537537,
"learning_rate": 0.0002979268130700371,
"loss": 1.9836992263793944,
"step": 4330
},
{
"epoch": 0.55552,
"grad_norm": 0.7530176043510437,
"learning_rate": 0.00029791678216911974,
"loss": 1.993238639831543,
"step": 4340
},
{
"epoch": 0.5568,
"grad_norm": 0.8118612766265869,
"learning_rate": 0.0002979067272296761,
"loss": 2.0035892486572267,
"step": 4350
},
{
"epoch": 0.55808,
"grad_norm": 0.7913742065429688,
"learning_rate": 0.0002978966482533404,
"loss": 2.002780532836914,
"step": 4360
},
{
"epoch": 0.55936,
"grad_norm": 0.694092333316803,
"learning_rate": 0.0002978865452417505,
"loss": 2.0032726287841798,
"step": 4370
},
{
"epoch": 0.56064,
"grad_norm": 0.6410046815872192,
"learning_rate": 0.0002978764181965483,
"loss": 2.001888465881348,
"step": 4380
},
{
"epoch": 0.56192,
"grad_norm": 0.7601751089096069,
"learning_rate": 0.00029786626711937966,
"loss": 2.0106828689575194,
"step": 4390
},
{
"epoch": 0.5632,
"grad_norm": 0.7212942838668823,
"learning_rate": 0.00029785609201189424,
"loss": 2.023110198974609,
"step": 4400
},
{
"epoch": 0.56448,
"grad_norm": 0.7451926469802856,
"learning_rate": 0.0002978458928757456,
"loss": 2.011552619934082,
"step": 4410
},
{
"epoch": 0.56576,
"grad_norm": 0.6781797409057617,
"learning_rate": 0.00029783566971259125,
"loss": 2.0140331268310545,
"step": 4420
},
{
"epoch": 0.56704,
"grad_norm": 0.7447178959846497,
"learning_rate": 0.0002978254225240926,
"loss": 2.0165130615234377,
"step": 4430
},
{
"epoch": 0.56832,
"grad_norm": 0.7245031595230103,
"learning_rate": 0.0002978151513119149,
"loss": 1.9940263748168945,
"step": 4440
},
{
"epoch": 0.5696,
"grad_norm": 0.756145715713501,
"learning_rate": 0.0002978048560777275,
"loss": 1.975267219543457,
"step": 4450
},
{
"epoch": 0.57088,
"grad_norm": 0.7302019000053406,
"learning_rate": 0.00029779453682320345,
"loss": 1.992108154296875,
"step": 4460
},
{
"epoch": 0.57216,
"grad_norm": 0.8035944104194641,
"learning_rate": 0.0002977841935500197,
"loss": 1.9908878326416015,
"step": 4470
},
{
"epoch": 0.57344,
"grad_norm": 0.7327392101287842,
"learning_rate": 0.00029777382625985734,
"loss": 2.000600051879883,
"step": 4480
},
{
"epoch": 0.57472,
"grad_norm": 0.6741060018539429,
"learning_rate": 0.00029776343495440104,
"loss": 2.0095552444458007,
"step": 4490
},
{
"epoch": 0.576,
"grad_norm": 0.6768994331359863,
"learning_rate": 0.00029775301963533957,
"loss": 1.983211135864258,
"step": 4500
},
{
"epoch": 0.57728,
"grad_norm": 0.7285369634628296,
"learning_rate": 0.0002977425803043657,
"loss": 2.011127471923828,
"step": 4510
},
{
"epoch": 0.57856,
"grad_norm": 0.6478345394134521,
"learning_rate": 0.00029773211696317573,
"loss": 1.9803707122802734,
"step": 4520
},
{
"epoch": 0.57984,
"grad_norm": 0.7388439774513245,
"learning_rate": 0.0002977216296134703,
"loss": 1.9949949264526368,
"step": 4530
},
{
"epoch": 0.58112,
"grad_norm": 0.6757745742797852,
"learning_rate": 0.0002977111182569536,
"loss": 2.025724983215332,
"step": 4540
},
{
"epoch": 0.5824,
"grad_norm": 0.6242362856864929,
"learning_rate": 0.00029770058289533397,
"loss": 2.0046735763549806,
"step": 4550
},
{
"epoch": 0.58368,
"grad_norm": 0.6959071159362793,
"learning_rate": 0.0002976900235303235,
"loss": 1.9909477233886719,
"step": 4560
},
{
"epoch": 0.58496,
"grad_norm": 1.1365975141525269,
"learning_rate": 0.0002976794401636383,
"loss": 2.0057315826416016,
"step": 4570
},
{
"epoch": 0.58624,
"grad_norm": 0.6782042384147644,
"learning_rate": 0.00029766883279699826,
"loss": 2.0111324310302736,
"step": 4580
},
{
"epoch": 0.58752,
"grad_norm": 1.024519681930542,
"learning_rate": 0.0002976582014321272,
"loss": 1.995925521850586,
"step": 4590
},
{
"epoch": 0.5888,
"grad_norm": 0.7553693056106567,
"learning_rate": 0.0002976475460707529,
"loss": 2.0124210357666015,
"step": 4600
},
{
"epoch": 0.59008,
"grad_norm": 0.6609204411506653,
"learning_rate": 0.00029763686671460703,
"loss": 1.9881935119628906,
"step": 4610
},
{
"epoch": 0.59136,
"grad_norm": 0.7214299440383911,
"learning_rate": 0.00029762616336542514,
"loss": 1.9801551818847656,
"step": 4620
},
{
"epoch": 0.59264,
"grad_norm": 0.6682612895965576,
"learning_rate": 0.00029761543602494655,
"loss": 2.023514175415039,
"step": 4630
},
{
"epoch": 0.59392,
"grad_norm": 0.7894415855407715,
"learning_rate": 0.0002976046846949148,
"loss": 2.010751724243164,
"step": 4640
},
{
"epoch": 0.5952,
"grad_norm": 0.6574305891990662,
"learning_rate": 0.0002975939093770769,
"loss": 1.976632308959961,
"step": 4650
},
{
"epoch": 0.59648,
"grad_norm": 0.6413716673851013,
"learning_rate": 0.00029758311007318416,
"loss": 2.0151065826416015,
"step": 4660
},
{
"epoch": 0.59776,
"grad_norm": 0.7089874148368835,
"learning_rate": 0.00029757228678499155,
"loss": 1.9980993270874023,
"step": 4670
},
{
"epoch": 0.59904,
"grad_norm": 0.6705209612846375,
"learning_rate": 0.000297561439514258,
"loss": 2.0038993835449217,
"step": 4680
},
{
"epoch": 0.60032,
"grad_norm": 0.6938350796699524,
"learning_rate": 0.00029755056826274637,
"loss": 1.9767511367797852,
"step": 4690
},
{
"epoch": 0.6016,
"grad_norm": 0.733703076839447,
"learning_rate": 0.0002975396730322234,
"loss": 2.0173906326293944,
"step": 4700
},
{
"epoch": 0.60288,
"grad_norm": 0.6944065093994141,
"learning_rate": 0.0002975287538244597,
"loss": 2.012947654724121,
"step": 4710
},
{
"epoch": 0.60416,
"grad_norm": 0.7177563905715942,
"learning_rate": 0.0002975178106412298,
"loss": 1.9930303573608399,
"step": 4720
},
{
"epoch": 0.60544,
"grad_norm": 0.7881021499633789,
"learning_rate": 0.000297506843484312,
"loss": 2.0142608642578126,
"step": 4730
},
{
"epoch": 0.60672,
"grad_norm": 0.7584002614021301,
"learning_rate": 0.00029749585235548887,
"loss": 2.0150390625,
"step": 4740
},
{
"epoch": 0.608,
"grad_norm": 0.7548288106918335,
"learning_rate": 0.00029748483725654636,
"loss": 2.001949691772461,
"step": 4750
},
{
"epoch": 0.60928,
"grad_norm": 0.7352362871170044,
"learning_rate": 0.00029747379818927474,
"loss": 2.014604187011719,
"step": 4760
},
{
"epoch": 0.61056,
"grad_norm": 0.736960768699646,
"learning_rate": 0.00029746273515546793,
"loss": 2.0182096481323244,
"step": 4770
},
{
"epoch": 0.61184,
"grad_norm": 0.8847310543060303,
"learning_rate": 0.00029745164815692387,
"loss": 1.9898075103759765,
"step": 4780
},
{
"epoch": 0.61312,
"grad_norm": 0.7176423072814941,
"learning_rate": 0.0002974405371954443,
"loss": 2.003055953979492,
"step": 4790
},
{
"epoch": 0.6144,
"grad_norm": 0.7516958713531494,
"learning_rate": 0.00029742940227283504,
"loss": 1.9840295791625977,
"step": 4800
},
{
"epoch": 0.61568,
"grad_norm": 0.7542930245399475,
"learning_rate": 0.0002974182433909055,
"loss": 2.0189092636108397,
"step": 4810
},
{
"epoch": 0.61696,
"grad_norm": 0.7811306118965149,
"learning_rate": 0.0002974070605514692,
"loss": 2.005120849609375,
"step": 4820
},
{
"epoch": 0.61824,
"grad_norm": 0.6935572028160095,
"learning_rate": 0.00029739585375634356,
"loss": 2.0136289596557617,
"step": 4830
},
{
"epoch": 0.61952,
"grad_norm": 0.9725347757339478,
"learning_rate": 0.0002973846230073498,
"loss": 2.0010726928710936,
"step": 4840
},
{
"epoch": 0.6208,
"grad_norm": 0.6915906071662903,
"learning_rate": 0.0002973733683063131,
"loss": 2.010584259033203,
"step": 4850
},
{
"epoch": 0.62208,
"grad_norm": 0.771823525428772,
"learning_rate": 0.00029736208965506247,
"loss": 2.0161518096923827,
"step": 4860
},
{
"epoch": 0.62336,
"grad_norm": 0.6337702870368958,
"learning_rate": 0.00029735078705543086,
"loss": 2.0008981704711912,
"step": 4870
},
{
"epoch": 0.62464,
"grad_norm": 0.6666226387023926,
"learning_rate": 0.0002973394605092551,
"loss": 2.010310173034668,
"step": 4880
},
{
"epoch": 0.62592,
"grad_norm": 0.7635229825973511,
"learning_rate": 0.00029732811001837594,
"loss": 2.0058834075927736,
"step": 4890
},
{
"epoch": 0.6272,
"grad_norm": 0.9127666354179382,
"learning_rate": 0.0002973167355846379,
"loss": 2.0224124908447267,
"step": 4900
},
{
"epoch": 0.62848,
"grad_norm": 0.8189308047294617,
"learning_rate": 0.0002973053372098896,
"loss": 2.007943344116211,
"step": 4910
},
{
"epoch": 0.62976,
"grad_norm": 0.7304124236106873,
"learning_rate": 0.0002972939148959834,
"loss": 2.000100517272949,
"step": 4920
},
{
"epoch": 0.63104,
"grad_norm": 0.6603564620018005,
"learning_rate": 0.00029728246864477555,
"loss": 2.0129776000976562,
"step": 4930
},
{
"epoch": 0.63232,
"grad_norm": 0.6280004382133484,
"learning_rate": 0.00029727099845812616,
"loss": 2.0045780181884765,
"step": 4940
},
{
"epoch": 0.6336,
"grad_norm": 0.7431449294090271,
"learning_rate": 0.00029725950433789954,
"loss": 1.98687744140625,
"step": 4950
},
{
"epoch": 0.63488,
"grad_norm": 0.8049905896186829,
"learning_rate": 0.0002972479862859634,
"loss": 2.004106330871582,
"step": 4960
},
{
"epoch": 0.63616,
"grad_norm": 0.7185505628585815,
"learning_rate": 0.00029723644430418956,
"loss": 1.998849868774414,
"step": 4970
},
{
"epoch": 0.63744,
"grad_norm": 0.8346318602561951,
"learning_rate": 0.0002972248783944539,
"loss": 1.9910211563110352,
"step": 4980
},
{
"epoch": 0.63872,
"grad_norm": 0.686622142791748,
"learning_rate": 0.0002972132885586361,
"loss": 1.9930900573730468,
"step": 4990
},
{
"epoch": 0.64,
"grad_norm": 0.7573023438453674,
"learning_rate": 0.00029720167479861946,
"loss": 2.01367244720459,
"step": 5000
},
{
"epoch": 0.64128,
"grad_norm": 0.7065608501434326,
"learning_rate": 0.0002971900371162915,
"loss": 1.9992128372192384,
"step": 5010
},
{
"epoch": 0.64256,
"grad_norm": 0.7289401292800903,
"learning_rate": 0.00029717837551354354,
"loss": 1.9842073440551757,
"step": 5020
},
{
"epoch": 0.64384,
"grad_norm": 0.6458973288536072,
"learning_rate": 0.00029716668999227064,
"loss": 2.0016260147094727,
"step": 5030
},
{
"epoch": 0.64512,
"grad_norm": 0.7114098072052002,
"learning_rate": 0.0002971549805543719,
"loss": 1.9759035110473633,
"step": 5040
},
{
"epoch": 0.6464,
"grad_norm": 0.7502028346061707,
"learning_rate": 0.0002971432472017503,
"loss": 1.9977811813354491,
"step": 5050
},
{
"epoch": 0.64768,
"grad_norm": 0.7050338387489319,
"learning_rate": 0.00029713148993631267,
"loss": 2.0023754119873045,
"step": 5060
},
{
"epoch": 0.64896,
"grad_norm": 0.7351704835891724,
"learning_rate": 0.0002971197087599697,
"loss": 1.9990671157836915,
"step": 5070
},
{
"epoch": 0.65024,
"grad_norm": 0.684573769569397,
"learning_rate": 0.000297107903674636,
"loss": 2.0173965454101563,
"step": 5080
},
{
"epoch": 0.65152,
"grad_norm": 0.7622045278549194,
"learning_rate": 0.0002970960746822301,
"loss": 1.9990699768066407,
"step": 5090
},
{
"epoch": 0.6528,
"grad_norm": 0.5931624174118042,
"learning_rate": 0.00029708422178467426,
"loss": 2.0040538787841795,
"step": 5100
},
{
"epoch": 0.65408,
"grad_norm": 0.6407083868980408,
"learning_rate": 0.0002970723449838949,
"loss": 2.00338191986084,
"step": 5110
},
{
"epoch": 0.65536,
"grad_norm": 0.7045226097106934,
"learning_rate": 0.0002970604442818221,
"loss": 1.996185874938965,
"step": 5120
},
{
"epoch": 0.65664,
"grad_norm": 0.6836894154548645,
"learning_rate": 0.0002970485196803898,
"loss": 1.994027328491211,
"step": 5130
},
{
"epoch": 0.65792,
"grad_norm": 0.6975918412208557,
"learning_rate": 0.000297036571181536,
"loss": 1.9909107208251953,
"step": 5140
},
{
"epoch": 0.6592,
"grad_norm": 0.7193573713302612,
"learning_rate": 0.00029702459878720255,
"loss": 1.9966390609741211,
"step": 5150
},
{
"epoch": 0.66048,
"grad_norm": 0.6120880246162415,
"learning_rate": 0.000297012602499335,
"loss": 1.9817461013793944,
"step": 5160
},
{
"epoch": 0.66176,
"grad_norm": 0.759924054145813,
"learning_rate": 0.000297000582319883,
"loss": 1.9850471496582032,
"step": 5170
},
{
"epoch": 0.66304,
"grad_norm": 0.6980900764465332,
"learning_rate": 0.0002969885382508,
"loss": 1.9931489944458007,
"step": 5180
},
{
"epoch": 0.66432,
"grad_norm": 0.7123715877532959,
"learning_rate": 0.00029697647029404325,
"loss": 2.000169563293457,
"step": 5190
},
{
"epoch": 0.6656,
"grad_norm": 0.6090735197067261,
"learning_rate": 0.000296964378451574,
"loss": 1.9939422607421875,
"step": 5200
},
{
"epoch": 0.66688,
"grad_norm": 0.7080000042915344,
"learning_rate": 0.0002969522627253574,
"loss": 1.9911760330200194,
"step": 5210
},
{
"epoch": 0.66816,
"grad_norm": 0.6564937233924866,
"learning_rate": 0.0002969401231173624,
"loss": 1.9953861236572266,
"step": 5220
},
{
"epoch": 0.66944,
"grad_norm": 0.6857775449752808,
"learning_rate": 0.0002969279596295618,
"loss": 2.0081436157226564,
"step": 5230
},
{
"epoch": 0.67072,
"grad_norm": 0.7370122671127319,
"learning_rate": 0.00029691577226393235,
"loss": 2.0082754135131835,
"step": 5240
},
{
"epoch": 0.672,
"grad_norm": 0.6324231624603271,
"learning_rate": 0.0002969035610224547,
"loss": 2.0048879623413085,
"step": 5250
},
{
"epoch": 0.67328,
"grad_norm": 0.7700392007827759,
"learning_rate": 0.0002968913259071133,
"loss": 2.0072811126708983,
"step": 5260
},
{
"epoch": 0.67456,
"grad_norm": 0.6762933731079102,
"learning_rate": 0.00029687906691989655,
"loss": 1.9954465866088866,
"step": 5270
},
{
"epoch": 0.67584,
"grad_norm": 0.7391619086265564,
"learning_rate": 0.00029686678406279674,
"loss": 1.9988208770751954,
"step": 5280
},
{
"epoch": 0.67712,
"grad_norm": 0.6812599301338196,
"learning_rate": 0.00029685447733780995,
"loss": 1.9795747756958009,
"step": 5290
},
{
"epoch": 0.6784,
"grad_norm": 0.6824685335159302,
"learning_rate": 0.00029684214674693623,
"loss": 2.0002880096435547,
"step": 5300
},
{
"epoch": 0.67968,
"grad_norm": 0.7883182168006897,
"learning_rate": 0.00029682979229217943,
"loss": 1.9892650604248048,
"step": 5310
},
{
"epoch": 0.68096,
"grad_norm": 0.8476542234420776,
"learning_rate": 0.0002968174139755474,
"loss": 2.0148120880126954,
"step": 5320
},
{
"epoch": 0.68224,
"grad_norm": 0.7227822542190552,
"learning_rate": 0.00029680501179905173,
"loss": 2.0036792755126953,
"step": 5330
},
{
"epoch": 0.68352,
"grad_norm": 0.6343380212783813,
"learning_rate": 0.00029679258576470794,
"loss": 2.0134708404541017,
"step": 5340
},
{
"epoch": 0.6848,
"grad_norm": 0.6354977488517761,
"learning_rate": 0.00029678013587453543,
"loss": 2.0034976959228517,
"step": 5350
},
{
"epoch": 0.68608,
"grad_norm": 0.7666685581207275,
"learning_rate": 0.0002967676621305575,
"loss": 1.9847345352172852,
"step": 5360
},
{
"epoch": 0.68736,
"grad_norm": 0.8441287875175476,
"learning_rate": 0.0002967551645348013,
"loss": 2.006691551208496,
"step": 5370
},
{
"epoch": 0.68864,
"grad_norm": 0.6251808404922485,
"learning_rate": 0.0002967426430892979,
"loss": 1.9834270477294922,
"step": 5380
},
{
"epoch": 0.68992,
"grad_norm": 0.636553168296814,
"learning_rate": 0.0002967300977960821,
"loss": 1.9941967010498047,
"step": 5390
},
{
"epoch": 0.6912,
"grad_norm": 0.6793889999389648,
"learning_rate": 0.0002967175286571928,
"loss": 2.016760063171387,
"step": 5400
},
{
"epoch": 0.69248,
"grad_norm": 0.7251984477043152,
"learning_rate": 0.0002967049356746726,
"loss": 1.9951417922973633,
"step": 5410
},
{
"epoch": 0.69376,
"grad_norm": 0.8330734372138977,
"learning_rate": 0.00029669231885056806,
"loss": 1.992695426940918,
"step": 5420
},
{
"epoch": 0.69504,
"grad_norm": 0.6929616928100586,
"learning_rate": 0.00029667967818692957,
"loss": 1.9796766281127929,
"step": 5430
},
{
"epoch": 0.69632,
"grad_norm": 0.73932945728302,
"learning_rate": 0.0002966670136858114,
"loss": 1.9927392959594727,
"step": 5440
},
{
"epoch": 0.6976,
"grad_norm": 0.728484034538269,
"learning_rate": 0.00029665432534927176,
"loss": 1.990020179748535,
"step": 5450
},
{
"epoch": 0.69888,
"grad_norm": 0.8292369842529297,
"learning_rate": 0.0002966416131793726,
"loss": 2.007571220397949,
"step": 5460
},
{
"epoch": 0.70016,
"grad_norm": 0.6958839297294617,
"learning_rate": 0.00029662887717818,
"loss": 1.9895818710327149,
"step": 5470
},
{
"epoch": 0.70144,
"grad_norm": 0.8625398874282837,
"learning_rate": 0.0002966161173477635,
"loss": 2.008391761779785,
"step": 5480
},
{
"epoch": 0.70272,
"grad_norm": 0.7190197110176086,
"learning_rate": 0.00029660333369019683,
"loss": 1.9892658233642577,
"step": 5490
},
{
"epoch": 0.704,
"grad_norm": 0.6254742741584778,
"learning_rate": 0.00029659052620755757,
"loss": 1.9928400039672851,
"step": 5500
},
{
"epoch": 0.70528,
"grad_norm": 0.7220839262008667,
"learning_rate": 0.0002965776949019271,
"loss": 1.9931129455566405,
"step": 5510
},
{
"epoch": 0.70656,
"grad_norm": 0.6875084042549133,
"learning_rate": 0.0002965648397753907,
"loss": 2.006221961975098,
"step": 5520
},
{
"epoch": 0.70784,
"grad_norm": 0.6967015862464905,
"learning_rate": 0.0002965519608300374,
"loss": 2.0144590377807616,
"step": 5530
},
{
"epoch": 0.70912,
"grad_norm": 0.7796663045883179,
"learning_rate": 0.0002965390580679603,
"loss": 1.9761390686035156,
"step": 5540
},
{
"epoch": 0.7104,
"grad_norm": 0.6932560801506042,
"learning_rate": 0.0002965261314912563,
"loss": 1.9977296829223632,
"step": 5550
},
{
"epoch": 0.71168,
"grad_norm": 0.6555649638175964,
"learning_rate": 0.00029651318110202605,
"loss": 1.9888612747192382,
"step": 5560
},
{
"epoch": 0.71296,
"grad_norm": 0.7425674796104431,
"learning_rate": 0.00029650020690237425,
"loss": 1.9898294448852538,
"step": 5570
},
{
"epoch": 0.71424,
"grad_norm": 0.754615068435669,
"learning_rate": 0.0002964872088944093,
"loss": 1.9967521667480468,
"step": 5580
},
{
"epoch": 0.71552,
"grad_norm": 0.6292251944541931,
"learning_rate": 0.0002964741870802437,
"loss": 2.0030208587646485,
"step": 5590
},
{
"epoch": 0.7168,
"grad_norm": 0.7437636256217957,
"learning_rate": 0.0002964611414619936,
"loss": 2.0063127517700194,
"step": 5600
},
{
"epoch": 0.71808,
"grad_norm": 0.6832345128059387,
"learning_rate": 0.000296448072041779,
"loss": 1.9996273040771484,
"step": 5610
},
{
"epoch": 0.71936,
"grad_norm": 0.6542320847511292,
"learning_rate": 0.000296434978821724,
"loss": 1.9821640014648438,
"step": 5620
},
{
"epoch": 0.72064,
"grad_norm": 0.6850648522377014,
"learning_rate": 0.0002964218618039564,
"loss": 1.9940942764282226,
"step": 5630
},
{
"epoch": 0.72192,
"grad_norm": 0.7427983283996582,
"learning_rate": 0.00029640872099060783,
"loss": 1.981001091003418,
"step": 5640
},
{
"epoch": 0.7232,
"grad_norm": 0.7683842182159424,
"learning_rate": 0.0002963955563838139,
"loss": 2.008561897277832,
"step": 5650
},
{
"epoch": 0.72448,
"grad_norm": 0.6730973124504089,
"learning_rate": 0.00029638236798571404,
"loss": 1.996862030029297,
"step": 5660
},
{
"epoch": 0.72576,
"grad_norm": 0.6657568216323853,
"learning_rate": 0.00029636915579845155,
"loss": 1.9981700897216796,
"step": 5670
},
{
"epoch": 0.72704,
"grad_norm": 0.7134661674499512,
"learning_rate": 0.0002963559198241736,
"loss": 1.9864439010620116,
"step": 5680
},
{
"epoch": 0.72832,
"grad_norm": 0.7284188866615295,
"learning_rate": 0.00029634266006503123,
"loss": 2.0042444229125977,
"step": 5690
},
{
"epoch": 0.7296,
"grad_norm": 0.7229200005531311,
"learning_rate": 0.0002963293765231793,
"loss": 1.9872091293334961,
"step": 5700
},
{
"epoch": 0.73088,
"grad_norm": 0.6603932976722717,
"learning_rate": 0.00029631606920077654,
"loss": 2.0120763778686523,
"step": 5710
},
{
"epoch": 0.73216,
"grad_norm": 0.8360553979873657,
"learning_rate": 0.00029630273809998567,
"loss": 1.9983232498168946,
"step": 5720
},
{
"epoch": 0.73344,
"grad_norm": 0.611983060836792,
"learning_rate": 0.0002962893832229731,
"loss": 1.9799180984497071,
"step": 5730
},
{
"epoch": 0.73472,
"grad_norm": 0.6826144456863403,
"learning_rate": 0.0002962760045719092,
"loss": 1.985218620300293,
"step": 5740
},
{
"epoch": 0.736,
"grad_norm": 0.6724496483802795,
"learning_rate": 0.0002962626021489682,
"loss": 1.9969600677490233,
"step": 5750
},
{
"epoch": 0.73728,
"grad_norm": 0.6387225389480591,
"learning_rate": 0.0002962491759563282,
"loss": 2.0010164260864256,
"step": 5760
},
{
"epoch": 0.73856,
"grad_norm": 0.7522284388542175,
"learning_rate": 0.00029623572599617106,
"loss": 2.0010658264160157,
"step": 5770
},
{
"epoch": 0.73984,
"grad_norm": 0.6354997158050537,
"learning_rate": 0.0002962222522706827,
"loss": 2.015804481506348,
"step": 5780
},
{
"epoch": 0.74112,
"grad_norm": 0.671524703502655,
"learning_rate": 0.00029620875478205273,
"loss": 2.0011077880859376,
"step": 5790
},
{
"epoch": 0.7424,
"grad_norm": 0.6422104835510254,
"learning_rate": 0.00029619523353247463,
"loss": 1.9921817779541016,
"step": 5800
},
{
"epoch": 0.74368,
"grad_norm": 0.6711571216583252,
"learning_rate": 0.0002961816885241458,
"loss": 1.9803316116333007,
"step": 5810
},
{
"epoch": 0.74496,
"grad_norm": 0.8022178411483765,
"learning_rate": 0.00029616811975926763,
"loss": 1.9788673400878907,
"step": 5820
},
{
"epoch": 0.74624,
"grad_norm": 0.6730902194976807,
"learning_rate": 0.0002961545272400451,
"loss": 1.9903987884521483,
"step": 5830
},
{
"epoch": 0.74752,
"grad_norm": 0.7900033593177795,
"learning_rate": 0.0002961409109686872,
"loss": 2.013372802734375,
"step": 5840
},
{
"epoch": 0.7488,
"grad_norm": 0.6188972592353821,
"learning_rate": 0.00029612727094740674,
"loss": 1.9988031387329102,
"step": 5850
},
{
"epoch": 0.75008,
"grad_norm": 0.6830716729164124,
"learning_rate": 0.00029611360717842047,
"loss": 1.9980751037597657,
"step": 5860
},
{
"epoch": 0.75136,
"grad_norm": 0.7211880683898926,
"learning_rate": 0.00029609991966394897,
"loss": 2.0045663833618166,
"step": 5870
},
{
"epoch": 0.75264,
"grad_norm": 0.6811301708221436,
"learning_rate": 0.0002960862084062166,
"loss": 2.0147485733032227,
"step": 5880
},
{
"epoch": 0.75392,
"grad_norm": 0.6604957580566406,
"learning_rate": 0.0002960724734074516,
"loss": 1.987942886352539,
"step": 5890
},
{
"epoch": 0.7552,
"grad_norm": 0.7545401453971863,
"learning_rate": 0.0002960587146698861,
"loss": 2.0053482055664062,
"step": 5900
},
{
"epoch": 0.75648,
"grad_norm": 0.6331950426101685,
"learning_rate": 0.00029604493219575616,
"loss": 1.9871101379394531,
"step": 5910
},
{
"epoch": 0.75776,
"grad_norm": 0.6771983504295349,
"learning_rate": 0.0002960311259873015,
"loss": 1.9838722229003907,
"step": 5920
},
{
"epoch": 0.75904,
"grad_norm": 0.7738776206970215,
"learning_rate": 0.000296017296046766,
"loss": 2.0045019149780274,
"step": 5930
},
{
"epoch": 0.76032,
"grad_norm": 0.7317860722541809,
"learning_rate": 0.0002960034423763971,
"loss": 2.015328788757324,
"step": 5940
},
{
"epoch": 0.7616,
"grad_norm": 0.6838796138763428,
"learning_rate": 0.00029598956497844616,
"loss": 1.993131446838379,
"step": 5950
},
{
"epoch": 0.76288,
"grad_norm": 0.7081223130226135,
"learning_rate": 0.00029597566385516856,
"loss": 1.9932073593139648,
"step": 5960
},
{
"epoch": 0.76416,
"grad_norm": 0.7130403518676758,
"learning_rate": 0.0002959617390088234,
"loss": 1.995915985107422,
"step": 5970
},
{
"epoch": 0.76544,
"grad_norm": 0.6577734351158142,
"learning_rate": 0.0002959477904416736,
"loss": 2.001402473449707,
"step": 5980
},
{
"epoch": 0.76672,
"grad_norm": 0.8433389663696289,
"learning_rate": 0.00029593381815598606,
"loss": 1.9908720016479493,
"step": 5990
},
{
"epoch": 0.768,
"grad_norm": 0.7578377723693848,
"learning_rate": 0.00029591982215403144,
"loss": 1.9951087951660156,
"step": 6000
},
{
"epoch": 0.76928,
"grad_norm": 0.6299430131912231,
"learning_rate": 0.0002959058024380843,
"loss": 1.9930356979370116,
"step": 6010
},
{
"epoch": 0.77056,
"grad_norm": 0.7663775682449341,
"learning_rate": 0.00029589175901042304,
"loss": 1.994969940185547,
"step": 6020
},
{
"epoch": 0.77184,
"grad_norm": 0.7576585412025452,
"learning_rate": 0.0002958776918733299,
"loss": 2.008650779724121,
"step": 6030
},
{
"epoch": 0.77312,
"grad_norm": 1.5998573303222656,
"learning_rate": 0.00029586360102909097,
"loss": 1.9937719345092773,
"step": 6040
},
{
"epoch": 0.7744,
"grad_norm": 0.8406224250793457,
"learning_rate": 0.0002958494864799962,
"loss": 2.0037378311157226,
"step": 6050
},
{
"epoch": 0.77568,
"grad_norm": 0.7180790901184082,
"learning_rate": 0.00029583534822833944,
"loss": 2.002363586425781,
"step": 6060
},
{
"epoch": 0.77696,
"grad_norm": 0.604666531085968,
"learning_rate": 0.0002958211862764183,
"loss": 1.9915838241577148,
"step": 6070
},
{
"epoch": 0.77824,
"grad_norm": 0.6440698504447937,
"learning_rate": 0.0002958070006265344,
"loss": 1.9986055374145508,
"step": 6080
},
{
"epoch": 0.77952,
"grad_norm": 0.6588718891143799,
"learning_rate": 0.000295792791280993,
"loss": 1.9881462097167968,
"step": 6090
},
{
"epoch": 0.7808,
"grad_norm": 0.7286064624786377,
"learning_rate": 0.0002957785582421034,
"loss": 2.000356674194336,
"step": 6100
},
{
"epoch": 0.78208,
"grad_norm": 0.8021141886711121,
"learning_rate": 0.0002957643015121785,
"loss": 1.9829841613769532,
"step": 6110
},
{
"epoch": 0.78336,
"grad_norm": 0.7439085245132446,
"learning_rate": 0.0002957500210935354,
"loss": 1.988357162475586,
"step": 6120
},
{
"epoch": 0.78464,
"grad_norm": 0.6714260578155518,
"learning_rate": 0.00029573571698849477,
"loss": 2.0017559051513674,
"step": 6130
},
{
"epoch": 0.78592,
"grad_norm": 0.654412031173706,
"learning_rate": 0.0002957213891993813,
"loss": 1.9880069732666015,
"step": 6140
},
{
"epoch": 0.7872,
"grad_norm": 0.6551602482795715,
"learning_rate": 0.00029570703772852336,
"loss": 2.0161787033081056,
"step": 6150
},
{
"epoch": 0.78848,
"grad_norm": 0.6666812300682068,
"learning_rate": 0.00029569266257825335,
"loss": 1.9900634765625,
"step": 6160
},
{
"epoch": 0.78976,
"grad_norm": 0.7876213192939758,
"learning_rate": 0.00029567826375090736,
"loss": 2.003603935241699,
"step": 6170
},
{
"epoch": 0.79104,
"grad_norm": 1.2557580471038818,
"learning_rate": 0.0002956638412488254,
"loss": 1.9852615356445313,
"step": 6180
},
{
"epoch": 0.79232,
"grad_norm": 0.6098614931106567,
"learning_rate": 0.0002956493950743514,
"loss": 1.9911855697631835,
"step": 6190
},
{
"epoch": 0.7936,
"grad_norm": 0.7123518586158752,
"learning_rate": 0.00029563492522983304,
"loss": 1.9931888580322266,
"step": 6200
},
{
"epoch": 0.79488,
"grad_norm": 0.6955991983413696,
"learning_rate": 0.0002956204317176218,
"loss": 2.0002986907958986,
"step": 6210
},
{
"epoch": 0.79616,
"grad_norm": 0.6723616123199463,
"learning_rate": 0.0002956059145400731,
"loss": 2.0050445556640626,
"step": 6220
},
{
"epoch": 0.79744,
"grad_norm": 0.7482268810272217,
"learning_rate": 0.0002955913736995463,
"loss": 1.996047592163086,
"step": 6230
},
{
"epoch": 0.79872,
"grad_norm": 0.6395532488822937,
"learning_rate": 0.00029557680919840433,
"loss": 2.012595558166504,
"step": 6240
},
{
"epoch": 0.8,
"grad_norm": 0.6754005551338196,
"learning_rate": 0.00029556222103901416,
"loss": 1.9906475067138671,
"step": 6250
},
{
"epoch": 0.80128,
"grad_norm": 0.6359081864356995,
"learning_rate": 0.00029554760922374665,
"loss": 1.979934310913086,
"step": 6260
},
{
"epoch": 0.80256,
"grad_norm": 0.5925866365432739,
"learning_rate": 0.0002955329737549764,
"loss": 2.0107406616210937,
"step": 6270
},
{
"epoch": 0.80384,
"grad_norm": 0.7143832445144653,
"learning_rate": 0.00029551831463508175,
"loss": 1.9844533920288085,
"step": 6280
},
{
"epoch": 0.80512,
"grad_norm": 0.7044582366943359,
"learning_rate": 0.0002955036318664452,
"loss": 1.9771629333496095,
"step": 6290
},
{
"epoch": 0.8064,
"grad_norm": 0.8326923847198486,
"learning_rate": 0.0002954889254514527,
"loss": 1.9971778869628907,
"step": 6300
},
{
"epoch": 0.80768,
"grad_norm": 0.682141900062561,
"learning_rate": 0.00029547419539249444,
"loss": 1.9877017974853515,
"step": 6310
},
{
"epoch": 0.80896,
"grad_norm": 0.664592981338501,
"learning_rate": 0.0002954594416919642,
"loss": 1.9961456298828124,
"step": 6320
},
{
"epoch": 0.81024,
"grad_norm": 0.6965521574020386,
"learning_rate": 0.0002954446643522596,
"loss": 1.9998743057250976,
"step": 6330
},
{
"epoch": 0.81152,
"grad_norm": 0.6839595437049866,
"learning_rate": 0.00029542986337578216,
"loss": 1.9827798843383788,
"step": 6340
},
{
"epoch": 0.8128,
"grad_norm": 0.7168360948562622,
"learning_rate": 0.0002954150387649373,
"loss": 1.9969907760620118,
"step": 6350
},
{
"epoch": 0.81408,
"grad_norm": 0.7123169302940369,
"learning_rate": 0.0002954001905221342,
"loss": 1.9951913833618165,
"step": 6360
},
{
"epoch": 0.81536,
"grad_norm": 0.6068657040596008,
"learning_rate": 0.0002953853186497859,
"loss": 2.000649642944336,
"step": 6370
},
{
"epoch": 0.81664,
"grad_norm": 0.6289076209068298,
"learning_rate": 0.0002953704231503093,
"loss": 2.0178243637084963,
"step": 6380
},
{
"epoch": 0.81792,
"grad_norm": 0.6308927536010742,
"learning_rate": 0.00029535550402612517,
"loss": 1.9915205001831056,
"step": 6390
},
{
"epoch": 0.8192,
"grad_norm": 0.616354763507843,
"learning_rate": 0.00029534056127965795,
"loss": 2.008121299743652,
"step": 6400
},
{
"epoch": 0.82048,
"grad_norm": 0.8025171756744385,
"learning_rate": 0.0002953255949133362,
"loss": 1.9940040588378907,
"step": 6410
},
{
"epoch": 0.82176,
"grad_norm": 0.5960074663162231,
"learning_rate": 0.00029531060492959206,
"loss": 2.007202911376953,
"step": 6420
},
{
"epoch": 0.82304,
"grad_norm": 0.7320867776870728,
"learning_rate": 0.0002952955913308616,
"loss": 2.001862144470215,
"step": 6430
},
{
"epoch": 0.82432,
"grad_norm": 0.6877573132514954,
"learning_rate": 0.0002952805541195848,
"loss": 1.9913444519042969,
"step": 6440
},
{
"epoch": 0.8256,
"grad_norm": 0.7313185334205627,
"learning_rate": 0.0002952654932982053,
"loss": 2.0005752563476564,
"step": 6450
},
{
"epoch": 0.82688,
"grad_norm": 0.6951473951339722,
"learning_rate": 0.00029525040886917083,
"loss": 1.9982521057128906,
"step": 6460
},
{
"epoch": 0.82816,
"grad_norm": 0.6680341362953186,
"learning_rate": 0.0002952353008349328,
"loss": 2.009914779663086,
"step": 6470
},
{
"epoch": 0.82944,
"grad_norm": 0.7091737389564514,
"learning_rate": 0.0002952201691979464,
"loss": 1.9832798004150392,
"step": 6480
},
{
"epoch": 0.83072,
"grad_norm": 0.7563579678535461,
"learning_rate": 0.00029520501396067077,
"loss": 1.9979364395141601,
"step": 6490
},
{
"epoch": 0.832,
"grad_norm": 0.6675616502761841,
"learning_rate": 0.00029518983512556886,
"loss": 2.0016298294067383,
"step": 6500
},
{
"epoch": 0.83328,
"grad_norm": 0.7811965346336365,
"learning_rate": 0.0002951746326951074,
"loss": 2.0019750595092773,
"step": 6510
},
{
"epoch": 0.83456,
"grad_norm": 0.6802430748939514,
"learning_rate": 0.00029515940667175706,
"loss": 1.9861677169799805,
"step": 6520
},
{
"epoch": 0.83584,
"grad_norm": 2.2662882804870605,
"learning_rate": 0.0002951441570579922,
"loss": 1.9777637481689454,
"step": 6530
},
{
"epoch": 0.83712,
"grad_norm": 0.6255841851234436,
"learning_rate": 0.0002951288838562912,
"loss": 1.9651605606079101,
"step": 6540
},
{
"epoch": 0.8384,
"grad_norm": 0.7399203777313232,
"learning_rate": 0.00029511358706913604,
"loss": 2.0162445068359376,
"step": 6550
},
{
"epoch": 0.83968,
"grad_norm": 0.6955289244651794,
"learning_rate": 0.0002950982666990127,
"loss": 2.0194833755493162,
"step": 6560
},
{
"epoch": 0.84096,
"grad_norm": 0.8125286102294922,
"learning_rate": 0.00029508292274841105,
"loss": 1.9961944580078126,
"step": 6570
},
{
"epoch": 0.84224,
"grad_norm": 0.6711224317550659,
"learning_rate": 0.00029506755521982453,
"loss": 1.9958179473876954,
"step": 6580
},
{
"epoch": 0.84352,
"grad_norm": 0.7141205668449402,
"learning_rate": 0.00029505216411575074,
"loss": 1.9968080520629883,
"step": 6590
},
{
"epoch": 0.8448,
"grad_norm": 0.6302902698516846,
"learning_rate": 0.0002950367494386909,
"loss": 1.9921850204467773,
"step": 6600
},
{
"epoch": 0.84608,
"grad_norm": 0.6486519575119019,
"learning_rate": 0.00029502131119114996,
"loss": 1.9878908157348634,
"step": 6610
},
{
"epoch": 0.84736,
"grad_norm": 0.6470666527748108,
"learning_rate": 0.00029500584937563703,
"loss": 2.011512565612793,
"step": 6620
},
{
"epoch": 0.84864,
"grad_norm": 0.6338657140731812,
"learning_rate": 0.0002949903639946649,
"loss": 2.005690574645996,
"step": 6630
},
{
"epoch": 0.84992,
"grad_norm": 0.7112775444984436,
"learning_rate": 0.00029497485505074997,
"loss": 1.9783432006835937,
"step": 6640
},
{
"epoch": 0.8512,
"grad_norm": 0.7458258271217346,
"learning_rate": 0.0002949593225464128,
"loss": 1.9891368865966796,
"step": 6650
},
{
"epoch": 0.85248,
"grad_norm": 0.7250550389289856,
"learning_rate": 0.0002949437664841775,
"loss": 2.01182861328125,
"step": 6660
},
{
"epoch": 0.85376,
"grad_norm": 0.7721503973007202,
"learning_rate": 0.0002949281868665723,
"loss": 1.9848461151123047,
"step": 6670
},
{
"epoch": 0.85504,
"grad_norm": 0.7179070115089417,
"learning_rate": 0.00029491258369612907,
"loss": 2.006007194519043,
"step": 6680
},
{
"epoch": 0.85632,
"grad_norm": 0.7209992408752441,
"learning_rate": 0.0002948969569753835,
"loss": 2.0215726852416993,
"step": 6690
},
{
"epoch": 0.8576,
"grad_norm": 0.6615075469017029,
"learning_rate": 0.00029488130670687516,
"loss": 1.9979560852050782,
"step": 6700
},
{
"epoch": 0.85888,
"grad_norm": 0.6356847286224365,
"learning_rate": 0.00029486563289314745,
"loss": 1.9941680908203125,
"step": 6710
},
{
"epoch": 0.86016,
"grad_norm": 0.6343762874603271,
"learning_rate": 0.00029484993553674754,
"loss": 1.9781364440917968,
"step": 6720
},
{
"epoch": 0.86144,
"grad_norm": 0.6580159664154053,
"learning_rate": 0.0002948342146402265,
"loss": 2.006149673461914,
"step": 6730
},
{
"epoch": 0.86272,
"grad_norm": 0.6490921974182129,
"learning_rate": 0.00029481847020613924,
"loss": 1.991459274291992,
"step": 6740
},
{
"epoch": 0.864,
"grad_norm": 0.6659132838249207,
"learning_rate": 0.0002948027022370443,
"loss": 1.977961540222168,
"step": 6750
},
{
"epoch": 0.86528,
"grad_norm": 0.6164100170135498,
"learning_rate": 0.0002947869107355044,
"loss": 2.012508773803711,
"step": 6760
},
{
"epoch": 0.86656,
"grad_norm": 0.6852055191993713,
"learning_rate": 0.0002947710957040857,
"loss": 1.9905057907104493,
"step": 6770
},
{
"epoch": 0.86784,
"grad_norm": 0.6740121841430664,
"learning_rate": 0.00029475525714535845,
"loss": 2.011355400085449,
"step": 6780
},
{
"epoch": 0.86912,
"grad_norm": 0.6753290891647339,
"learning_rate": 0.0002947393950618967,
"loss": 1.9824066162109375,
"step": 6790
},
{
"epoch": 0.8704,
"grad_norm": 0.6230936646461487,
"learning_rate": 0.0002947235094562781,
"loss": 1.9923259735107421,
"step": 6800
},
{
"epoch": 0.87168,
"grad_norm": 0.6286994218826294,
"learning_rate": 0.00029470760033108433,
"loss": 1.9875059127807617,
"step": 6810
},
{
"epoch": 0.87296,
"grad_norm": 0.8378221988677979,
"learning_rate": 0.0002946916676889009,
"loss": 1.992207145690918,
"step": 6820
},
{
"epoch": 0.87424,
"grad_norm": 0.6988818645477295,
"learning_rate": 0.00029467571153231705,
"loss": 1.9979055404663086,
"step": 6830
},
{
"epoch": 0.87552,
"grad_norm": 0.6570968627929688,
"learning_rate": 0.0002946597318639259,
"loss": 1.9981571197509767,
"step": 6840
},
{
"epoch": 0.8768,
"grad_norm": 0.6940664052963257,
"learning_rate": 0.0002946437286863243,
"loss": 1.9830677032470703,
"step": 6850
},
{
"epoch": 0.87808,
"grad_norm": 0.7300937175750732,
"learning_rate": 0.0002946277020021131,
"loss": 1.9862096786499024,
"step": 6860
},
{
"epoch": 0.87936,
"grad_norm": 0.7412726283073425,
"learning_rate": 0.0002946116518138968,
"loss": 1.99698486328125,
"step": 6870
},
{
"epoch": 0.88064,
"grad_norm": 0.71746426820755,
"learning_rate": 0.0002945955781242838,
"loss": 2.0053390502929687,
"step": 6880
},
{
"epoch": 0.88192,
"grad_norm": 0.6349291801452637,
"learning_rate": 0.00029457948093588617,
"loss": 1.999542236328125,
"step": 6890
},
{
"epoch": 0.8832,
"grad_norm": 0.654897153377533,
"learning_rate": 0.0002945633602513201,
"loss": 1.995721435546875,
"step": 6900
},
{
"epoch": 0.88448,
"grad_norm": 0.7566924691200256,
"learning_rate": 0.0002945472160732053,
"loss": 1.9950716018676757,
"step": 6910
},
{
"epoch": 0.88576,
"grad_norm": 0.7700677514076233,
"learning_rate": 0.00029453104840416555,
"loss": 1.9884449005126954,
"step": 6920
},
{
"epoch": 0.88704,
"grad_norm": 0.6426023244857788,
"learning_rate": 0.0002945148572468282,
"loss": 1.9813749313354492,
"step": 6930
},
{
"epoch": 0.88832,
"grad_norm": 0.6777772307395935,
"learning_rate": 0.00029449864260382456,
"loss": 1.9967033386230468,
"step": 6940
},
{
"epoch": 0.8896,
"grad_norm": 0.6357970237731934,
"learning_rate": 0.0002944824044777898,
"loss": 1.9828590393066405,
"step": 6950
},
{
"epoch": 0.89088,
"grad_norm": 0.7022501230239868,
"learning_rate": 0.00029446614287136275,
"loss": 1.9888254165649415,
"step": 6960
},
{
"epoch": 0.89216,
"grad_norm": 0.7219241857528687,
"learning_rate": 0.0002944498577871862,
"loss": 2.0122438430786134,
"step": 6970
},
{
"epoch": 0.89344,
"grad_norm": 0.9110201597213745,
"learning_rate": 0.00029443354922790673,
"loss": 2.006223678588867,
"step": 6980
},
{
"epoch": 0.89472,
"grad_norm": 0.6884856224060059,
"learning_rate": 0.00029441721719617467,
"loss": 1.995973777770996,
"step": 6990
},
{
"epoch": 0.896,
"grad_norm": 0.645409882068634,
"learning_rate": 0.0002944008616946442,
"loss": 2.0004106521606446,
"step": 7000
},
{
"epoch": 0.89728,
"grad_norm": 0.711455762386322,
"learning_rate": 0.00029438448272597325,
"loss": 1.9932981491088868,
"step": 7010
},
{
"epoch": 0.89856,
"grad_norm": 0.7177769541740417,
"learning_rate": 0.00029436808029282375,
"loss": 1.997603988647461,
"step": 7020
},
{
"epoch": 0.89984,
"grad_norm": 0.6674689054489136,
"learning_rate": 0.0002943516543978613,
"loss": 1.993707275390625,
"step": 7030
},
{
"epoch": 0.90112,
"grad_norm": 0.6319096684455872,
"learning_rate": 0.00029433520504375526,
"loss": 1.9995529174804687,
"step": 7040
},
{
"epoch": 0.9024,
"grad_norm": 0.709757387638092,
"learning_rate": 0.00029431873223317894,
"loss": 2.0230655670166016,
"step": 7050
},
{
"epoch": 0.90368,
"grad_norm": 0.6909634470939636,
"learning_rate": 0.00029430223596880935,
"loss": 1.9975494384765624,
"step": 7060
},
{
"epoch": 0.90496,
"grad_norm": 0.7052473425865173,
"learning_rate": 0.00029428571625332746,
"loss": 1.9993263244628907,
"step": 7070
},
{
"epoch": 0.90624,
"grad_norm": 0.7152456641197205,
"learning_rate": 0.00029426917308941785,
"loss": 1.9970455169677734,
"step": 7080
},
{
"epoch": 0.90752,
"grad_norm": 0.6790175437927246,
"learning_rate": 0.0002942526064797691,
"loss": 2.0116363525390626,
"step": 7090
},
{
"epoch": 0.9088,
"grad_norm": 0.7606917023658752,
"learning_rate": 0.0002942360164270734,
"loss": 1.9900932312011719,
"step": 7100
},
{
"epoch": 0.91008,
"grad_norm": 0.7293239235877991,
"learning_rate": 0.00029421940293402693,
"loss": 2.0022926330566406,
"step": 7110
},
{
"epoch": 0.91136,
"grad_norm": 0.6462413668632507,
"learning_rate": 0.00029420276600332964,
"loss": 1.998992919921875,
"step": 7120
},
{
"epoch": 0.91264,
"grad_norm": 0.686337947845459,
"learning_rate": 0.0002941861056376853,
"loss": 1.9897741317749023,
"step": 7130
},
{
"epoch": 0.91392,
"grad_norm": 0.6641421914100647,
"learning_rate": 0.00029416942183980126,
"loss": 1.9985929489135743,
"step": 7140
},
{
"epoch": 0.9152,
"grad_norm": 0.7737864851951599,
"learning_rate": 0.000294152714612389,
"loss": 1.9727941513061524,
"step": 7150
},
{
"epoch": 0.91648,
"grad_norm": 0.64877849817276,
"learning_rate": 0.00029413598395816376,
"loss": 2.022666358947754,
"step": 7160
},
{
"epoch": 0.91776,
"grad_norm": 0.6656960248947144,
"learning_rate": 0.0002941192298798443,
"loss": 2.000381851196289,
"step": 7170
},
{
"epoch": 0.91904,
"grad_norm": 0.7156676650047302,
"learning_rate": 0.0002941024523801536,
"loss": 2.003289222717285,
"step": 7180
},
{
"epoch": 0.92032,
"grad_norm": 0.8479704260826111,
"learning_rate": 0.00029408565146181806,
"loss": 1.9910587310791015,
"step": 7190
},
{
"epoch": 0.9216,
"grad_norm": 0.6156359314918518,
"learning_rate": 0.0002940688271275682,
"loss": 1.978067970275879,
"step": 7200
},
{
"epoch": 0.92288,
"grad_norm": 0.6154444813728333,
"learning_rate": 0.0002940519793801381,
"loss": 1.9935375213623048,
"step": 7210
},
{
"epoch": 0.92416,
"grad_norm": 0.6085964441299438,
"learning_rate": 0.0002940351082222658,
"loss": 1.9991844177246094,
"step": 7220
},
{
"epoch": 0.92544,
"grad_norm": 0.5994155406951904,
"learning_rate": 0.0002940182136566931,
"loss": 1.9809038162231445,
"step": 7230
},
{
"epoch": 0.92672,
"grad_norm": 0.6245939135551453,
"learning_rate": 0.0002940012956861656,
"loss": 2.0163652420043947,
"step": 7240
},
{
"epoch": 0.928,
"grad_norm": 0.6001088619232178,
"learning_rate": 0.0002939843543134327,
"loss": 1.9949073791503906,
"step": 7250
},
{
"epoch": 0.92928,
"grad_norm": 0.6890986561775208,
"learning_rate": 0.00029396738954124763,
"loss": 1.9710325241088866,
"step": 7260
},
{
"epoch": 0.93056,
"grad_norm": 0.6825648546218872,
"learning_rate": 0.0002939504013723673,
"loss": 1.9972740173339845,
"step": 7270
},
{
"epoch": 0.93184,
"grad_norm": 0.6806229948997498,
"learning_rate": 0.0002939333898095527,
"loss": 2.008091354370117,
"step": 7280
},
{
"epoch": 0.93312,
"grad_norm": 0.6244903206825256,
"learning_rate": 0.0002939163548555683,
"loss": 2.005999755859375,
"step": 7290
},
{
"epoch": 0.9344,
"grad_norm": 0.6636385917663574,
"learning_rate": 0.0002938992965131825,
"loss": 1.9983814239501954,
"step": 7300
},
{
"epoch": 0.93568,
"grad_norm": 0.6596930623054504,
"learning_rate": 0.00029388221478516765,
"loss": 1.9942008972167968,
"step": 7310
},
{
"epoch": 0.93696,
"grad_norm": 0.6415631175041199,
"learning_rate": 0.00029386510967429973,
"loss": 1.996099090576172,
"step": 7320
},
{
"epoch": 0.93824,
"grad_norm": 0.616016685962677,
"learning_rate": 0.0002938479811833585,
"loss": 1.974690818786621,
"step": 7330
},
{
"epoch": 0.93952,
"grad_norm": 0.7595481872558594,
"learning_rate": 0.00029383082931512755,
"loss": 2.010562515258789,
"step": 7340
},
{
"epoch": 0.9408,
"grad_norm": 0.7548149228096008,
"learning_rate": 0.00029381365407239436,
"loss": 2.013906478881836,
"step": 7350
},
{
"epoch": 0.94208,
"grad_norm": 0.6517251133918762,
"learning_rate": 0.0002937964554579502,
"loss": 1.99677734375,
"step": 7360
},
{
"epoch": 0.94336,
"grad_norm": 0.6647694706916809,
"learning_rate": 0.00029377923347458996,
"loss": 1.9947694778442382,
"step": 7370
},
{
"epoch": 0.94464,
"grad_norm": 0.6167952418327332,
"learning_rate": 0.00029376198812511246,
"loss": 1.9989194869995117,
"step": 7380
},
{
"epoch": 0.94592,
"grad_norm": 0.6266070008277893,
"learning_rate": 0.00029374471941232043,
"loss": 2.007636070251465,
"step": 7390
},
{
"epoch": 0.9472,
"grad_norm": 0.675845742225647,
"learning_rate": 0.00029372742733902016,
"loss": 2.0036916732788086,
"step": 7400
},
{
"epoch": 0.94848,
"grad_norm": 0.674096405506134,
"learning_rate": 0.0002937101119080219,
"loss": 1.9969058990478517,
"step": 7410
},
{
"epoch": 0.94976,
"grad_norm": 0.7183719873428345,
"learning_rate": 0.0002936927731221397,
"loss": 1.9890649795532227,
"step": 7420
},
{
"epoch": 0.95104,
"grad_norm": 0.7150754928588867,
"learning_rate": 0.0002936754109841912,
"loss": 1.9729717254638672,
"step": 7430
},
{
"epoch": 0.95232,
"grad_norm": 0.7129974961280823,
"learning_rate": 0.0002936580254969981,
"loss": 1.9953811645507813,
"step": 7440
},
{
"epoch": 0.9536,
"grad_norm": 0.6096296906471252,
"learning_rate": 0.0002936406166633858,
"loss": 2.001888465881348,
"step": 7450
},
{
"epoch": 0.95488,
"grad_norm": 0.6857330203056335,
"learning_rate": 0.0002936231844861834,
"loss": 1.9927192687988282,
"step": 7460
},
{
"epoch": 0.95616,
"grad_norm": 0.6103941798210144,
"learning_rate": 0.0002936057289682239,
"loss": 1.9911823272705078,
"step": 7470
},
{
"epoch": 0.95744,
"grad_norm": 0.597206175327301,
"learning_rate": 0.0002935882501123441,
"loss": 1.988555335998535,
"step": 7480
},
{
"epoch": 0.95872,
"grad_norm": 0.7180148959159851,
"learning_rate": 0.0002935707479213845,
"loss": 1.978421974182129,
"step": 7490
},
{
"epoch": 0.96,
"grad_norm": 0.6061798930168152,
"learning_rate": 0.0002935532223981895,
"loss": 2.000789833068848,
"step": 7500
},
{
"epoch": 0.96128,
"grad_norm": 0.6952294707298279,
"learning_rate": 0.0002935356735456072,
"loss": 2.0071075439453123,
"step": 7510
},
{
"epoch": 0.96256,
"grad_norm": 0.6120730042457581,
"learning_rate": 0.0002935181013664895,
"loss": 2.0083288192749023,
"step": 7520
},
{
"epoch": 0.96384,
"grad_norm": 0.7803784608840942,
"learning_rate": 0.0002935005058636923,
"loss": 2.009001541137695,
"step": 7530
},
{
"epoch": 0.96512,
"grad_norm": 0.681745171546936,
"learning_rate": 0.0002934828870400749,
"loss": 1.9958309173583983,
"step": 7540
},
{
"epoch": 0.9664,
"grad_norm": 0.7455606460571289,
"learning_rate": 0.0002934652448985007,
"loss": 2.0131885528564455,
"step": 7550
},
{
"epoch": 0.96768,
"grad_norm": 0.7167068719863892,
"learning_rate": 0.0002934475794418368,
"loss": 1.995752716064453,
"step": 7560
},
{
"epoch": 0.96896,
"grad_norm": 0.6225274205207825,
"learning_rate": 0.000293429890672954,
"loss": 1.9854461669921875,
"step": 7570
},
{
"epoch": 0.97024,
"grad_norm": 0.6639164686203003,
"learning_rate": 0.00029341217859472706,
"loss": 1.978256607055664,
"step": 7580
},
{
"epoch": 0.97152,
"grad_norm": 0.6081394553184509,
"learning_rate": 0.0002933944432100344,
"loss": 1.9813619613647462,
"step": 7590
},
{
"epoch": 0.9728,
"grad_norm": 0.6417889595031738,
"learning_rate": 0.00029337668452175826,
"loss": 1.9765779495239257,
"step": 7600
},
{
"epoch": 0.97408,
"grad_norm": 0.6452786326408386,
"learning_rate": 0.00029335890253278475,
"loss": 1.9978923797607422,
"step": 7610
},
{
"epoch": 0.97536,
"grad_norm": 0.793510913848877,
"learning_rate": 0.0002933410972460036,
"loss": 1.9694734573364259,
"step": 7620
},
{
"epoch": 0.97664,
"grad_norm": 0.6846933364868164,
"learning_rate": 0.0002933232686643085,
"loss": 1.9795555114746093,
"step": 7630
},
{
"epoch": 0.97792,
"grad_norm": 0.6672417521476746,
"learning_rate": 0.00029330541679059666,
"loss": 2.008864974975586,
"step": 7640
},
{
"epoch": 0.9792,
"grad_norm": 0.6564120054244995,
"learning_rate": 0.0002932875416277694,
"loss": 1.9894664764404297,
"step": 7650
},
{
"epoch": 0.98048,
"grad_norm": 0.6869664788246155,
"learning_rate": 0.0002932696431787317,
"loss": 1.9923513412475586,
"step": 7660
},
{
"epoch": 0.98176,
"grad_norm": 0.601786732673645,
"learning_rate": 0.0002932517214463923,
"loss": 1.987978744506836,
"step": 7670
},
{
"epoch": 0.98304,
"grad_norm": 0.6519659757614136,
"learning_rate": 0.0002932337764336637,
"loss": 2.0119016647338865,
"step": 7680
},
{
"epoch": 0.98432,
"grad_norm": 0.7457683682441711,
"learning_rate": 0.00029321580814346213,
"loss": 2.004359817504883,
"step": 7690
},
{
"epoch": 0.9856,
"grad_norm": 0.6067475080490112,
"learning_rate": 0.00029319781657870784,
"loss": 1.9842674255371093,
"step": 7700
},
{
"epoch": 0.98688,
"grad_norm": 0.7065998315811157,
"learning_rate": 0.00029317980174232457,
"loss": 1.9875959396362304,
"step": 7710
},
{
"epoch": 0.98816,
"grad_norm": 0.6263285279273987,
"learning_rate": 0.0002931617636372401,
"loss": 2.007324981689453,
"step": 7720
},
{
"epoch": 0.98944,
"grad_norm": 0.6340311765670776,
"learning_rate": 0.00029314370226638577,
"loss": 1.9956018447875976,
"step": 7730
},
{
"epoch": 0.99072,
"grad_norm": 0.5806658267974854,
"learning_rate": 0.0002931256176326969,
"loss": 2.0026828765869142,
"step": 7740
},
{
"epoch": 0.992,
"grad_norm": 0.6932752132415771,
"learning_rate": 0.00029310750973911244,
"loss": 1.9967496871948243,
"step": 7750
},
{
"epoch": 0.99328,
"grad_norm": 0.5850876569747925,
"learning_rate": 0.00029308937858857516,
"loss": 1.9920431137084962,
"step": 7760
},
{
"epoch": 0.99456,
"grad_norm": 0.6746029853820801,
"learning_rate": 0.0002930712241840317,
"loss": 2.0010097503662108,
"step": 7770
},
{
"epoch": 0.99584,
"grad_norm": 0.6813861131668091,
"learning_rate": 0.00029305304652843224,
"loss": 1.9804662704467773,
"step": 7780
},
{
"epoch": 0.99712,
"grad_norm": 0.6215769052505493,
"learning_rate": 0.00029303484562473106,
"loss": 1.995717430114746,
"step": 7790
},
{
"epoch": 0.9984,
"grad_norm": 0.5907732248306274,
"learning_rate": 0.000293016621475886,
"loss": 2.0071685791015623,
"step": 7800
},
{
"epoch": 0.99968,
"grad_norm": 0.5828958749771118,
"learning_rate": 0.0002929983740848588,
"loss": 1.9976682662963867,
"step": 7810
},
{
"epoch": 1.0,
"eval_loss": 2.000473737716675,
"eval_runtime": 17.9727,
"eval_samples_per_second": 556.399,
"eval_steps_per_second": 69.55,
"step": 7813
},
{
"epoch": 1.000896,
"grad_norm": 0.7024282217025757,
"learning_rate": 0.00029298010345461475,
"loss": 1.9894508361816405,
"step": 7820
},
{
"epoch": 1.002176,
"grad_norm": 0.5675840973854065,
"learning_rate": 0.0002929618095881232,
"loss": 1.9923004150390624,
"step": 7830
},
{
"epoch": 1.003456,
"grad_norm": 0.6930027008056641,
"learning_rate": 0.0002929434924883572,
"loss": 1.9826398849487306,
"step": 7840
},
{
"epoch": 1.004736,
"grad_norm": 0.622359037399292,
"learning_rate": 0.00029292515215829343,
"loss": 1.9939605712890625,
"step": 7850
},
{
"epoch": 1.006016,
"grad_norm": 0.6805294752120972,
"learning_rate": 0.0002929067886009125,
"loss": 2.0002468109130858,
"step": 7860
},
{
"epoch": 1.007296,
"grad_norm": 0.6081849932670593,
"learning_rate": 0.00029288840181919873,
"loss": 1.9870527267456055,
"step": 7870
},
{
"epoch": 1.008576,
"grad_norm": 0.6336544752120972,
"learning_rate": 0.00029286999181614025,
"loss": 1.9954706192016602,
"step": 7880
},
{
"epoch": 1.009856,
"grad_norm": 0.5647273063659668,
"learning_rate": 0.00029285155859472885,
"loss": 1.9976259231567384,
"step": 7890
},
{
"epoch": 1.011136,
"grad_norm": 0.696118950843811,
"learning_rate": 0.00029283310215796036,
"loss": 1.9746929168701173,
"step": 7900
},
{
"epoch": 1.012416,
"grad_norm": 0.6657605171203613,
"learning_rate": 0.000292814622508834,
"loss": 2.003103256225586,
"step": 7910
},
{
"epoch": 1.013696,
"grad_norm": 0.6706362366676331,
"learning_rate": 0.00029279611965035316,
"loss": 1.9854976654052734,
"step": 7920
},
{
"epoch": 1.014976,
"grad_norm": 0.6378129124641418,
"learning_rate": 0.0002927775935855247,
"loss": 1.9934816360473633,
"step": 7930
},
{
"epoch": 1.016256,
"grad_norm": 0.7083684802055359,
"learning_rate": 0.0002927590443173594,
"loss": 1.9854513168334962,
"step": 7940
},
{
"epoch": 1.017536,
"grad_norm": 0.6245972514152527,
"learning_rate": 0.0002927404718488717,
"loss": 1.9725086212158203,
"step": 7950
},
{
"epoch": 1.018816,
"grad_norm": 0.8558565974235535,
"learning_rate": 0.00029272187618307994,
"loss": 2.0068031311035157,
"step": 7960
},
{
"epoch": 1.020096,
"grad_norm": 0.7024584412574768,
"learning_rate": 0.00029270325732300617,
"loss": 2.0012155532836915,
"step": 7970
},
{
"epoch": 1.021376,
"grad_norm": 0.7526901364326477,
"learning_rate": 0.00029268461527167625,
"loss": 1.9808631896972657,
"step": 7980
},
{
"epoch": 1.022656,
"grad_norm": 0.6876883506774902,
"learning_rate": 0.0002926659500321197,
"loss": 2.0097496032714846,
"step": 7990
},
{
"epoch": 1.023936,
"grad_norm": 0.6317782402038574,
"learning_rate": 0.00029264726160737,
"loss": 1.9856227874755858,
"step": 8000
},
{
"epoch": 1.025216,
"grad_norm": 0.6349713802337646,
"learning_rate": 0.0002926285500004641,
"loss": 1.9954814910888672,
"step": 8010
},
{
"epoch": 1.026496,
"grad_norm": 0.7267403602600098,
"learning_rate": 0.00029260981521444303,
"loss": 1.9805095672607422,
"step": 8020
},
{
"epoch": 1.027776,
"grad_norm": 0.6417277455329895,
"learning_rate": 0.0002925910572523514,
"loss": 1.9915693283081055,
"step": 8030
},
{
"epoch": 1.029056,
"grad_norm": 0.6444427967071533,
"learning_rate": 0.0002925722761172377,
"loss": 2.005472946166992,
"step": 8040
},
{
"epoch": 1.030336,
"grad_norm": 0.6285523772239685,
"learning_rate": 0.00029255347181215405,
"loss": 1.9932641983032227,
"step": 8050
},
{
"epoch": 1.031616,
"grad_norm": 0.7181846499443054,
"learning_rate": 0.0002925346443401564,
"loss": 1.9999160766601562,
"step": 8060
},
{
"epoch": 1.032896,
"grad_norm": 0.6741138696670532,
"learning_rate": 0.0002925157937043045,
"loss": 1.9825817108154298,
"step": 8070
},
{
"epoch": 1.034176,
"grad_norm": 0.6893488764762878,
"learning_rate": 0.0002924969199076619,
"loss": 1.991935920715332,
"step": 8080
},
{
"epoch": 1.035456,
"grad_norm": 0.6443804502487183,
"learning_rate": 0.0002924780229532958,
"loss": 1.990732765197754,
"step": 8090
},
{
"epoch": 1.036736,
"grad_norm": 0.6640390157699585,
"learning_rate": 0.0002924591028442772,
"loss": 1.987821388244629,
"step": 8100
},
{
"epoch": 1.038016,
"grad_norm": 0.661410927772522,
"learning_rate": 0.00029244015958368094,
"loss": 1.9827268600463868,
"step": 8110
},
{
"epoch": 1.039296,
"grad_norm": 0.6270850300788879,
"learning_rate": 0.0002924211931745855,
"loss": 2.002053070068359,
"step": 8120
},
{
"epoch": 1.040576,
"grad_norm": 0.6214422583580017,
"learning_rate": 0.0002924022036200732,
"loss": 2.00252742767334,
"step": 8130
},
{
"epoch": 1.041856,
"grad_norm": 0.729099452495575,
"learning_rate": 0.00029238319092323014,
"loss": 2.0230823516845704,
"step": 8140
},
{
"epoch": 1.043136,
"grad_norm": 0.7051316499710083,
"learning_rate": 0.00029236415508714607,
"loss": 1.9967733383178712,
"step": 8150
},
{
"epoch": 1.044416,
"grad_norm": 0.6399348974227905,
"learning_rate": 0.0002923450961149147,
"loss": 2.0062068939208983,
"step": 8160
},
{
"epoch": 1.045696,
"grad_norm": 0.6973909735679626,
"learning_rate": 0.0002923260140096332,
"loss": 1.9918655395507812,
"step": 8170
},
{
"epoch": 1.046976,
"grad_norm": 0.6704668998718262,
"learning_rate": 0.00029230690877440283,
"loss": 1.979511833190918,
"step": 8180
},
{
"epoch": 1.048256,
"grad_norm": 0.8503730297088623,
"learning_rate": 0.0002922877804123284,
"loss": 1.989525032043457,
"step": 8190
},
{
"epoch": 1.049536,
"grad_norm": 0.6308216452598572,
"learning_rate": 0.0002922686289265186,
"loss": 1.9986820220947266,
"step": 8200
},
{
"epoch": 1.050816,
"grad_norm": 0.6530852317810059,
"learning_rate": 0.0002922494543200857,
"loss": 1.9965543746948242,
"step": 8210
},
{
"epoch": 1.052096,
"grad_norm": 0.6364430785179138,
"learning_rate": 0.0002922302565961459,
"loss": 1.9840011596679688,
"step": 8220
},
{
"epoch": 1.053376,
"grad_norm": 0.7692019939422607,
"learning_rate": 0.0002922110357578191,
"loss": 1.9859504699707031,
"step": 8230
},
{
"epoch": 1.054656,
"grad_norm": 0.7825514674186707,
"learning_rate": 0.00029219179180822887,
"loss": 1.9674690246582032,
"step": 8240
},
{
"epoch": 1.055936,
"grad_norm": 0.6661249995231628,
"learning_rate": 0.00029217252475050276,
"loss": 1.9999160766601562,
"step": 8250
},
{
"epoch": 1.057216,
"grad_norm": 0.7364896535873413,
"learning_rate": 0.0002921532345877718,
"loss": 2.0021907806396486,
"step": 8260
},
{
"epoch": 1.058496,
"grad_norm": 0.6803957223892212,
"learning_rate": 0.000292133921323171,
"loss": 1.9869365692138672,
"step": 8270
},
{
"epoch": 1.059776,
"grad_norm": 0.821689784526825,
"learning_rate": 0.00029211458495983903,
"loss": 1.976166343688965,
"step": 8280
},
{
"epoch": 1.061056,
"grad_norm": 0.7540163993835449,
"learning_rate": 0.00029209522550091826,
"loss": 2.0094572067260743,
"step": 8290
},
{
"epoch": 1.062336,
"grad_norm": 0.7056066989898682,
"learning_rate": 0.0002920758429495549,
"loss": 2.001365089416504,
"step": 8300
},
{
"epoch": 1.0636160000000001,
"grad_norm": 0.5745490789413452,
"learning_rate": 0.0002920564373088989,
"loss": 1.9914764404296874,
"step": 8310
},
{
"epoch": 1.064896,
"grad_norm": 0.6503936648368835,
"learning_rate": 0.0002920370085821039,
"loss": 2.0032411575317384,
"step": 8320
},
{
"epoch": 1.066176,
"grad_norm": 0.7628610730171204,
"learning_rate": 0.0002920175567723274,
"loss": 1.9816133499145507,
"step": 8330
},
{
"epoch": 1.067456,
"grad_norm": 0.6931906342506409,
"learning_rate": 0.0002919980818827306,
"loss": 2.023828125,
"step": 8340
},
{
"epoch": 1.068736,
"grad_norm": 0.6251834630966187,
"learning_rate": 0.0002919785839164783,
"loss": 2.0116418838500976,
"step": 8350
},
{
"epoch": 1.070016,
"grad_norm": 0.6859288215637207,
"learning_rate": 0.0002919590628767393,
"loss": 2.012284278869629,
"step": 8360
},
{
"epoch": 1.071296,
"grad_norm": 0.6031617522239685,
"learning_rate": 0.00029193951876668603,
"loss": 2.012980079650879,
"step": 8370
},
{
"epoch": 1.072576,
"grad_norm": 0.5843190550804138,
"learning_rate": 0.00029191995158949467,
"loss": 2.0026559829711914,
"step": 8380
},
{
"epoch": 1.073856,
"grad_norm": 0.7164151668548584,
"learning_rate": 0.0002919003613483452,
"loss": 1.9945117950439453,
"step": 8390
},
{
"epoch": 1.075136,
"grad_norm": 0.5945529341697693,
"learning_rate": 0.0002918807480464212,
"loss": 2.0126720428466798,
"step": 8400
},
{
"epoch": 1.076416,
"grad_norm": 0.6829405426979065,
"learning_rate": 0.0002918611116869102,
"loss": 1.9972929000854491,
"step": 8410
},
{
"epoch": 1.077696,
"grad_norm": 0.6316304206848145,
"learning_rate": 0.0002918414522730034,
"loss": 1.985988235473633,
"step": 8420
},
{
"epoch": 1.078976,
"grad_norm": 0.7297128438949585,
"learning_rate": 0.0002918217698078956,
"loss": 1.9967208862304688,
"step": 8430
},
{
"epoch": 1.0802559999999999,
"grad_norm": 0.7054782509803772,
"learning_rate": 0.00029180206429478556,
"loss": 1.988363265991211,
"step": 8440
},
{
"epoch": 1.081536,
"grad_norm": 0.6320076584815979,
"learning_rate": 0.00029178233573687567,
"loss": 1.9792478561401368,
"step": 8450
},
{
"epoch": 1.082816,
"grad_norm": 0.644890546798706,
"learning_rate": 0.00029176258413737214,
"loss": 2.0043672561645507,
"step": 8460
},
{
"epoch": 1.084096,
"grad_norm": 0.6674224734306335,
"learning_rate": 0.00029174280949948486,
"loss": 1.9823286056518554,
"step": 8470
},
{
"epoch": 1.085376,
"grad_norm": 0.6665927767753601,
"learning_rate": 0.0002917230118264275,
"loss": 1.9845500946044923,
"step": 8480
},
{
"epoch": 1.086656,
"grad_norm": 0.6852653622627258,
"learning_rate": 0.00029170319112141734,
"loss": 1.9879318237304688,
"step": 8490
},
{
"epoch": 1.087936,
"grad_norm": 0.6454859972000122,
"learning_rate": 0.0002916833473876757,
"loss": 1.9943349838256836,
"step": 8500
},
{
"epoch": 1.089216,
"grad_norm": 0.7632401585578918,
"learning_rate": 0.0002916634806284273,
"loss": 2.003750228881836,
"step": 8510
},
{
"epoch": 1.090496,
"grad_norm": 0.719601035118103,
"learning_rate": 0.0002916435908469009,
"loss": 1.976084327697754,
"step": 8520
},
{
"epoch": 1.091776,
"grad_norm": 0.7195095419883728,
"learning_rate": 0.0002916236780463287,
"loss": 1.9817279815673827,
"step": 8530
},
{
"epoch": 1.093056,
"grad_norm": 0.5842824578285217,
"learning_rate": 0.00029160374222994697,
"loss": 1.9941551208496093,
"step": 8540
},
{
"epoch": 1.094336,
"grad_norm": 0.6005828976631165,
"learning_rate": 0.0002915837834009955,
"loss": 1.9818029403686523,
"step": 8550
},
{
"epoch": 1.095616,
"grad_norm": 0.7141455411911011,
"learning_rate": 0.0002915638015627178,
"loss": 1.9895195007324218,
"step": 8560
},
{
"epoch": 1.096896,
"grad_norm": 0.627389132976532,
"learning_rate": 0.0002915437967183613,
"loss": 1.995197868347168,
"step": 8570
},
{
"epoch": 1.098176,
"grad_norm": 0.6275233626365662,
"learning_rate": 0.00029152376887117703,
"loss": 1.997283935546875,
"step": 8580
},
{
"epoch": 1.099456,
"grad_norm": 0.7087017297744751,
"learning_rate": 0.0002915037180244198,
"loss": 1.9955589294433593,
"step": 8590
},
{
"epoch": 1.100736,
"grad_norm": 0.6075368523597717,
"learning_rate": 0.0002914836441813481,
"loss": 2.01346435546875,
"step": 8600
},
{
"epoch": 1.1020159999999999,
"grad_norm": 0.6734739542007446,
"learning_rate": 0.00029146354734522426,
"loss": 2.009580612182617,
"step": 8610
},
{
"epoch": 1.103296,
"grad_norm": 0.7390580177307129,
"learning_rate": 0.00029144342751931434,
"loss": 1.99266357421875,
"step": 8620
},
{
"epoch": 1.104576,
"grad_norm": 0.6550483107566833,
"learning_rate": 0.000291423284706888,
"loss": 2.00741024017334,
"step": 8630
},
{
"epoch": 1.105856,
"grad_norm": 0.7504582405090332,
"learning_rate": 0.00029140311891121874,
"loss": 1.9887908935546874,
"step": 8640
},
{
"epoch": 1.107136,
"grad_norm": 0.5810806751251221,
"learning_rate": 0.0002913829301355838,
"loss": 1.9858449935913085,
"step": 8650
},
{
"epoch": 1.108416,
"grad_norm": 0.6569668054580688,
"learning_rate": 0.0002913627183832641,
"loss": 1.9724803924560548,
"step": 8660
},
{
"epoch": 1.109696,
"grad_norm": 0.6474798917770386,
"learning_rate": 0.0002913424836575444,
"loss": 1.9916406631469727,
"step": 8670
},
{
"epoch": 1.110976,
"grad_norm": 0.6266366243362427,
"learning_rate": 0.0002913222259617131,
"loss": 1.9882991790771485,
"step": 8680
},
{
"epoch": 1.112256,
"grad_norm": 0.596480667591095,
"learning_rate": 0.0002913019452990624,
"loss": 2.0080114364624024,
"step": 8690
},
{
"epoch": 1.113536,
"grad_norm": 0.6068896651268005,
"learning_rate": 0.000291281641672888,
"loss": 1.991042709350586,
"step": 8700
},
{
"epoch": 1.114816,
"grad_norm": 0.7131031155586243,
"learning_rate": 0.0002912613150864897,
"loss": 1.998366928100586,
"step": 8710
},
{
"epoch": 1.116096,
"grad_norm": 0.7099978923797607,
"learning_rate": 0.0002912409655431708,
"loss": 2.005122947692871,
"step": 8720
},
{
"epoch": 1.117376,
"grad_norm": 0.6002253293991089,
"learning_rate": 0.00029122059304623844,
"loss": 2.008624458312988,
"step": 8730
},
{
"epoch": 1.118656,
"grad_norm": 0.6592414379119873,
"learning_rate": 0.0002912001975990034,
"loss": 2.001871109008789,
"step": 8740
},
{
"epoch": 1.119936,
"grad_norm": 0.6449405550956726,
"learning_rate": 0.0002911797792047801,
"loss": 1.9964082717895508,
"step": 8750
},
{
"epoch": 1.121216,
"grad_norm": 0.6155027747154236,
"learning_rate": 0.00029115933786688696,
"loss": 2.0084716796875,
"step": 8760
},
{
"epoch": 1.122496,
"grad_norm": 0.6817499995231628,
"learning_rate": 0.00029113887358864595,
"loss": 1.9915023803710938,
"step": 8770
},
{
"epoch": 1.1237759999999999,
"grad_norm": 0.6094062924385071,
"learning_rate": 0.00029111838637338276,
"loss": 2.0121707916259766,
"step": 8780
},
{
"epoch": 1.125056,
"grad_norm": 0.6841493248939514,
"learning_rate": 0.00029109787622442687,
"loss": 1.9927202224731446,
"step": 8790
},
{
"epoch": 1.126336,
"grad_norm": 0.6776390671730042,
"learning_rate": 0.0002910773431451115,
"loss": 1.9925752639770509,
"step": 8800
},
{
"epoch": 1.127616,
"grad_norm": 0.6800190210342407,
"learning_rate": 0.00029105678713877347,
"loss": 1.9946041107177734,
"step": 8810
},
{
"epoch": 1.1288960000000001,
"grad_norm": 0.7016764283180237,
"learning_rate": 0.0002910362082087535,
"loss": 1.976521110534668,
"step": 8820
},
{
"epoch": 1.130176,
"grad_norm": 0.743949294090271,
"learning_rate": 0.00029101560635839594,
"loss": 1.9865568161010743,
"step": 8830
},
{
"epoch": 1.131456,
"grad_norm": 0.7083392143249512,
"learning_rate": 0.0002909949815910488,
"loss": 2.003726005554199,
"step": 8840
},
{
"epoch": 1.132736,
"grad_norm": 0.6515411138534546,
"learning_rate": 0.00029097433391006404,
"loss": 1.9914260864257813,
"step": 8850
},
{
"epoch": 1.134016,
"grad_norm": 0.5899383425712585,
"learning_rate": 0.000290953663318797,
"loss": 1.990021324157715,
"step": 8860
},
{
"epoch": 1.135296,
"grad_norm": 0.8539913892745972,
"learning_rate": 0.0002909329698206071,
"loss": 2.0082475662231447,
"step": 8870
},
{
"epoch": 1.136576,
"grad_norm": 0.6628640294075012,
"learning_rate": 0.00029091225341885724,
"loss": 2.0038652420043945,
"step": 8880
},
{
"epoch": 1.137856,
"grad_norm": 0.5988782644271851,
"learning_rate": 0.0002908915141169142,
"loss": 1.9990652084350586,
"step": 8890
},
{
"epoch": 1.139136,
"grad_norm": 0.6021246910095215,
"learning_rate": 0.0002908707519181483,
"loss": 2.0028762817382812,
"step": 8900
},
{
"epoch": 1.140416,
"grad_norm": 0.6528365612030029,
"learning_rate": 0.00029084996682593364,
"loss": 2.0019399642944338,
"step": 8910
},
{
"epoch": 1.141696,
"grad_norm": 0.7681286334991455,
"learning_rate": 0.0002908291588436483,
"loss": 2.0067686080932616,
"step": 8920
},
{
"epoch": 1.142976,
"grad_norm": 0.6195328235626221,
"learning_rate": 0.00029080832797467366,
"loss": 2.0036157608032226,
"step": 8930
},
{
"epoch": 1.144256,
"grad_norm": 0.623600423336029,
"learning_rate": 0.0002907874742223951,
"loss": 1.9990995407104493,
"step": 8940
},
{
"epoch": 1.1455359999999999,
"grad_norm": 0.5216490030288696,
"learning_rate": 0.00029076659759020174,
"loss": 1.9925945281982422,
"step": 8950
},
{
"epoch": 1.146816,
"grad_norm": 0.6754294633865356,
"learning_rate": 0.0002907456980814862,
"loss": 1.9840730667114257,
"step": 8960
},
{
"epoch": 1.148096,
"grad_norm": 0.6998392939567566,
"learning_rate": 0.00029072477569964494,
"loss": 1.9936521530151368,
"step": 8970
},
{
"epoch": 1.149376,
"grad_norm": 0.7083913087844849,
"learning_rate": 0.00029070383044807815,
"loss": 1.9848180770874024,
"step": 8980
},
{
"epoch": 1.1506560000000001,
"grad_norm": 0.6396852135658264,
"learning_rate": 0.0002906828623301898,
"loss": 1.9986888885498046,
"step": 8990
},
{
"epoch": 1.151936,
"grad_norm": 0.6948009133338928,
"learning_rate": 0.0002906618713493874,
"loss": 1.9894752502441406,
"step": 9000
},
{
"epoch": 1.153216,
"grad_norm": 0.6531941890716553,
"learning_rate": 0.0002906408575090823,
"loss": 2.0062957763671876,
"step": 9010
},
{
"epoch": 1.154496,
"grad_norm": 0.5760480165481567,
"learning_rate": 0.00029061982081268956,
"loss": 1.9995208740234376,
"step": 9020
},
{
"epoch": 1.155776,
"grad_norm": 0.626785159111023,
"learning_rate": 0.00029059876126362797,
"loss": 1.9724435806274414,
"step": 9030
},
{
"epoch": 1.157056,
"grad_norm": 0.6878816485404968,
"learning_rate": 0.00029057767886531995,
"loss": 2.0013551712036133,
"step": 9040
},
{
"epoch": 1.158336,
"grad_norm": 0.6750949621200562,
"learning_rate": 0.00029055657362119164,
"loss": 1.9839204788208007,
"step": 9050
},
{
"epoch": 1.159616,
"grad_norm": 0.6423714756965637,
"learning_rate": 0.00029053544553467303,
"loss": 1.99649658203125,
"step": 9060
},
{
"epoch": 1.160896,
"grad_norm": 0.6204116940498352,
"learning_rate": 0.0002905142946091977,
"loss": 1.9851703643798828,
"step": 9070
},
{
"epoch": 1.162176,
"grad_norm": 0.6263810396194458,
"learning_rate": 0.00029049312084820293,
"loss": 2.0322357177734376,
"step": 9080
},
{
"epoch": 1.163456,
"grad_norm": 0.7048192024230957,
"learning_rate": 0.0002904719242551297,
"loss": 2.0227006912231444,
"step": 9090
},
{
"epoch": 1.164736,
"grad_norm": 0.699733316898346,
"learning_rate": 0.0002904507048334229,
"loss": 1.9934453964233398,
"step": 9100
},
{
"epoch": 1.166016,
"grad_norm": 0.7169121503829956,
"learning_rate": 0.0002904294625865309,
"loss": 2.0101224899291994,
"step": 9110
},
{
"epoch": 1.167296,
"grad_norm": 0.6990874409675598,
"learning_rate": 0.00029040819751790587,
"loss": 1.9892080307006836,
"step": 9120
},
{
"epoch": 1.168576,
"grad_norm": 0.7468820214271545,
"learning_rate": 0.00029038690963100367,
"loss": 2.000370407104492,
"step": 9130
},
{
"epoch": 1.169856,
"grad_norm": 0.5932368636131287,
"learning_rate": 0.00029036559892928383,
"loss": 1.9932567596435546,
"step": 9140
},
{
"epoch": 1.171136,
"grad_norm": 0.6190387606620789,
"learning_rate": 0.0002903442654162097,
"loss": 1.986810302734375,
"step": 9150
},
{
"epoch": 1.1724160000000001,
"grad_norm": 0.6284942030906677,
"learning_rate": 0.0002903229090952482,
"loss": 2.0007484436035154,
"step": 9160
},
{
"epoch": 1.173696,
"grad_norm": 0.6459634900093079,
"learning_rate": 0.00029030152996987015,
"loss": 1.9838754653930664,
"step": 9170
},
{
"epoch": 1.174976,
"grad_norm": 0.6110875010490417,
"learning_rate": 0.00029028012804354985,
"loss": 1.9833248138427735,
"step": 9180
},
{
"epoch": 1.176256,
"grad_norm": 0.6577744483947754,
"learning_rate": 0.00029025870331976544,
"loss": 1.987434196472168,
"step": 9190
},
{
"epoch": 1.177536,
"grad_norm": 0.7174404859542847,
"learning_rate": 0.00029023725580199877,
"loss": 1.9872659683227538,
"step": 9200
},
{
"epoch": 1.178816,
"grad_norm": 0.6075583100318909,
"learning_rate": 0.0002902157854937353,
"loss": 1.996384048461914,
"step": 9210
},
{
"epoch": 1.180096,
"grad_norm": 0.6371153593063354,
"learning_rate": 0.00029019429239846424,
"loss": 2.0035045623779295,
"step": 9220
},
{
"epoch": 1.181376,
"grad_norm": 0.6514714360237122,
"learning_rate": 0.0002901727765196786,
"loss": 1.9920272827148438,
"step": 9230
},
{
"epoch": 1.182656,
"grad_norm": 0.7493175268173218,
"learning_rate": 0.0002901512378608749,
"loss": 1.9803998947143555,
"step": 9240
},
{
"epoch": 1.183936,
"grad_norm": 0.5912330746650696,
"learning_rate": 0.0002901296764255536,
"loss": 1.9990440368652345,
"step": 9250
},
{
"epoch": 1.185216,
"grad_norm": 0.7372444868087769,
"learning_rate": 0.00029010809221721864,
"loss": 1.983697509765625,
"step": 9260
},
{
"epoch": 1.186496,
"grad_norm": 0.6901136040687561,
"learning_rate": 0.00029008648523937776,
"loss": 1.9996004104614258,
"step": 9270
},
{
"epoch": 1.187776,
"grad_norm": 0.7002010345458984,
"learning_rate": 0.00029006485549554244,
"loss": 1.9926166534423828,
"step": 9280
},
{
"epoch": 1.189056,
"grad_norm": 0.6769751310348511,
"learning_rate": 0.0002900432029892278,
"loss": 1.9752182006835937,
"step": 9290
},
{
"epoch": 1.190336,
"grad_norm": 0.770754337310791,
"learning_rate": 0.0002900215277239526,
"loss": 1.9828071594238281,
"step": 9300
},
{
"epoch": 1.191616,
"grad_norm": 0.6528109908103943,
"learning_rate": 0.0002899998297032395,
"loss": 2.0097742080688477,
"step": 9310
},
{
"epoch": 1.192896,
"grad_norm": 0.6705182194709778,
"learning_rate": 0.0002899781089306145,
"loss": 1.9903186798095702,
"step": 9320
},
{
"epoch": 1.194176,
"grad_norm": 0.5876357555389404,
"learning_rate": 0.0002899563654096078,
"loss": 1.9866683959960938,
"step": 9330
},
{
"epoch": 1.195456,
"grad_norm": 0.6873459815979004,
"learning_rate": 0.00028993459914375294,
"loss": 1.9888439178466797,
"step": 9340
},
{
"epoch": 1.196736,
"grad_norm": 0.6179608702659607,
"learning_rate": 0.0002899128101365871,
"loss": 1.9880025863647461,
"step": 9350
},
{
"epoch": 1.198016,
"grad_norm": 0.5779441595077515,
"learning_rate": 0.00028989099839165153,
"loss": 1.9844137191772462,
"step": 9360
},
{
"epoch": 1.199296,
"grad_norm": 0.5996869802474976,
"learning_rate": 0.0002898691639124907,
"loss": 1.9843496322631835,
"step": 9370
},
{
"epoch": 1.200576,
"grad_norm": 0.6474190950393677,
"learning_rate": 0.00028984730670265316,
"loss": 1.9760330200195313,
"step": 9380
},
{
"epoch": 1.201856,
"grad_norm": 0.6795530319213867,
"learning_rate": 0.00028982542676569094,
"loss": 1.9977497100830077,
"step": 9390
},
{
"epoch": 1.203136,
"grad_norm": 0.689729630947113,
"learning_rate": 0.0002898035241051599,
"loss": 1.9970844268798829,
"step": 9400
},
{
"epoch": 1.204416,
"grad_norm": 0.5615735650062561,
"learning_rate": 0.00028978159872461955,
"loss": 2.000540924072266,
"step": 9410
},
{
"epoch": 1.205696,
"grad_norm": 0.6494594812393188,
"learning_rate": 0.00028975965062763293,
"loss": 2.015325736999512,
"step": 9420
},
{
"epoch": 1.206976,
"grad_norm": 0.6551460027694702,
"learning_rate": 0.000289737679817767,
"loss": 1.9717287063598632,
"step": 9430
},
{
"epoch": 1.208256,
"grad_norm": 0.596483588218689,
"learning_rate": 0.00028971568629859235,
"loss": 1.9834039688110352,
"step": 9440
},
{
"epoch": 1.209536,
"grad_norm": 0.6951877474784851,
"learning_rate": 0.0002896936700736831,
"loss": 1.998183250427246,
"step": 9450
},
{
"epoch": 1.210816,
"grad_norm": 0.5358121991157532,
"learning_rate": 0.0002896716311466174,
"loss": 1.974911880493164,
"step": 9460
},
{
"epoch": 1.212096,
"grad_norm": 0.6566754579544067,
"learning_rate": 0.00028964956952097677,
"loss": 1.999298095703125,
"step": 9470
},
{
"epoch": 1.213376,
"grad_norm": 0.6111728549003601,
"learning_rate": 0.00028962748520034646,
"loss": 1.9865041732788087,
"step": 9480
},
{
"epoch": 1.214656,
"grad_norm": 0.592664897441864,
"learning_rate": 0.00028960537818831553,
"loss": 1.998788833618164,
"step": 9490
},
{
"epoch": 1.215936,
"grad_norm": 0.8113380074501038,
"learning_rate": 0.00028958324848847675,
"loss": 2.0102975845336912,
"step": 9500
},
{
"epoch": 1.217216,
"grad_norm": 0.6786605715751648,
"learning_rate": 0.0002895610961044264,
"loss": 1.9990224838256836,
"step": 9510
},
{
"epoch": 1.218496,
"grad_norm": 0.5709884166717529,
"learning_rate": 0.00028953892103976463,
"loss": 2.008403015136719,
"step": 9520
},
{
"epoch": 1.219776,
"grad_norm": 0.6305531859397888,
"learning_rate": 0.0002895167232980951,
"loss": 1.9691112518310547,
"step": 9530
},
{
"epoch": 1.221056,
"grad_norm": 0.6338509917259216,
"learning_rate": 0.00028949450288302537,
"loss": 1.9768163681030273,
"step": 9540
},
{
"epoch": 1.222336,
"grad_norm": 0.6110090613365173,
"learning_rate": 0.00028947225979816647,
"loss": 1.9901458740234375,
"step": 9550
},
{
"epoch": 1.223616,
"grad_norm": 0.7046611309051514,
"learning_rate": 0.0002894499940471333,
"loss": 2.0120988845825196,
"step": 9560
},
{
"epoch": 1.224896,
"grad_norm": 0.5922809839248657,
"learning_rate": 0.0002894277056335442,
"loss": 1.9899206161499023,
"step": 9570
},
{
"epoch": 1.226176,
"grad_norm": 0.6404845118522644,
"learning_rate": 0.0002894053945610215,
"loss": 1.9906450271606446,
"step": 9580
},
{
"epoch": 1.227456,
"grad_norm": 0.8395019769668579,
"learning_rate": 0.00028938306083319096,
"loss": 1.9806875228881835,
"step": 9590
},
{
"epoch": 1.228736,
"grad_norm": 0.6999068856239319,
"learning_rate": 0.0002893607044536822,
"loss": 2.0015453338623046,
"step": 9600
},
{
"epoch": 1.230016,
"grad_norm": 0.6708416938781738,
"learning_rate": 0.0002893383254261284,
"loss": 2.011116600036621,
"step": 9610
},
{
"epoch": 1.231296,
"grad_norm": 0.6128595471382141,
"learning_rate": 0.0002893159237541665,
"loss": 2.00012092590332,
"step": 9620
},
{
"epoch": 1.232576,
"grad_norm": 0.5683632493019104,
"learning_rate": 0.000289293499441437,
"loss": 1.9906721115112305,
"step": 9630
},
{
"epoch": 1.233856,
"grad_norm": 0.6938204765319824,
"learning_rate": 0.0002892710524915842,
"loss": 2.0040435791015625,
"step": 9640
},
{
"epoch": 1.235136,
"grad_norm": 0.653827965259552,
"learning_rate": 0.000289248582908256,
"loss": 1.9979835510253907,
"step": 9650
},
{
"epoch": 1.236416,
"grad_norm": 0.6616809368133545,
"learning_rate": 0.00028922609069510416,
"loss": 1.9993171691894531,
"step": 9660
},
{
"epoch": 1.237696,
"grad_norm": 0.8354722857475281,
"learning_rate": 0.0002892035758557838,
"loss": 2.000519371032715,
"step": 9670
},
{
"epoch": 1.238976,
"grad_norm": 0.742372453212738,
"learning_rate": 0.00028918103839395404,
"loss": 1.9554183959960938,
"step": 9680
},
{
"epoch": 1.240256,
"grad_norm": 0.6502337455749512,
"learning_rate": 0.0002891584783132774,
"loss": 1.9820390701293946,
"step": 9690
},
{
"epoch": 1.241536,
"grad_norm": 0.6170443892478943,
"learning_rate": 0.0002891358956174204,
"loss": 2.0008007049560548,
"step": 9700
},
{
"epoch": 1.242816,
"grad_norm": 0.7270113229751587,
"learning_rate": 0.00028911329031005277,
"loss": 1.9921909332275392,
"step": 9710
},
{
"epoch": 1.244096,
"grad_norm": 0.6173141598701477,
"learning_rate": 0.0002890906623948484,
"loss": 1.9868398666381837,
"step": 9720
},
{
"epoch": 1.245376,
"grad_norm": 0.5539853572845459,
"learning_rate": 0.0002890680118754845,
"loss": 1.9938194274902343,
"step": 9730
},
{
"epoch": 1.246656,
"grad_norm": 0.6551358103752136,
"learning_rate": 0.0002890453387556422,
"loss": 2.000285339355469,
"step": 9740
},
{
"epoch": 1.247936,
"grad_norm": 0.6417705416679382,
"learning_rate": 0.00028902264303900616,
"loss": 2.01000919342041,
"step": 9750
},
{
"epoch": 1.249216,
"grad_norm": 0.6623634099960327,
"learning_rate": 0.00028899992472926475,
"loss": 2.0037622451782227,
"step": 9760
},
{
"epoch": 1.250496,
"grad_norm": 0.7787668108940125,
"learning_rate": 0.00028897718383011004,
"loss": 2.0113265991210936,
"step": 9770
},
{
"epoch": 1.251776,
"grad_norm": 0.6396425366401672,
"learning_rate": 0.00028895442034523763,
"loss": 2.008376121520996,
"step": 9780
},
{
"epoch": 1.253056,
"grad_norm": 0.5482383370399475,
"learning_rate": 0.00028893163427834705,
"loss": 1.99007568359375,
"step": 9790
},
{
"epoch": 1.254336,
"grad_norm": 0.6803271174430847,
"learning_rate": 0.0002889088256331412,
"loss": 2.008478355407715,
"step": 9800
},
{
"epoch": 1.255616,
"grad_norm": 0.6334392428398132,
"learning_rate": 0.000288885994413327,
"loss": 1.9965988159179688,
"step": 9810
},
{
"epoch": 1.256896,
"grad_norm": 0.6643745303153992,
"learning_rate": 0.00028886314062261456,
"loss": 1.9996068954467774,
"step": 9820
},
{
"epoch": 1.258176,
"grad_norm": 0.6333228349685669,
"learning_rate": 0.0002888402642647182,
"loss": 1.9843921661376953,
"step": 9830
},
{
"epoch": 1.2594560000000001,
"grad_norm": 0.8311046957969666,
"learning_rate": 0.0002888173653433555,
"loss": 2.009435844421387,
"step": 9840
},
{
"epoch": 1.260736,
"grad_norm": 0.6435722708702087,
"learning_rate": 0.00028879444386224794,
"loss": 2.004991912841797,
"step": 9850
},
{
"epoch": 1.262016,
"grad_norm": 0.6507195234298706,
"learning_rate": 0.00028877149982512053,
"loss": 2.001881980895996,
"step": 9860
},
{
"epoch": 1.263296,
"grad_norm": 0.6838963031768799,
"learning_rate": 0.0002887485332357019,
"loss": 1.9808956146240235,
"step": 9870
},
{
"epoch": 1.264576,
"grad_norm": 0.6148912906646729,
"learning_rate": 0.00028872554409772465,
"loss": 2.0023317337036133,
"step": 9880
},
{
"epoch": 1.2658559999999999,
"grad_norm": 0.6489004492759705,
"learning_rate": 0.00028870253241492465,
"loss": 1.9983543395996093,
"step": 9890
},
{
"epoch": 1.267136,
"grad_norm": 0.562017560005188,
"learning_rate": 0.0002886794981910417,
"loss": 1.9948774337768556,
"step": 9900
},
{
"epoch": 1.268416,
"grad_norm": 0.7083085775375366,
"learning_rate": 0.0002886564414298192,
"loss": 1.9873723983764648,
"step": 9910
},
{
"epoch": 1.269696,
"grad_norm": 0.6206421852111816,
"learning_rate": 0.0002886333621350041,
"loss": 1.99036865234375,
"step": 9920
},
{
"epoch": 1.270976,
"grad_norm": 0.6814326643943787,
"learning_rate": 0.00028861026031034724,
"loss": 1.9877098083496094,
"step": 9930
},
{
"epoch": 1.272256,
"grad_norm": 0.6762963533401489,
"learning_rate": 0.0002885871359596029,
"loss": 1.9892280578613282,
"step": 9940
},
{
"epoch": 1.273536,
"grad_norm": 0.6174524426460266,
"learning_rate": 0.0002885639890865291,
"loss": 2.004766082763672,
"step": 9950
},
{
"epoch": 1.274816,
"grad_norm": 0.8179512023925781,
"learning_rate": 0.0002885408196948875,
"loss": 1.9910087585449219,
"step": 9960
},
{
"epoch": 1.276096,
"grad_norm": 0.6321988105773926,
"learning_rate": 0.00028851762778844353,
"loss": 1.9934940338134766,
"step": 9970
},
{
"epoch": 1.277376,
"grad_norm": 0.6512481570243835,
"learning_rate": 0.0002884944133709661,
"loss": 1.993656539916992,
"step": 9980
},
{
"epoch": 1.278656,
"grad_norm": 0.5633252263069153,
"learning_rate": 0.000288471176446228,
"loss": 1.9791446685791017,
"step": 9990
},
{
"epoch": 1.279936,
"grad_norm": 0.6259685754776001,
"learning_rate": 0.0002884479170180055,
"loss": 1.9953647613525392,
"step": 10000
},
{
"epoch": 1.2812160000000001,
"grad_norm": 0.6940027475357056,
"learning_rate": 0.0002884246350900785,
"loss": 1.9884658813476563,
"step": 10010
},
{
"epoch": 1.282496,
"grad_norm": 0.5981363654136658,
"learning_rate": 0.00028840133066623076,
"loss": 1.9940732955932616,
"step": 10020
},
{
"epoch": 1.283776,
"grad_norm": 0.6378276348114014,
"learning_rate": 0.0002883780037502495,
"loss": 2.0147527694702148,
"step": 10030
},
{
"epoch": 1.285056,
"grad_norm": 0.5869522094726562,
"learning_rate": 0.00028835465434592564,
"loss": 2.0134244918823243,
"step": 10040
},
{
"epoch": 1.286336,
"grad_norm": 0.5895775556564331,
"learning_rate": 0.00028833128245705385,
"loss": 1.9956907272338866,
"step": 10050
},
{
"epoch": 1.2876159999999999,
"grad_norm": 0.6918666362762451,
"learning_rate": 0.0002883078880874324,
"loss": 1.994941520690918,
"step": 10060
},
{
"epoch": 1.288896,
"grad_norm": 0.5689588785171509,
"learning_rate": 0.00028828447124086304,
"loss": 2.004123497009277,
"step": 10070
},
{
"epoch": 1.290176,
"grad_norm": 0.6616405248641968,
"learning_rate": 0.00028826103192115154,
"loss": 1.989501953125,
"step": 10080
},
{
"epoch": 1.291456,
"grad_norm": 0.7383025884628296,
"learning_rate": 0.000288237570132107,
"loss": 1.9705989837646485,
"step": 10090
},
{
"epoch": 1.292736,
"grad_norm": 0.6901052594184875,
"learning_rate": 0.0002882140858775423,
"loss": 1.9954898834228516,
"step": 10100
},
{
"epoch": 1.294016,
"grad_norm": 0.6705789566040039,
"learning_rate": 0.00028819057916127397,
"loss": 1.9858028411865234,
"step": 10110
},
{
"epoch": 1.295296,
"grad_norm": 0.6146973967552185,
"learning_rate": 0.00028816704998712217,
"loss": 2.0052791595458985,
"step": 10120
},
{
"epoch": 1.296576,
"grad_norm": 0.7097535729408264,
"learning_rate": 0.0002881434983589107,
"loss": 1.979853057861328,
"step": 10130
},
{
"epoch": 1.297856,
"grad_norm": 0.6322594285011292,
"learning_rate": 0.0002881199242804671,
"loss": 2.0004344940185548,
"step": 10140
},
{
"epoch": 1.299136,
"grad_norm": 0.6610602140426636,
"learning_rate": 0.0002880963277556224,
"loss": 1.9636198043823243,
"step": 10150
},
{
"epoch": 1.300416,
"grad_norm": 0.6865135431289673,
"learning_rate": 0.00028807270878821134,
"loss": 2.0157705307006837,
"step": 10160
},
{
"epoch": 1.301696,
"grad_norm": 0.6560506224632263,
"learning_rate": 0.00028804906738207244,
"loss": 1.99947509765625,
"step": 10170
},
{
"epoch": 1.3029760000000001,
"grad_norm": 0.6947329640388489,
"learning_rate": 0.00028802540354104766,
"loss": 1.9987869262695312,
"step": 10180
},
{
"epoch": 1.304256,
"grad_norm": 0.7200315594673157,
"learning_rate": 0.0002880017172689828,
"loss": 2.00171012878418,
"step": 10190
},
{
"epoch": 1.305536,
"grad_norm": 0.6494705080986023,
"learning_rate": 0.0002879780085697271,
"loss": 1.983466911315918,
"step": 10200
},
{
"epoch": 1.306816,
"grad_norm": 0.6532952785491943,
"learning_rate": 0.00028795427744713356,
"loss": 1.9807058334350587,
"step": 10210
},
{
"epoch": 1.308096,
"grad_norm": 0.6909893155097961,
"learning_rate": 0.0002879305239050589,
"loss": 2.0161617279052733,
"step": 10220
},
{
"epoch": 1.3093759999999999,
"grad_norm": 0.6190416812896729,
"learning_rate": 0.0002879067479473633,
"loss": 2.024698829650879,
"step": 10230
},
{
"epoch": 1.310656,
"grad_norm": 0.618144154548645,
"learning_rate": 0.0002878829495779108,
"loss": 1.9919988632202148,
"step": 10240
},
{
"epoch": 1.311936,
"grad_norm": 0.7293611168861389,
"learning_rate": 0.0002878591288005689,
"loss": 2.0231300354003907,
"step": 10250
},
{
"epoch": 1.313216,
"grad_norm": 0.5665643215179443,
"learning_rate": 0.00028783528561920874,
"loss": 1.9832281112670898,
"step": 10260
},
{
"epoch": 1.314496,
"grad_norm": 0.6041682958602905,
"learning_rate": 0.0002878114200377053,
"loss": 2.0007213592529296,
"step": 10270
},
{
"epoch": 1.315776,
"grad_norm": 0.676340639591217,
"learning_rate": 0.000287787532059937,
"loss": 1.9854997634887694,
"step": 10280
},
{
"epoch": 1.317056,
"grad_norm": 0.6602581739425659,
"learning_rate": 0.00028776362168978593,
"loss": 2.013218116760254,
"step": 10290
},
{
"epoch": 1.318336,
"grad_norm": 0.6657641530036926,
"learning_rate": 0.0002877396889311379,
"loss": 2.0098697662353517,
"step": 10300
},
{
"epoch": 1.319616,
"grad_norm": 0.6588369607925415,
"learning_rate": 0.0002877157337878823,
"loss": 1.9871788024902344,
"step": 10310
},
{
"epoch": 1.320896,
"grad_norm": 0.587067723274231,
"learning_rate": 0.00028769175626391225,
"loss": 2.006184387207031,
"step": 10320
},
{
"epoch": 1.322176,
"grad_norm": 0.6068325638771057,
"learning_rate": 0.00028766775636312434,
"loss": 2.021990203857422,
"step": 10330
},
{
"epoch": 1.323456,
"grad_norm": 0.6358674168586731,
"learning_rate": 0.00028764373408941883,
"loss": 1.9817586898803712,
"step": 10340
},
{
"epoch": 1.3247360000000001,
"grad_norm": 0.6737171411514282,
"learning_rate": 0.0002876196894466998,
"loss": 1.9924598693847657,
"step": 10350
},
{
"epoch": 1.326016,
"grad_norm": 0.6510801315307617,
"learning_rate": 0.00028759562243887483,
"loss": 1.9732032775878907,
"step": 10360
},
{
"epoch": 1.327296,
"grad_norm": 0.6623960733413696,
"learning_rate": 0.00028757153306985505,
"loss": 2.0112014770507813,
"step": 10370
},
{
"epoch": 1.328576,
"grad_norm": 0.7083103656768799,
"learning_rate": 0.0002875474213435554,
"loss": 1.9748468399047852,
"step": 10380
},
{
"epoch": 1.329856,
"grad_norm": 0.6988223195075989,
"learning_rate": 0.0002875232872638943,
"loss": 2.0001197814941407,
"step": 10390
},
{
"epoch": 1.3311359999999999,
"grad_norm": 0.621229887008667,
"learning_rate": 0.0002874991308347939,
"loss": 1.9709728240966797,
"step": 10400
},
{
"epoch": 1.332416,
"grad_norm": 0.6560629606246948,
"learning_rate": 0.00028747495206018,
"loss": 1.9833150863647462,
"step": 10410
},
{
"epoch": 1.333696,
"grad_norm": 0.6206739544868469,
"learning_rate": 0.00028745075094398193,
"loss": 1.9948457717895507,
"step": 10420
},
{
"epoch": 1.334976,
"grad_norm": 0.6081636548042297,
"learning_rate": 0.0002874265274901327,
"loss": 1.9810810089111328,
"step": 10430
},
{
"epoch": 1.336256,
"grad_norm": 0.7739059329032898,
"learning_rate": 0.00028740228170256906,
"loss": 2.004429244995117,
"step": 10440
},
{
"epoch": 1.337536,
"grad_norm": 0.6709536910057068,
"learning_rate": 0.00028737801358523116,
"loss": 1.9931043624877929,
"step": 10450
},
{
"epoch": 1.338816,
"grad_norm": 0.6346166133880615,
"learning_rate": 0.00028735372314206295,
"loss": 1.9890787124633789,
"step": 10460
},
{
"epoch": 1.340096,
"grad_norm": 0.5924286842346191,
"learning_rate": 0.00028732941037701195,
"loss": 1.9799274444580077,
"step": 10470
},
{
"epoch": 1.341376,
"grad_norm": 0.6107586026191711,
"learning_rate": 0.0002873050752940293,
"loss": 1.9958015441894532,
"step": 10480
},
{
"epoch": 1.342656,
"grad_norm": 0.7260807156562805,
"learning_rate": 0.00028728071789706997,
"loss": 1.9838800430297852,
"step": 10490
},
{
"epoch": 1.343936,
"grad_norm": 0.6761348843574524,
"learning_rate": 0.00028725633819009214,
"loss": 1.998250389099121,
"step": 10500
},
{
"epoch": 1.345216,
"grad_norm": 0.6748042702674866,
"learning_rate": 0.00028723193617705796,
"loss": 1.9905017852783202,
"step": 10510
},
{
"epoch": 1.3464960000000001,
"grad_norm": 0.6004123091697693,
"learning_rate": 0.0002872075118619331,
"loss": 2.0023746490478516,
"step": 10520
},
{
"epoch": 1.347776,
"grad_norm": 0.7450584173202515,
"learning_rate": 0.0002871830652486867,
"loss": 2.0077308654785155,
"step": 10530
},
{
"epoch": 1.349056,
"grad_norm": 0.6224455833435059,
"learning_rate": 0.0002871585963412919,
"loss": 1.985758399963379,
"step": 10540
},
{
"epoch": 1.350336,
"grad_norm": 0.705573558807373,
"learning_rate": 0.00028713410514372514,
"loss": 1.9857095718383788,
"step": 10550
},
{
"epoch": 1.351616,
"grad_norm": 0.6310495734214783,
"learning_rate": 0.0002871095916599666,
"loss": 1.9932771682739259,
"step": 10560
},
{
"epoch": 1.3528959999999999,
"grad_norm": 0.6771069765090942,
"learning_rate": 0.000287085055894,
"loss": 1.979155921936035,
"step": 10570
},
{
"epoch": 1.354176,
"grad_norm": 0.7155138850212097,
"learning_rate": 0.00028706049784981276,
"loss": 1.9811246871948243,
"step": 10580
},
{
"epoch": 1.355456,
"grad_norm": 0.6692302227020264,
"learning_rate": 0.0002870359175313959,
"loss": 1.9715784072875977,
"step": 10590
},
{
"epoch": 1.356736,
"grad_norm": 0.698147177696228,
"learning_rate": 0.00028701131494274413,
"loss": 1.9810985565185546,
"step": 10600
},
{
"epoch": 1.3580160000000001,
"grad_norm": 0.6109893321990967,
"learning_rate": 0.00028698669008785565,
"loss": 1.9850080490112305,
"step": 10610
},
{
"epoch": 1.359296,
"grad_norm": 0.6750178337097168,
"learning_rate": 0.0002869620429707323,
"loss": 1.9807966232299805,
"step": 10620
},
{
"epoch": 1.360576,
"grad_norm": 0.6399105191230774,
"learning_rate": 0.00028693737359537975,
"loss": 1.996908950805664,
"step": 10630
},
{
"epoch": 1.361856,
"grad_norm": 0.7156611680984497,
"learning_rate": 0.0002869126819658069,
"loss": 2.015245819091797,
"step": 10640
},
{
"epoch": 1.363136,
"grad_norm": 0.6653876304626465,
"learning_rate": 0.0002868879680860266,
"loss": 1.982522964477539,
"step": 10650
},
{
"epoch": 1.364416,
"grad_norm": 0.5759459733963013,
"learning_rate": 0.00028686323196005516,
"loss": 1.9852264404296875,
"step": 10660
},
{
"epoch": 1.365696,
"grad_norm": 0.663753092288971,
"learning_rate": 0.0002868384735919126,
"loss": 1.990663719177246,
"step": 10670
},
{
"epoch": 1.366976,
"grad_norm": 0.6780126690864563,
"learning_rate": 0.0002868136929856224,
"loss": 1.9865737915039063,
"step": 10680
},
{
"epoch": 1.368256,
"grad_norm": 0.58465975522995,
"learning_rate": 0.0002867888901452119,
"loss": 1.9981576919555664,
"step": 10690
},
{
"epoch": 1.369536,
"grad_norm": 0.588762104511261,
"learning_rate": 0.00028676406507471174,
"loss": 2.019969940185547,
"step": 10700
},
{
"epoch": 1.370816,
"grad_norm": 0.6932179927825928,
"learning_rate": 0.0002867392177781564,
"loss": 2.012518310546875,
"step": 10710
},
{
"epoch": 1.372096,
"grad_norm": 0.7421392202377319,
"learning_rate": 0.00028671434825958396,
"loss": 2.0192115783691404,
"step": 10720
},
{
"epoch": 1.373376,
"grad_norm": 0.6568295955657959,
"learning_rate": 0.00028668945652303603,
"loss": 2.0088897705078126,
"step": 10730
},
{
"epoch": 1.3746559999999999,
"grad_norm": 0.5931066274642944,
"learning_rate": 0.0002866645425725578,
"loss": 1.990027618408203,
"step": 10740
},
{
"epoch": 1.375936,
"grad_norm": 0.6859118938446045,
"learning_rate": 0.0002866396064121983,
"loss": 1.9899696350097655,
"step": 10750
},
{
"epoch": 1.377216,
"grad_norm": 0.568245530128479,
"learning_rate": 0.0002866146480460098,
"loss": 1.991337776184082,
"step": 10760
},
{
"epoch": 1.378496,
"grad_norm": 0.5934443473815918,
"learning_rate": 0.0002865896674780485,
"loss": 1.9911518096923828,
"step": 10770
},
{
"epoch": 1.3797760000000001,
"grad_norm": 0.5719492435455322,
"learning_rate": 0.000286564664712374,
"loss": 2.007747840881348,
"step": 10780
},
{
"epoch": 1.381056,
"grad_norm": 1.5354629755020142,
"learning_rate": 0.0002865396397530497,
"loss": 2.0097402572631835,
"step": 10790
},
{
"epoch": 1.382336,
"grad_norm": 0.6930305361747742,
"learning_rate": 0.00028651459260414244,
"loss": 1.992599105834961,
"step": 10800
},
{
"epoch": 1.383616,
"grad_norm": 3.082179069519043,
"learning_rate": 0.00028648952326972276,
"loss": 1.9868263244628905,
"step": 10810
},
{
"epoch": 1.384896,
"grad_norm": 0.6655099987983704,
"learning_rate": 0.00028646443175386476,
"loss": 2.0117923736572267,
"step": 10820
},
{
"epoch": 1.386176,
"grad_norm": 0.6517053842544556,
"learning_rate": 0.00028643931806064617,
"loss": 1.9878664016723633,
"step": 10830
},
{
"epoch": 1.387456,
"grad_norm": 0.6760900020599365,
"learning_rate": 0.00028641418219414825,
"loss": 2.0264522552490236,
"step": 10840
},
{
"epoch": 1.388736,
"grad_norm": 0.6285338997840881,
"learning_rate": 0.000286389024158456,
"loss": 1.9868438720703125,
"step": 10850
},
{
"epoch": 1.390016,
"grad_norm": 0.7342556118965149,
"learning_rate": 0.0002863638439576579,
"loss": 1.976963424682617,
"step": 10860
},
{
"epoch": 1.391296,
"grad_norm": 0.6236872673034668,
"learning_rate": 0.0002863386415958462,
"loss": 1.9964305877685546,
"step": 10870
},
{
"epoch": 1.392576,
"grad_norm": 0.5662305355072021,
"learning_rate": 0.00028631341707711645,
"loss": 1.9916418075561524,
"step": 10880
},
{
"epoch": 1.393856,
"grad_norm": 0.7787207365036011,
"learning_rate": 0.0002862881704055681,
"loss": 2.005059814453125,
"step": 10890
},
{
"epoch": 1.395136,
"grad_norm": 0.6719783544540405,
"learning_rate": 0.00028626290158530404,
"loss": 2.0124752044677736,
"step": 10900
},
{
"epoch": 1.3964159999999999,
"grad_norm": 2.0237138271331787,
"learning_rate": 0.00028623761062043085,
"loss": 1.9972211837768554,
"step": 10910
},
{
"epoch": 1.397696,
"grad_norm": 0.7342656850814819,
"learning_rate": 0.00028621229751505864,
"loss": 1.9924606323242187,
"step": 10920
},
{
"epoch": 1.398976,
"grad_norm": 0.6957292556762695,
"learning_rate": 0.0002861869622733011,
"loss": 1.9881515502929688,
"step": 10930
},
{
"epoch": 1.400256,
"grad_norm": 0.6907182931900024,
"learning_rate": 0.00028616160489927563,
"loss": 1.999223518371582,
"step": 10940
},
{
"epoch": 1.4015360000000001,
"grad_norm": 0.7615217566490173,
"learning_rate": 0.00028613622539710315,
"loss": 2.0106784820556642,
"step": 10950
},
{
"epoch": 1.402816,
"grad_norm": 0.5999549627304077,
"learning_rate": 0.00028611082377090805,
"loss": 2.0066198348999023,
"step": 10960
},
{
"epoch": 1.404096,
"grad_norm": 0.6976864337921143,
"learning_rate": 0.0002860854000248186,
"loss": 1.969540786743164,
"step": 10970
},
{
"epoch": 1.405376,
"grad_norm": 0.7127543091773987,
"learning_rate": 0.00028605995416296643,
"loss": 2.0016845703125,
"step": 10980
},
{
"epoch": 1.406656,
"grad_norm": 0.678183913230896,
"learning_rate": 0.0002860344861894869,
"loss": 1.998734474182129,
"step": 10990
},
{
"epoch": 1.407936,
"grad_norm": 0.6985042095184326,
"learning_rate": 0.00028600899610851883,
"loss": 2.0049741744995115,
"step": 11000
},
{
"epoch": 1.409216,
"grad_norm": 0.6266785860061646,
"learning_rate": 0.00028598348392420476,
"loss": 2.0135120391845702,
"step": 11010
},
{
"epoch": 1.410496,
"grad_norm": 0.6285012364387512,
"learning_rate": 0.00028595794964069083,
"loss": 1.9838949203491212,
"step": 11020
},
{
"epoch": 1.411776,
"grad_norm": 0.6828831434249878,
"learning_rate": 0.0002859323932621266,
"loss": 2.0116504669189452,
"step": 11030
},
{
"epoch": 1.413056,
"grad_norm": 0.7426384091377258,
"learning_rate": 0.00028590681479266546,
"loss": 2.002168083190918,
"step": 11040
},
{
"epoch": 1.414336,
"grad_norm": 0.8292524814605713,
"learning_rate": 0.0002858812142364641,
"loss": 1.9958446502685547,
"step": 11050
},
{
"epoch": 1.415616,
"grad_norm": 0.5959202647209167,
"learning_rate": 0.0002858555915976831,
"loss": 2.0103025436401367,
"step": 11060
},
{
"epoch": 1.416896,
"grad_norm": 0.655495285987854,
"learning_rate": 0.00028582994688048654,
"loss": 2.0029670715332033,
"step": 11070
},
{
"epoch": 1.4181759999999999,
"grad_norm": 0.6991884708404541,
"learning_rate": 0.0002858042800890419,
"loss": 2.000274658203125,
"step": 11080
},
{
"epoch": 1.419456,
"grad_norm": 0.6688516139984131,
"learning_rate": 0.0002857785912275204,
"loss": 1.9912334442138673,
"step": 11090
},
{
"epoch": 1.420736,
"grad_norm": 0.6340951323509216,
"learning_rate": 0.00028575288030009693,
"loss": 1.9949312210083008,
"step": 11100
},
{
"epoch": 1.422016,
"grad_norm": 0.741901695728302,
"learning_rate": 0.00028572714731094984,
"loss": 1.997260284423828,
"step": 11110
},
{
"epoch": 1.4232960000000001,
"grad_norm": 0.6392485499382019,
"learning_rate": 0.0002857013922642611,
"loss": 1.991695022583008,
"step": 11120
},
{
"epoch": 1.424576,
"grad_norm": 0.6262326836585999,
"learning_rate": 0.00028567561516421616,
"loss": 1.9889442443847656,
"step": 11130
},
{
"epoch": 1.425856,
"grad_norm": 0.6146108508110046,
"learning_rate": 0.0002856498160150043,
"loss": 1.9918178558349608,
"step": 11140
},
{
"epoch": 1.427136,
"grad_norm": 0.6391763687133789,
"learning_rate": 0.00028562399482081816,
"loss": 1.9855514526367188,
"step": 11150
},
{
"epoch": 1.428416,
"grad_norm": 0.7085586190223694,
"learning_rate": 0.0002855981515858541,
"loss": 2.00579776763916,
"step": 11160
},
{
"epoch": 1.429696,
"grad_norm": 0.6206077933311462,
"learning_rate": 0.0002855722863143119,
"loss": 1.997480583190918,
"step": 11170
},
{
"epoch": 1.430976,
"grad_norm": 0.6236110329627991,
"learning_rate": 0.0002855463990103951,
"loss": 1.9995306015014649,
"step": 11180
},
{
"epoch": 1.432256,
"grad_norm": 0.6080915331840515,
"learning_rate": 0.0002855204896783108,
"loss": 2.0111207962036133,
"step": 11190
},
{
"epoch": 1.433536,
"grad_norm": 0.6906118988990784,
"learning_rate": 0.0002854945583222695,
"loss": 1.9987607955932618,
"step": 11200
},
{
"epoch": 1.434816,
"grad_norm": 0.6722120046615601,
"learning_rate": 0.0002854686049464855,
"loss": 1.9931190490722657,
"step": 11210
},
{
"epoch": 1.436096,
"grad_norm": 0.6693927645683289,
"learning_rate": 0.0002854426295551765,
"loss": 1.9924522399902345,
"step": 11220
},
{
"epoch": 1.437376,
"grad_norm": 0.6863765716552734,
"learning_rate": 0.000285416632152564,
"loss": 2.0015438079833983,
"step": 11230
},
{
"epoch": 1.438656,
"grad_norm": 0.6704138517379761,
"learning_rate": 0.0002853906127428728,
"loss": 1.9822942733764648,
"step": 11240
},
{
"epoch": 1.4399359999999999,
"grad_norm": 0.6855907440185547,
"learning_rate": 0.0002853645713303314,
"loss": 1.9638769149780273,
"step": 11250
},
{
"epoch": 1.441216,
"grad_norm": 0.7492919564247131,
"learning_rate": 0.00028533850791917205,
"loss": 1.9975135803222657,
"step": 11260
},
{
"epoch": 1.442496,
"grad_norm": 0.6287387609481812,
"learning_rate": 0.0002853124225136302,
"loss": 1.9622228622436524,
"step": 11270
},
{
"epoch": 1.443776,
"grad_norm": 0.64750075340271,
"learning_rate": 0.0002852863151179453,
"loss": 1.9830795288085938,
"step": 11280
},
{
"epoch": 1.4450560000000001,
"grad_norm": 0.6543371677398682,
"learning_rate": 0.00028526018573636007,
"loss": 1.9904781341552735,
"step": 11290
},
{
"epoch": 1.446336,
"grad_norm": 0.6812680959701538,
"learning_rate": 0.0002852340343731209,
"loss": 1.9876131057739257,
"step": 11300
},
{
"epoch": 1.447616,
"grad_norm": 0.7219147086143494,
"learning_rate": 0.0002852078610324777,
"loss": 1.9981233596801757,
"step": 11310
},
{
"epoch": 1.448896,
"grad_norm": 0.6542626023292542,
"learning_rate": 0.0002851816657186841,
"loss": 2.0076507568359374,
"step": 11320
},
{
"epoch": 1.450176,
"grad_norm": 0.7153387665748596,
"learning_rate": 0.0002851554484359971,
"loss": 2.0010080337524414,
"step": 11330
},
{
"epoch": 1.451456,
"grad_norm": 0.6429722905158997,
"learning_rate": 0.00028512920918867746,
"loss": 1.9750185012817383,
"step": 11340
},
{
"epoch": 1.452736,
"grad_norm": 0.7414952516555786,
"learning_rate": 0.00028510294798098936,
"loss": 1.989020538330078,
"step": 11350
},
{
"epoch": 1.454016,
"grad_norm": 0.6578467488288879,
"learning_rate": 0.00028507666481720063,
"loss": 1.9759967803955079,
"step": 11360
},
{
"epoch": 1.455296,
"grad_norm": 0.621391773223877,
"learning_rate": 0.00028505035970158266,
"loss": 1.9875198364257813,
"step": 11370
},
{
"epoch": 1.456576,
"grad_norm": 0.7765076160430908,
"learning_rate": 0.00028502403263841035,
"loss": 2.0161365509033202,
"step": 11380
},
{
"epoch": 1.457856,
"grad_norm": 0.6582878232002258,
"learning_rate": 0.0002849976836319623,
"loss": 2.0019020080566405,
"step": 11390
},
{
"epoch": 1.459136,
"grad_norm": 0.5817230343818665,
"learning_rate": 0.0002849713126865205,
"loss": 1.9872003555297852,
"step": 11400
},
{
"epoch": 1.460416,
"grad_norm": 0.6162863373756409,
"learning_rate": 0.00028494491980637064,
"loss": 1.995553207397461,
"step": 11410
},
{
"epoch": 1.4616959999999999,
"grad_norm": 0.6032384037971497,
"learning_rate": 0.0002849185049958019,
"loss": 1.9747112274169922,
"step": 11420
},
{
"epoch": 1.462976,
"grad_norm": 0.6233032941818237,
"learning_rate": 0.0002848920682591071,
"loss": 1.9824819564819336,
"step": 11430
},
{
"epoch": 1.464256,
"grad_norm": 0.6629032492637634,
"learning_rate": 0.00028486560960058256,
"loss": 2.005998992919922,
"step": 11440
},
{
"epoch": 1.465536,
"grad_norm": 0.544144332408905,
"learning_rate": 0.0002848391290245281,
"loss": 1.9982406616210937,
"step": 11450
},
{
"epoch": 1.4668160000000001,
"grad_norm": 0.650945782661438,
"learning_rate": 0.0002848126265352473,
"loss": 1.980784797668457,
"step": 11460
},
{
"epoch": 1.468096,
"grad_norm": 0.6223903894424438,
"learning_rate": 0.0002847861021370471,
"loss": 1.9930503845214844,
"step": 11470
},
{
"epoch": 1.469376,
"grad_norm": 0.6246350407600403,
"learning_rate": 0.00028475955583423816,
"loss": 1.9865238189697265,
"step": 11480
},
{
"epoch": 1.470656,
"grad_norm": 0.632250189781189,
"learning_rate": 0.00028473298763113453,
"loss": 1.9947776794433594,
"step": 11490
},
{
"epoch": 1.471936,
"grad_norm": 0.58414626121521,
"learning_rate": 0.00028470639753205395,
"loss": 2.005775260925293,
"step": 11500
},
{
"epoch": 1.473216,
"grad_norm": 0.595113217830658,
"learning_rate": 0.0002846797855413177,
"loss": 1.99066104888916,
"step": 11510
},
{
"epoch": 1.474496,
"grad_norm": 0.6954291462898254,
"learning_rate": 0.00028465315166325053,
"loss": 1.9903976440429687,
"step": 11520
},
{
"epoch": 1.475776,
"grad_norm": 0.6561645269393921,
"learning_rate": 0.00028462649590218093,
"loss": 2.022213935852051,
"step": 11530
},
{
"epoch": 1.477056,
"grad_norm": 0.684343159198761,
"learning_rate": 0.00028459981826244073,
"loss": 2.0005161285400392,
"step": 11540
},
{
"epoch": 1.478336,
"grad_norm": 0.7423267364501953,
"learning_rate": 0.00028457311874836546,
"loss": 1.9980703353881837,
"step": 11550
},
{
"epoch": 1.479616,
"grad_norm": 0.6137770414352417,
"learning_rate": 0.00028454639736429413,
"loss": 2.006209373474121,
"step": 11560
},
{
"epoch": 1.480896,
"grad_norm": 0.6377004981040955,
"learning_rate": 0.00028451965411456937,
"loss": 1.994059181213379,
"step": 11570
},
{
"epoch": 1.482176,
"grad_norm": 0.6084611415863037,
"learning_rate": 0.00028449288900353736,
"loss": 2.0026636123657227,
"step": 11580
},
{
"epoch": 1.4834559999999999,
"grad_norm": 0.6679419875144958,
"learning_rate": 0.0002844661020355477,
"loss": 2.0045246124267577,
"step": 11590
},
{
"epoch": 1.484736,
"grad_norm": 0.6842729449272156,
"learning_rate": 0.0002844392932149537,
"loss": 1.994192886352539,
"step": 11600
},
{
"epoch": 1.486016,
"grad_norm": 0.6864992380142212,
"learning_rate": 0.0002844124625461122,
"loss": 1.993010902404785,
"step": 11610
},
{
"epoch": 1.487296,
"grad_norm": 0.6094993352890015,
"learning_rate": 0.0002843856100333835,
"loss": 1.9936738967895509,
"step": 11620
},
{
"epoch": 1.4885760000000001,
"grad_norm": 0.7267225980758667,
"learning_rate": 0.0002843587356811315,
"loss": 1.9862861633300781,
"step": 11630
},
{
"epoch": 1.489856,
"grad_norm": 0.5966879725456238,
"learning_rate": 0.0002843318394937237,
"loss": 2.006046104431152,
"step": 11640
},
{
"epoch": 1.491136,
"grad_norm": 0.7184135913848877,
"learning_rate": 0.0002843049214755311,
"loss": 1.9959741592407227,
"step": 11650
},
{
"epoch": 1.492416,
"grad_norm": 0.6486974954605103,
"learning_rate": 0.0002842779816309282,
"loss": 1.9820661544799805,
"step": 11660
},
{
"epoch": 1.493696,
"grad_norm": 0.566840410232544,
"learning_rate": 0.00028425101996429314,
"loss": 1.9912609100341796,
"step": 11670
},
{
"epoch": 1.494976,
"grad_norm": 0.6932626962661743,
"learning_rate": 0.00028422403648000757,
"loss": 1.984910011291504,
"step": 11680
},
{
"epoch": 1.496256,
"grad_norm": 0.6808951497077942,
"learning_rate": 0.00028419703118245666,
"loss": 1.9870758056640625,
"step": 11690
},
{
"epoch": 1.497536,
"grad_norm": 0.6252458095550537,
"learning_rate": 0.00028417000407602916,
"loss": 1.9854507446289062,
"step": 11700
},
{
"epoch": 1.498816,
"grad_norm": 0.602453887462616,
"learning_rate": 0.00028414295516511733,
"loss": 1.9923028945922852,
"step": 11710
},
{
"epoch": 1.500096,
"grad_norm": 0.8114696741104126,
"learning_rate": 0.00028411588445411703,
"loss": 1.9952844619750976,
"step": 11720
},
{
"epoch": 1.501376,
"grad_norm": 0.6450442671775818,
"learning_rate": 0.00028408879194742755,
"loss": 1.9972211837768554,
"step": 11730
},
{
"epoch": 1.502656,
"grad_norm": 0.6286420226097107,
"learning_rate": 0.0002840616776494519,
"loss": 1.987843894958496,
"step": 11740
},
{
"epoch": 1.503936,
"grad_norm": 0.6167110204696655,
"learning_rate": 0.0002840345415645964,
"loss": 1.9556808471679688,
"step": 11750
},
{
"epoch": 1.5052159999999999,
"grad_norm": 0.7153220176696777,
"learning_rate": 0.00028400738369727113,
"loss": 1.9784412384033203,
"step": 11760
},
{
"epoch": 1.5064959999999998,
"grad_norm": 0.6213808655738831,
"learning_rate": 0.00028398020405188967,
"loss": 1.9949676513671875,
"step": 11770
},
{
"epoch": 1.507776,
"grad_norm": 0.6814656257629395,
"learning_rate": 0.00028395300263286895,
"loss": 1.999637985229492,
"step": 11780
},
{
"epoch": 1.509056,
"grad_norm": 0.6773201823234558,
"learning_rate": 0.0002839257794446297,
"loss": 2.0208032608032225,
"step": 11790
},
{
"epoch": 1.5103360000000001,
"grad_norm": 0.5861405730247498,
"learning_rate": 0.000283898534491596,
"loss": 2.0031352996826173,
"step": 11800
},
{
"epoch": 1.511616,
"grad_norm": 0.6029302477836609,
"learning_rate": 0.0002838712677781955,
"loss": 1.9905710220336914,
"step": 11810
},
{
"epoch": 1.512896,
"grad_norm": 0.6933667659759521,
"learning_rate": 0.0002838439793088595,
"loss": 1.9951766967773437,
"step": 11820
},
{
"epoch": 1.514176,
"grad_norm": 0.7042934894561768,
"learning_rate": 0.0002838166690880228,
"loss": 2.0045038223266602,
"step": 11830
},
{
"epoch": 1.515456,
"grad_norm": 0.7290188074111938,
"learning_rate": 0.0002837893371201235,
"loss": 1.997245979309082,
"step": 11840
},
{
"epoch": 1.5167359999999999,
"grad_norm": 0.6674178838729858,
"learning_rate": 0.0002837619834096036,
"loss": 1.9995222091674805,
"step": 11850
},
{
"epoch": 1.518016,
"grad_norm": 0.6585103273391724,
"learning_rate": 0.00028373460796090835,
"loss": 1.98739013671875,
"step": 11860
},
{
"epoch": 1.519296,
"grad_norm": 0.6088100075721741,
"learning_rate": 0.0002837072107784867,
"loss": 2.0116018295288085,
"step": 11870
},
{
"epoch": 1.5205760000000001,
"grad_norm": 0.6550244092941284,
"learning_rate": 0.00028367979186679106,
"loss": 2.0045097351074217,
"step": 11880
},
{
"epoch": 1.521856,
"grad_norm": 0.6974179148674011,
"learning_rate": 0.0002836523512302773,
"loss": 1.9653329849243164,
"step": 11890
},
{
"epoch": 1.523136,
"grad_norm": 0.6559057235717773,
"learning_rate": 0.00028362488887340506,
"loss": 1.997995376586914,
"step": 11900
},
{
"epoch": 1.524416,
"grad_norm": 0.612674355506897,
"learning_rate": 0.00028359740480063725,
"loss": 1.9864452362060547,
"step": 11910
},
{
"epoch": 1.525696,
"grad_norm": 0.5927303433418274,
"learning_rate": 0.00028356989901644046,
"loss": 1.9702861785888672,
"step": 11920
},
{
"epoch": 1.5269759999999999,
"grad_norm": 0.6803772449493408,
"learning_rate": 0.0002835423715252847,
"loss": 1.977646827697754,
"step": 11930
},
{
"epoch": 1.5282559999999998,
"grad_norm": 0.6468073725700378,
"learning_rate": 0.0002835148223316436,
"loss": 2.0037322998046876,
"step": 11940
},
{
"epoch": 1.529536,
"grad_norm": 0.6533370018005371,
"learning_rate": 0.0002834872514399943,
"loss": 2.019442558288574,
"step": 11950
},
{
"epoch": 1.530816,
"grad_norm": 0.6377413868904114,
"learning_rate": 0.00028345965885481745,
"loss": 1.9905149459838867,
"step": 11960
},
{
"epoch": 1.5320960000000001,
"grad_norm": 0.58443284034729,
"learning_rate": 0.0002834320445805972,
"loss": 1.9770112991333009,
"step": 11970
},
{
"epoch": 1.533376,
"grad_norm": 0.6253820061683655,
"learning_rate": 0.0002834044086218213,
"loss": 2.0002887725830076,
"step": 11980
},
{
"epoch": 1.534656,
"grad_norm": 0.6281173229217529,
"learning_rate": 0.00028337675098298094,
"loss": 1.989990234375,
"step": 11990
},
{
"epoch": 1.535936,
"grad_norm": 0.6314921975135803,
"learning_rate": 0.00028334907166857083,
"loss": 2.0067394256591795,
"step": 12000
},
{
"epoch": 1.537216,
"grad_norm": 0.6377007365226746,
"learning_rate": 0.0002833213706830893,
"loss": 1.973414993286133,
"step": 12010
},
{
"epoch": 1.5384959999999999,
"grad_norm": 0.5816660523414612,
"learning_rate": 0.0002832936480310382,
"loss": 1.9892141342163085,
"step": 12020
},
{
"epoch": 1.539776,
"grad_norm": 0.6100146174430847,
"learning_rate": 0.00028326590371692274,
"loss": 1.9866313934326172,
"step": 12030
},
{
"epoch": 1.541056,
"grad_norm": 0.72185218334198,
"learning_rate": 0.00028323813774525173,
"loss": 1.9793666839599608,
"step": 12040
},
{
"epoch": 1.5423360000000002,
"grad_norm": 0.6151171326637268,
"learning_rate": 0.00028321035012053764,
"loss": 1.9886417388916016,
"step": 12050
},
{
"epoch": 1.543616,
"grad_norm": 0.5789729356765747,
"learning_rate": 0.0002831825408472963,
"loss": 1.9868101119995116,
"step": 12060
},
{
"epoch": 1.544896,
"grad_norm": 0.6294896602630615,
"learning_rate": 0.0002831547099300471,
"loss": 1.998661422729492,
"step": 12070
},
{
"epoch": 1.546176,
"grad_norm": 0.6025799512863159,
"learning_rate": 0.00028312685737331294,
"loss": 2.003857231140137,
"step": 12080
},
{
"epoch": 1.547456,
"grad_norm": 0.615364670753479,
"learning_rate": 0.00028309898318162023,
"loss": 1.9846534729003906,
"step": 12090
},
{
"epoch": 1.548736,
"grad_norm": 0.7120283246040344,
"learning_rate": 0.00028307108735949895,
"loss": 1.982862663269043,
"step": 12100
},
{
"epoch": 1.5500159999999998,
"grad_norm": 0.6011831164360046,
"learning_rate": 0.00028304316991148253,
"loss": 1.9920019149780273,
"step": 12110
},
{
"epoch": 1.551296,
"grad_norm": 0.5913985371589661,
"learning_rate": 0.0002830152308421079,
"loss": 1.9857685089111328,
"step": 12120
},
{
"epoch": 1.552576,
"grad_norm": 0.6578329205513,
"learning_rate": 0.00028298727015591566,
"loss": 1.9907623291015626,
"step": 12130
},
{
"epoch": 1.5538560000000001,
"grad_norm": 0.7379078269004822,
"learning_rate": 0.00028295928785744974,
"loss": 1.9933242797851562,
"step": 12140
},
{
"epoch": 1.555136,
"grad_norm": 0.6420323252677917,
"learning_rate": 0.0002829312839512576,
"loss": 2.002197265625,
"step": 12150
},
{
"epoch": 1.556416,
"grad_norm": 0.5930525660514832,
"learning_rate": 0.00028290325844189043,
"loss": 1.9843338012695313,
"step": 12160
},
{
"epoch": 1.557696,
"grad_norm": 0.6576258540153503,
"learning_rate": 0.0002828752113339026,
"loss": 1.9914953231811523,
"step": 12170
},
{
"epoch": 1.558976,
"grad_norm": 0.6151474714279175,
"learning_rate": 0.00028284714263185215,
"loss": 1.983308982849121,
"step": 12180
},
{
"epoch": 1.5602559999999999,
"grad_norm": 0.6413775086402893,
"learning_rate": 0.00028281905234030073,
"loss": 1.9880573272705078,
"step": 12190
},
{
"epoch": 1.561536,
"grad_norm": 0.6724574565887451,
"learning_rate": 0.00028279094046381343,
"loss": 1.9918071746826171,
"step": 12200
},
{
"epoch": 1.562816,
"grad_norm": 0.5884711146354675,
"learning_rate": 0.0002827628070069587,
"loss": 1.997279167175293,
"step": 12210
},
{
"epoch": 1.5640960000000002,
"grad_norm": 0.6631693840026855,
"learning_rate": 0.00028273465197430867,
"loss": 1.9837032318115235,
"step": 12220
},
{
"epoch": 1.565376,
"grad_norm": 0.6248396039009094,
"learning_rate": 0.00028270647537043884,
"loss": 2.00390510559082,
"step": 12230
},
{
"epoch": 1.566656,
"grad_norm": 0.7199519872665405,
"learning_rate": 0.0002826782771999285,
"loss": 1.9853185653686523,
"step": 12240
},
{
"epoch": 1.567936,
"grad_norm": 0.6237788796424866,
"learning_rate": 0.00028265005746736004,
"loss": 1.9767860412597655,
"step": 12250
},
{
"epoch": 1.569216,
"grad_norm": 0.7757836580276489,
"learning_rate": 0.0002826218161773197,
"loss": 2.0048952102661133,
"step": 12260
},
{
"epoch": 1.570496,
"grad_norm": 0.6837695837020874,
"learning_rate": 0.00028259355333439695,
"loss": 1.9857776641845704,
"step": 12270
},
{
"epoch": 1.5717759999999998,
"grad_norm": 0.593316376209259,
"learning_rate": 0.000282565268943185,
"loss": 2.0092344284057617,
"step": 12280
},
{
"epoch": 1.573056,
"grad_norm": 0.61705082654953,
"learning_rate": 0.00028253696300828044,
"loss": 1.9957685470581055,
"step": 12290
},
{
"epoch": 1.574336,
"grad_norm": 0.6312508583068848,
"learning_rate": 0.00028250863553428335,
"loss": 1.9954248428344727,
"step": 12300
},
{
"epoch": 1.5756160000000001,
"grad_norm": 0.6899939775466919,
"learning_rate": 0.00028248028652579733,
"loss": 1.9767358779907227,
"step": 12310
},
{
"epoch": 1.576896,
"grad_norm": 0.5813516974449158,
"learning_rate": 0.0002824519159874295,
"loss": 2.0123149871826174,
"step": 12320
},
{
"epoch": 1.578176,
"grad_norm": 0.5960056185722351,
"learning_rate": 0.00028242352392379044,
"loss": 1.996229362487793,
"step": 12330
},
{
"epoch": 1.579456,
"grad_norm": 0.7136126160621643,
"learning_rate": 0.0002823951103394943,
"loss": 1.9698562622070312,
"step": 12340
},
{
"epoch": 1.580736,
"grad_norm": 0.599897027015686,
"learning_rate": 0.0002823666752391587,
"loss": 2.002450180053711,
"step": 12350
},
{
"epoch": 1.5820159999999999,
"grad_norm": 0.6239819526672363,
"learning_rate": 0.0002823382186274046,
"loss": 1.9910825729370116,
"step": 12360
},
{
"epoch": 1.583296,
"grad_norm": 0.6745089888572693,
"learning_rate": 0.00028230974050885677,
"loss": 2.003436279296875,
"step": 12370
},
{
"epoch": 1.584576,
"grad_norm": 0.578188419342041,
"learning_rate": 0.0002822812408881432,
"loss": 1.988958740234375,
"step": 12380
},
{
"epoch": 1.5858560000000002,
"grad_norm": 0.6390565037727356,
"learning_rate": 0.0002822527197698955,
"loss": 1.984975814819336,
"step": 12390
},
{
"epoch": 1.587136,
"grad_norm": 0.6547807455062866,
"learning_rate": 0.00028222417715874874,
"loss": 1.9811216354370118,
"step": 12400
},
{
"epoch": 1.588416,
"grad_norm": 0.5896871089935303,
"learning_rate": 0.00028219561305934147,
"loss": 1.9872743606567382,
"step": 12410
},
{
"epoch": 1.589696,
"grad_norm": 0.597897469997406,
"learning_rate": 0.00028216702747631576,
"loss": 1.9848625183105468,
"step": 12420
},
{
"epoch": 1.590976,
"grad_norm": 0.5630715489387512,
"learning_rate": 0.0002821384204143172,
"loss": 1.985508918762207,
"step": 12430
},
{
"epoch": 1.592256,
"grad_norm": 0.7521433234214783,
"learning_rate": 0.0002821097918779948,
"loss": 1.9825284957885743,
"step": 12440
},
{
"epoch": 1.5935359999999998,
"grad_norm": 0.6286043524742126,
"learning_rate": 0.0002820811418720011,
"loss": 1.992232322692871,
"step": 12450
},
{
"epoch": 1.594816,
"grad_norm": 0.7475308179855347,
"learning_rate": 0.0002820524704009921,
"loss": 1.9865779876708984,
"step": 12460
},
{
"epoch": 1.596096,
"grad_norm": 0.7135626077651978,
"learning_rate": 0.00028202377746962737,
"loss": 1.9769983291625977,
"step": 12470
},
{
"epoch": 1.5973760000000001,
"grad_norm": 0.6866909265518188,
"learning_rate": 0.0002819950630825699,
"loss": 1.996453094482422,
"step": 12480
},
{
"epoch": 1.598656,
"grad_norm": 0.5828826427459717,
"learning_rate": 0.0002819663272444861,
"loss": 1.9783212661743164,
"step": 12490
},
{
"epoch": 1.599936,
"grad_norm": 0.6047083735466003,
"learning_rate": 0.000281937569960046,
"loss": 2.0021230697631838,
"step": 12500
},
{
"epoch": 1.601216,
"grad_norm": 0.6071474552154541,
"learning_rate": 0.0002819087912339231,
"loss": 1.9867988586425782,
"step": 12510
},
{
"epoch": 1.602496,
"grad_norm": 0.5802363753318787,
"learning_rate": 0.00028187999107079424,
"loss": 1.9826335906982422,
"step": 12520
},
{
"epoch": 1.6037759999999999,
"grad_norm": 0.5885127186775208,
"learning_rate": 0.00028185116947533997,
"loss": 1.9890375137329102,
"step": 12530
},
{
"epoch": 1.605056,
"grad_norm": 0.6466164588928223,
"learning_rate": 0.000281822326452244,
"loss": 1.9949928283691407,
"step": 12540
},
{
"epoch": 1.606336,
"grad_norm": 0.5743044018745422,
"learning_rate": 0.0002817934620061939,
"loss": 2.002297019958496,
"step": 12550
},
{
"epoch": 1.6076160000000002,
"grad_norm": 0.6383411884307861,
"learning_rate": 0.0002817645761418805,
"loss": 2.0104206085205076,
"step": 12560
},
{
"epoch": 1.608896,
"grad_norm": 0.672825276851654,
"learning_rate": 0.0002817356688639981,
"loss": 2.001966667175293,
"step": 12570
},
{
"epoch": 1.610176,
"grad_norm": 0.5477749109268188,
"learning_rate": 0.00028170674017724464,
"loss": 1.985647201538086,
"step": 12580
},
{
"epoch": 1.611456,
"grad_norm": 0.56830894947052,
"learning_rate": 0.0002816777900863214,
"loss": 2.0019514083862306,
"step": 12590
},
{
"epoch": 1.612736,
"grad_norm": 0.6407654881477356,
"learning_rate": 0.0002816488185959331,
"loss": 1.9946565628051758,
"step": 12600
},
{
"epoch": 1.614016,
"grad_norm": 0.6300269365310669,
"learning_rate": 0.000281619825710788,
"loss": 2.0234329223632814,
"step": 12610
},
{
"epoch": 1.6152959999999998,
"grad_norm": 0.5886912941932678,
"learning_rate": 0.0002815908114355979,
"loss": 1.9737695693969726,
"step": 12620
},
{
"epoch": 1.616576,
"grad_norm": 0.6044586896896362,
"learning_rate": 0.000281561775775078,
"loss": 1.9872671127319337,
"step": 12630
},
{
"epoch": 1.617856,
"grad_norm": 0.661132276058197,
"learning_rate": 0.00028153271873394706,
"loss": 1.9847532272338868,
"step": 12640
},
{
"epoch": 1.6191360000000001,
"grad_norm": 0.7229477763175964,
"learning_rate": 0.0002815036403169272,
"loss": 2.003045845031738,
"step": 12650
},
{
"epoch": 1.620416,
"grad_norm": 0.6306692361831665,
"learning_rate": 0.00028147454052874403,
"loss": 1.9856027603149413,
"step": 12660
},
{
"epoch": 1.621696,
"grad_norm": 0.606751561164856,
"learning_rate": 0.0002814454193741267,
"loss": 1.9890954971313477,
"step": 12670
},
{
"epoch": 1.622976,
"grad_norm": 0.6360762715339661,
"learning_rate": 0.00028141627685780783,
"loss": 1.995732879638672,
"step": 12680
},
{
"epoch": 1.624256,
"grad_norm": 0.6043411493301392,
"learning_rate": 0.00028138711298452345,
"loss": 1.9911890029907227,
"step": 12690
},
{
"epoch": 1.6255359999999999,
"grad_norm": 0.6502869725227356,
"learning_rate": 0.00028135792775901315,
"loss": 1.967578125,
"step": 12700
},
{
"epoch": 1.626816,
"grad_norm": 0.5480800271034241,
"learning_rate": 0.0002813287211860199,
"loss": 1.9979866027832032,
"step": 12710
},
{
"epoch": 1.628096,
"grad_norm": 0.6363219022750854,
"learning_rate": 0.00028129949327029006,
"loss": 1.9978750228881836,
"step": 12720
},
{
"epoch": 1.6293760000000002,
"grad_norm": 0.6919147968292236,
"learning_rate": 0.00028127024401657373,
"loss": 1.9816198348999023,
"step": 12730
},
{
"epoch": 1.630656,
"grad_norm": 0.7384796738624573,
"learning_rate": 0.0002812409734296242,
"loss": 1.999420166015625,
"step": 12740
},
{
"epoch": 1.631936,
"grad_norm": 0.6270831227302551,
"learning_rate": 0.0002812116815141985,
"loss": 1.990450668334961,
"step": 12750
},
{
"epoch": 1.633216,
"grad_norm": 0.6797302961349487,
"learning_rate": 0.00028118236827505687,
"loss": 1.9965476989746094,
"step": 12760
},
{
"epoch": 1.634496,
"grad_norm": 0.626695454120636,
"learning_rate": 0.000281153033716963,
"loss": 1.9968223571777344,
"step": 12770
},
{
"epoch": 1.635776,
"grad_norm": 0.6879094839096069,
"learning_rate": 0.00028112367784468445,
"loss": 1.984914207458496,
"step": 12780
},
{
"epoch": 1.6370559999999998,
"grad_norm": 0.586521565914154,
"learning_rate": 0.0002810943006629917,
"loss": 1.9820964813232422,
"step": 12790
},
{
"epoch": 1.638336,
"grad_norm": 0.59141606092453,
"learning_rate": 0.0002810649021766591,
"loss": 1.9659967422485352,
"step": 12800
},
{
"epoch": 1.639616,
"grad_norm": 0.6080377101898193,
"learning_rate": 0.00028103548239046417,
"loss": 2.012135124206543,
"step": 12810
},
{
"epoch": 1.6408960000000001,
"grad_norm": 0.6091150641441345,
"learning_rate": 0.00028100604130918815,
"loss": 1.9932632446289062,
"step": 12820
},
{
"epoch": 1.642176,
"grad_norm": 0.6203098297119141,
"learning_rate": 0.0002809765789376156,
"loss": 1.995589256286621,
"step": 12830
},
{
"epoch": 1.643456,
"grad_norm": 0.6006268262863159,
"learning_rate": 0.0002809470952805346,
"loss": 1.992354965209961,
"step": 12840
},
{
"epoch": 1.644736,
"grad_norm": 0.6625576019287109,
"learning_rate": 0.0002809175903427365,
"loss": 1.9785432815551758,
"step": 12850
},
{
"epoch": 1.646016,
"grad_norm": 0.6188414692878723,
"learning_rate": 0.0002808880641290164,
"loss": 1.981410598754883,
"step": 12860
},
{
"epoch": 1.6472959999999999,
"grad_norm": 0.6317717432975769,
"learning_rate": 0.0002808585166441728,
"loss": 1.984215545654297,
"step": 12870
},
{
"epoch": 1.648576,
"grad_norm": 0.6587024927139282,
"learning_rate": 0.0002808289478930073,
"loss": 1.9941490173339844,
"step": 12880
},
{
"epoch": 1.649856,
"grad_norm": 0.6256428956985474,
"learning_rate": 0.0002807993578803255,
"loss": 1.9906015396118164,
"step": 12890
},
{
"epoch": 1.6511360000000002,
"grad_norm": 0.605486273765564,
"learning_rate": 0.00028076974661093605,
"loss": 1.99066162109375,
"step": 12900
},
{
"epoch": 1.652416,
"grad_norm": 0.6108291149139404,
"learning_rate": 0.0002807401140896512,
"loss": 1.9964624404907227,
"step": 12910
},
{
"epoch": 1.653696,
"grad_norm": 0.631614089012146,
"learning_rate": 0.00028071046032128666,
"loss": 1.9938804626464843,
"step": 12920
},
{
"epoch": 1.654976,
"grad_norm": 0.6231254935264587,
"learning_rate": 0.00028068078531066166,
"loss": 2.000991630554199,
"step": 12930
},
{
"epoch": 1.656256,
"grad_norm": 0.6264880895614624,
"learning_rate": 0.00028065108906259864,
"loss": 1.9956014633178711,
"step": 12940
},
{
"epoch": 1.657536,
"grad_norm": 0.6015536785125732,
"learning_rate": 0.0002806213715819238,
"loss": 1.9887815475463868,
"step": 12950
},
{
"epoch": 1.6588159999999998,
"grad_norm": 0.6260659694671631,
"learning_rate": 0.00028059163287346653,
"loss": 1.9906013488769532,
"step": 12960
},
{
"epoch": 1.660096,
"grad_norm": 0.6274285316467285,
"learning_rate": 0.0002805618729420598,
"loss": 2.019454002380371,
"step": 12970
},
{
"epoch": 1.661376,
"grad_norm": 0.6511818766593933,
"learning_rate": 0.00028053209179254015,
"loss": 2.0094207763671874,
"step": 12980
},
{
"epoch": 1.6626560000000001,
"grad_norm": 0.6439974904060364,
"learning_rate": 0.0002805022894297472,
"loss": 1.9952024459838866,
"step": 12990
},
{
"epoch": 1.663936,
"grad_norm": 0.6719408631324768,
"learning_rate": 0.0002804724658585245,
"loss": 2.0143915176391602,
"step": 13000
},
{
"epoch": 1.665216,
"grad_norm": 0.5965760946273804,
"learning_rate": 0.0002804426210837186,
"loss": 2.0003263473510744,
"step": 13010
},
{
"epoch": 1.666496,
"grad_norm": 0.5631793737411499,
"learning_rate": 0.00028041275511017975,
"loss": 1.977505874633789,
"step": 13020
},
{
"epoch": 1.667776,
"grad_norm": 0.5862990021705627,
"learning_rate": 0.0002803828679427615,
"loss": 2.0093894958496095,
"step": 13030
},
{
"epoch": 1.6690559999999999,
"grad_norm": 0.6070151329040527,
"learning_rate": 0.00028035295958632114,
"loss": 2.000132942199707,
"step": 13040
},
{
"epoch": 1.670336,
"grad_norm": 0.5861722826957703,
"learning_rate": 0.000280323030045719,
"loss": 1.9736347198486328,
"step": 13050
},
{
"epoch": 1.671616,
"grad_norm": 0.8306954503059387,
"learning_rate": 0.00028029307932581917,
"loss": 2.0125246047973633,
"step": 13060
},
{
"epoch": 1.6728960000000002,
"grad_norm": 0.6431558132171631,
"learning_rate": 0.00028026310743148894,
"loss": 2.0154903411865233,
"step": 13070
},
{
"epoch": 1.674176,
"grad_norm": 0.7464187145233154,
"learning_rate": 0.0002802331143675992,
"loss": 1.9874631881713867,
"step": 13080
},
{
"epoch": 1.675456,
"grad_norm": 0.6642718315124512,
"learning_rate": 0.0002802031001390243,
"loss": 2.01281623840332,
"step": 13090
},
{
"epoch": 1.676736,
"grad_norm": 0.633001446723938,
"learning_rate": 0.00028017306475064196,
"loss": 1.9952484130859376,
"step": 13100
},
{
"epoch": 1.678016,
"grad_norm": 0.6135600209236145,
"learning_rate": 0.00028014300820733325,
"loss": 1.971938705444336,
"step": 13110
},
{
"epoch": 1.679296,
"grad_norm": 0.5765730738639832,
"learning_rate": 0.0002801129305139829,
"loss": 1.997808074951172,
"step": 13120
},
{
"epoch": 1.6805759999999998,
"grad_norm": 0.544807493686676,
"learning_rate": 0.0002800828316754788,
"loss": 1.982986831665039,
"step": 13130
},
{
"epoch": 1.681856,
"grad_norm": 0.6663402915000916,
"learning_rate": 0.00028005271169671256,
"loss": 1.9861152648925782,
"step": 13140
},
{
"epoch": 1.683136,
"grad_norm": 0.6248042583465576,
"learning_rate": 0.00028002257058257907,
"loss": 2.0063156127929687,
"step": 13150
},
{
"epoch": 1.6844160000000001,
"grad_norm": 0.5508193373680115,
"learning_rate": 0.0002799924083379766,
"loss": 1.9891693115234375,
"step": 13160
},
{
"epoch": 1.685696,
"grad_norm": 0.5529075264930725,
"learning_rate": 0.000279962224967807,
"loss": 1.9941001892089845,
"step": 13170
},
{
"epoch": 1.686976,
"grad_norm": 0.6517800092697144,
"learning_rate": 0.0002799320204769755,
"loss": 1.987466049194336,
"step": 13180
},
{
"epoch": 1.688256,
"grad_norm": 0.6514473557472229,
"learning_rate": 0.0002799017948703907,
"loss": 1.9915737152099608,
"step": 13190
},
{
"epoch": 1.689536,
"grad_norm": 0.6751680374145508,
"learning_rate": 0.0002798715481529647,
"loss": 1.9961626052856445,
"step": 13200
},
{
"epoch": 1.6908159999999999,
"grad_norm": 0.710557758808136,
"learning_rate": 0.00027984128032961306,
"loss": 1.9766324996948241,
"step": 13210
},
{
"epoch": 1.692096,
"grad_norm": 0.6588106155395508,
"learning_rate": 0.0002798109914052546,
"loss": 2.002887725830078,
"step": 13220
},
{
"epoch": 1.693376,
"grad_norm": 0.6127070188522339,
"learning_rate": 0.00027978068138481184,
"loss": 2.004521369934082,
"step": 13230
},
{
"epoch": 1.6946560000000002,
"grad_norm": 0.6921170949935913,
"learning_rate": 0.0002797503502732105,
"loss": 1.9970478057861327,
"step": 13240
},
{
"epoch": 1.695936,
"grad_norm": 0.5995293855667114,
"learning_rate": 0.00027971999807537977,
"loss": 1.985312271118164,
"step": 13250
},
{
"epoch": 1.697216,
"grad_norm": 0.6018744111061096,
"learning_rate": 0.00027968962479625234,
"loss": 1.9960086822509766,
"step": 13260
},
{
"epoch": 1.698496,
"grad_norm": 0.6067647337913513,
"learning_rate": 0.00027965923044076426,
"loss": 1.990829086303711,
"step": 13270
},
{
"epoch": 1.699776,
"grad_norm": 0.606651782989502,
"learning_rate": 0.00027962881501385515,
"loss": 1.9917610168457032,
"step": 13280
},
{
"epoch": 1.701056,
"grad_norm": 0.6090924739837646,
"learning_rate": 0.0002795983785204678,
"loss": 1.9930423736572265,
"step": 13290
},
{
"epoch": 1.7023359999999998,
"grad_norm": 0.5838082432746887,
"learning_rate": 0.00027956792096554855,
"loss": 2.0029977798461913,
"step": 13300
},
{
"epoch": 1.703616,
"grad_norm": 0.6076372265815735,
"learning_rate": 0.00027953744235404733,
"loss": 1.967214012145996,
"step": 13310
},
{
"epoch": 1.704896,
"grad_norm": 0.6525275111198425,
"learning_rate": 0.00027950694269091725,
"loss": 1.9967252731323242,
"step": 13320
},
{
"epoch": 1.7061760000000001,
"grad_norm": 0.5798757672309875,
"learning_rate": 0.00027947642198111484,
"loss": 1.9798770904541017,
"step": 13330
},
{
"epoch": 1.707456,
"grad_norm": 0.5803235173225403,
"learning_rate": 0.0002794458802296003,
"loss": 1.9929985046386718,
"step": 13340
},
{
"epoch": 1.708736,
"grad_norm": 0.55645352602005,
"learning_rate": 0.000279415317441337,
"loss": 2.0056087493896486,
"step": 13350
},
{
"epoch": 1.710016,
"grad_norm": 0.6510474681854248,
"learning_rate": 0.0002793847336212918,
"loss": 1.9835859298706056,
"step": 13360
},
{
"epoch": 1.711296,
"grad_norm": 0.6064742207527161,
"learning_rate": 0.000279354128774435,
"loss": 2.0099687576293945,
"step": 13370
},
{
"epoch": 1.7125759999999999,
"grad_norm": 0.6000496745109558,
"learning_rate": 0.0002793235029057403,
"loss": 1.9745065689086914,
"step": 13380
},
{
"epoch": 1.713856,
"grad_norm": 0.6242145895957947,
"learning_rate": 0.00027929285602018496,
"loss": 1.9916412353515625,
"step": 13390
},
{
"epoch": 1.715136,
"grad_norm": 0.5768315196037292,
"learning_rate": 0.0002792621881227494,
"loss": 1.9905712127685546,
"step": 13400
},
{
"epoch": 1.7164160000000002,
"grad_norm": 0.6390241980552673,
"learning_rate": 0.00027923149921841754,
"loss": 1.96690673828125,
"step": 13410
},
{
"epoch": 1.7176960000000001,
"grad_norm": 0.6659237146377563,
"learning_rate": 0.0002792007893121768,
"loss": 1.9732889175415038,
"step": 13420
},
{
"epoch": 1.718976,
"grad_norm": 0.642673671245575,
"learning_rate": 0.00027917005840901805,
"loss": 2.0019617080688477,
"step": 13430
},
{
"epoch": 1.720256,
"grad_norm": 0.599125325679779,
"learning_rate": 0.00027913930651393535,
"loss": 1.9965097427368164,
"step": 13440
},
{
"epoch": 1.721536,
"grad_norm": 0.6379650831222534,
"learning_rate": 0.00027910853363192643,
"loss": 1.9848491668701171,
"step": 13450
},
{
"epoch": 1.722816,
"grad_norm": 0.6716325879096985,
"learning_rate": 0.0002790777397679923,
"loss": 1.978513526916504,
"step": 13460
},
{
"epoch": 1.7240959999999999,
"grad_norm": 0.6485804915428162,
"learning_rate": 0.0002790469249271372,
"loss": 1.9878868103027343,
"step": 13470
},
{
"epoch": 1.725376,
"grad_norm": 0.7345725893974304,
"learning_rate": 0.0002790160891143692,
"loss": 1.992083740234375,
"step": 13480
},
{
"epoch": 1.726656,
"grad_norm": 0.6288764476776123,
"learning_rate": 0.0002789852323346995,
"loss": 2.0068994522094727,
"step": 13490
},
{
"epoch": 1.7279360000000001,
"grad_norm": 0.6416577696800232,
"learning_rate": 0.0002789543545931426,
"loss": 1.9909046173095704,
"step": 13500
},
{
"epoch": 1.729216,
"grad_norm": 0.6367363333702087,
"learning_rate": 0.00027892345589471677,
"loss": 2.0024709701538086,
"step": 13510
},
{
"epoch": 1.730496,
"grad_norm": 0.6156037449836731,
"learning_rate": 0.0002788925362444433,
"loss": 1.9947532653808593,
"step": 13520
},
{
"epoch": 1.731776,
"grad_norm": 0.5910305380821228,
"learning_rate": 0.00027886159564734713,
"loss": 1.9946659088134766,
"step": 13530
},
{
"epoch": 1.733056,
"grad_norm": 0.5788825154304504,
"learning_rate": 0.00027883063410845665,
"loss": 1.9830217361450195,
"step": 13540
},
{
"epoch": 1.7343359999999999,
"grad_norm": 0.6356834173202515,
"learning_rate": 0.00027879965163280337,
"loss": 1.989356803894043,
"step": 13550
},
{
"epoch": 1.735616,
"grad_norm": 0.6032944321632385,
"learning_rate": 0.00027876864822542243,
"loss": 1.9937528610229491,
"step": 13560
},
{
"epoch": 1.736896,
"grad_norm": 0.6071460247039795,
"learning_rate": 0.00027873762389135236,
"loss": 2.0006227493286133,
"step": 13570
},
{
"epoch": 1.7381760000000002,
"grad_norm": 0.7040545344352722,
"learning_rate": 0.000278706578635635,
"loss": 2.0084535598754885,
"step": 13580
},
{
"epoch": 1.7394560000000001,
"grad_norm": 0.6047849655151367,
"learning_rate": 0.0002786755124633156,
"loss": 1.9948095321655273,
"step": 13590
},
{
"epoch": 1.740736,
"grad_norm": 0.6119142770767212,
"learning_rate": 0.0002786444253794429,
"loss": 1.996514129638672,
"step": 13600
},
{
"epoch": 1.742016,
"grad_norm": 0.5880126357078552,
"learning_rate": 0.000278613317389069,
"loss": 1.9876934051513673,
"step": 13610
},
{
"epoch": 1.743296,
"grad_norm": 0.5863587856292725,
"learning_rate": 0.0002785821884972493,
"loss": 2.005519485473633,
"step": 13620
},
{
"epoch": 1.744576,
"grad_norm": 0.6820835471153259,
"learning_rate": 0.00027855103870904274,
"loss": 1.9719966888427733,
"step": 13630
},
{
"epoch": 1.7458559999999999,
"grad_norm": 0.6092546582221985,
"learning_rate": 0.0002785198680295115,
"loss": 2.006068992614746,
"step": 13640
},
{
"epoch": 1.747136,
"grad_norm": 0.5908424258232117,
"learning_rate": 0.00027848867646372143,
"loss": 1.9989572525024415,
"step": 13650
},
{
"epoch": 1.748416,
"grad_norm": 0.6115155220031738,
"learning_rate": 0.00027845746401674145,
"loss": 1.9978038787841796,
"step": 13660
},
{
"epoch": 1.7496960000000001,
"grad_norm": 0.6529688835144043,
"learning_rate": 0.000278426230693644,
"loss": 1.9970335006713866,
"step": 13670
},
{
"epoch": 1.750976,
"grad_norm": 0.6464643478393555,
"learning_rate": 0.00027839497649950496,
"loss": 1.9870317459106446,
"step": 13680
},
{
"epoch": 1.752256,
"grad_norm": 0.6392168998718262,
"learning_rate": 0.00027836370143940364,
"loss": 1.975596809387207,
"step": 13690
},
{
"epoch": 1.753536,
"grad_norm": 0.5521542429924011,
"learning_rate": 0.0002783324055184226,
"loss": 1.982096290588379,
"step": 13700
},
{
"epoch": 1.754816,
"grad_norm": 0.5614274740219116,
"learning_rate": 0.0002783010887416479,
"loss": 1.9764310836791992,
"step": 13710
},
{
"epoch": 1.7560959999999999,
"grad_norm": 0.6286564469337463,
"learning_rate": 0.00027826975111416884,
"loss": 1.9740509033203124,
"step": 13720
},
{
"epoch": 1.757376,
"grad_norm": 0.6732945442199707,
"learning_rate": 0.00027823839264107833,
"loss": 1.9812362670898438,
"step": 13730
},
{
"epoch": 1.758656,
"grad_norm": 0.8319885730743408,
"learning_rate": 0.0002782070133274726,
"loss": 1.9909189224243165,
"step": 13740
},
{
"epoch": 1.759936,
"grad_norm": 0.6059410572052002,
"learning_rate": 0.0002781756131784511,
"loss": 1.980379867553711,
"step": 13750
},
{
"epoch": 1.7612160000000001,
"grad_norm": 0.5295083522796631,
"learning_rate": 0.0002781441921991168,
"loss": 1.9844615936279297,
"step": 13760
},
{
"epoch": 1.762496,
"grad_norm": 0.6505018472671509,
"learning_rate": 0.00027811275039457613,
"loss": 1.9884096145629884,
"step": 13770
},
{
"epoch": 1.763776,
"grad_norm": 0.7103697657585144,
"learning_rate": 0.0002780812877699388,
"loss": 1.9862951278686523,
"step": 13780
},
{
"epoch": 1.765056,
"grad_norm": 0.5813751816749573,
"learning_rate": 0.00027804980433031784,
"loss": 1.9884418487548827,
"step": 13790
},
{
"epoch": 1.766336,
"grad_norm": 0.6926559209823608,
"learning_rate": 0.00027801830008082986,
"loss": 1.9918901443481445,
"step": 13800
},
{
"epoch": 1.7676159999999999,
"grad_norm": 0.5946470499038696,
"learning_rate": 0.00027798677502659464,
"loss": 1.9839643478393554,
"step": 13810
},
{
"epoch": 1.768896,
"grad_norm": 0.557457447052002,
"learning_rate": 0.00027795522917273545,
"loss": 1.9885560989379882,
"step": 13820
},
{
"epoch": 1.770176,
"grad_norm": 0.7370789647102356,
"learning_rate": 0.00027792366252437896,
"loss": 1.986794662475586,
"step": 13830
},
{
"epoch": 1.7714560000000001,
"grad_norm": 0.5929363369941711,
"learning_rate": 0.0002778920750866553,
"loss": 1.9836645126342773,
"step": 13840
},
{
"epoch": 1.772736,
"grad_norm": 0.5790180563926697,
"learning_rate": 0.00027786046686469756,
"loss": 1.9910282135009765,
"step": 13850
},
{
"epoch": 1.774016,
"grad_norm": 0.5799253582954407,
"learning_rate": 0.0002778288378636428,
"loss": 1.9691133499145508,
"step": 13860
},
{
"epoch": 1.775296,
"grad_norm": 0.7639281749725342,
"learning_rate": 0.00027779718808863113,
"loss": 2.0000869750976564,
"step": 13870
},
{
"epoch": 1.776576,
"grad_norm": 0.5788276791572571,
"learning_rate": 0.0002777655175448059,
"loss": 1.9941930770874023,
"step": 13880
},
{
"epoch": 1.7778559999999999,
"grad_norm": 0.7169702053070068,
"learning_rate": 0.0002777338262373142,
"loss": 2.001001739501953,
"step": 13890
},
{
"epoch": 1.779136,
"grad_norm": 0.5163822174072266,
"learning_rate": 0.00027770211417130626,
"loss": 1.9909893035888673,
"step": 13900
},
{
"epoch": 1.780416,
"grad_norm": 0.7218848466873169,
"learning_rate": 0.0002776703813519357,
"loss": 2.000100517272949,
"step": 13910
},
{
"epoch": 1.781696,
"grad_norm": 0.6570733785629272,
"learning_rate": 0.0002776386277843595,
"loss": 1.9942224502563477,
"step": 13920
},
{
"epoch": 1.7829760000000001,
"grad_norm": 0.6489054560661316,
"learning_rate": 0.00027760685347373815,
"loss": 1.9865427017211914,
"step": 13930
},
{
"epoch": 1.784256,
"grad_norm": 0.6423578262329102,
"learning_rate": 0.0002775750584252354,
"loss": 1.9879083633422852,
"step": 13940
},
{
"epoch": 1.785536,
"grad_norm": 0.5846525430679321,
"learning_rate": 0.0002775432426440183,
"loss": 1.998491096496582,
"step": 13950
},
{
"epoch": 1.786816,
"grad_norm": 0.6934936046600342,
"learning_rate": 0.0002775114061352574,
"loss": 2.0036827087402345,
"step": 13960
},
{
"epoch": 1.788096,
"grad_norm": 0.7417251467704773,
"learning_rate": 0.00027747954890412665,
"loss": 1.993621826171875,
"step": 13970
},
{
"epoch": 1.7893759999999999,
"grad_norm": 0.6612619161605835,
"learning_rate": 0.00027744767095580323,
"loss": 1.9732223510742188,
"step": 13980
},
{
"epoch": 1.790656,
"grad_norm": 0.6654414534568787,
"learning_rate": 0.0002774157722954677,
"loss": 1.9863597869873046,
"step": 13990
},
{
"epoch": 1.791936,
"grad_norm": 0.6348403096199036,
"learning_rate": 0.00027738385292830407,
"loss": 2.0069671630859376,
"step": 14000
},
{
"epoch": 1.7932160000000001,
"grad_norm": 0.6972537636756897,
"learning_rate": 0.00027735191285949975,
"loss": 1.9975904464721679,
"step": 14010
},
{
"epoch": 1.794496,
"grad_norm": 0.5936768651008606,
"learning_rate": 0.0002773199520942453,
"loss": 1.9938947677612304,
"step": 14020
},
{
"epoch": 1.795776,
"grad_norm": 0.5596235990524292,
"learning_rate": 0.00027728797063773493,
"loss": 1.9974689483642578,
"step": 14030
},
{
"epoch": 1.797056,
"grad_norm": 0.6101556420326233,
"learning_rate": 0.000277255968495166,
"loss": 2.0115772247314454,
"step": 14040
},
{
"epoch": 1.798336,
"grad_norm": 0.6465120911598206,
"learning_rate": 0.0002772239456717393,
"loss": 1.9835803985595704,
"step": 14050
},
{
"epoch": 1.7996159999999999,
"grad_norm": 0.7538608312606812,
"learning_rate": 0.000277191902172659,
"loss": 2.005080223083496,
"step": 14060
},
{
"epoch": 1.800896,
"grad_norm": 0.7185438871383667,
"learning_rate": 0.0002771598380031326,
"loss": 1.9749935150146485,
"step": 14070
},
{
"epoch": 1.802176,
"grad_norm": 0.5687502026557922,
"learning_rate": 0.000277127753168371,
"loss": 1.9848592758178711,
"step": 14080
},
{
"epoch": 1.803456,
"grad_norm": 0.6506163477897644,
"learning_rate": 0.0002770956476735884,
"loss": 1.9971248626708984,
"step": 14090
},
{
"epoch": 1.8047360000000001,
"grad_norm": 0.6640498042106628,
"learning_rate": 0.0002770635215240024,
"loss": 2.0049943923950195,
"step": 14100
},
{
"epoch": 1.806016,
"grad_norm": 0.5507321953773499,
"learning_rate": 0.00027703137472483396,
"loss": 1.981342887878418,
"step": 14110
},
{
"epoch": 1.807296,
"grad_norm": 0.5848096013069153,
"learning_rate": 0.00027699920728130735,
"loss": 2.0065704345703126,
"step": 14120
},
{
"epoch": 1.808576,
"grad_norm": 0.7137436270713806,
"learning_rate": 0.0002769670191986502,
"loss": 1.9926156997680664,
"step": 14130
},
{
"epoch": 1.809856,
"grad_norm": 0.6607388854026794,
"learning_rate": 0.0002769348104820936,
"loss": 2.004817008972168,
"step": 14140
},
{
"epoch": 1.8111359999999999,
"grad_norm": 0.589976966381073,
"learning_rate": 0.0002769025811368718,
"loss": 1.9768518447875976,
"step": 14150
},
{
"epoch": 1.812416,
"grad_norm": 0.6693985462188721,
"learning_rate": 0.00027687033116822264,
"loss": 1.9946104049682618,
"step": 14160
},
{
"epoch": 1.813696,
"grad_norm": 0.6808820962905884,
"learning_rate": 0.0002768380605813871,
"loss": 1.9783695220947266,
"step": 14170
},
{
"epoch": 1.8149760000000001,
"grad_norm": 0.6001582741737366,
"learning_rate": 0.00027680576938160967,
"loss": 1.9821128845214844,
"step": 14180
},
{
"epoch": 1.816256,
"grad_norm": 0.5757496953010559,
"learning_rate": 0.000276773457574138,
"loss": 1.9941371917724608,
"step": 14190
},
{
"epoch": 1.817536,
"grad_norm": 0.6027823090553284,
"learning_rate": 0.00027674112516422333,
"loss": 1.975214958190918,
"step": 14200
},
{
"epoch": 1.818816,
"grad_norm": 0.7105240225791931,
"learning_rate": 0.0002767087721571201,
"loss": 1.9998865127563477,
"step": 14210
},
{
"epoch": 1.820096,
"grad_norm": 0.6610202193260193,
"learning_rate": 0.00027667639855808606,
"loss": 1.9810190200805664,
"step": 14220
},
{
"epoch": 1.8213759999999999,
"grad_norm": 0.6623238325119019,
"learning_rate": 0.0002766440043723824,
"loss": 2.0051544189453123,
"step": 14230
},
{
"epoch": 1.822656,
"grad_norm": 0.6435370445251465,
"learning_rate": 0.00027661158960527363,
"loss": 2.0050176620483398,
"step": 14240
},
{
"epoch": 1.823936,
"grad_norm": 0.6199061870574951,
"learning_rate": 0.0002765791542620276,
"loss": 1.9975845336914062,
"step": 14250
},
{
"epoch": 1.825216,
"grad_norm": 0.620334804058075,
"learning_rate": 0.00027654669834791554,
"loss": 2.000550079345703,
"step": 14260
},
{
"epoch": 1.8264960000000001,
"grad_norm": 0.614678144454956,
"learning_rate": 0.0002765142218682119,
"loss": 1.9887870788574218,
"step": 14270
},
{
"epoch": 1.827776,
"grad_norm": 0.6724008917808533,
"learning_rate": 0.0002764817248281946,
"loss": 2.0075620651245116,
"step": 14280
},
{
"epoch": 1.829056,
"grad_norm": 0.5586716532707214,
"learning_rate": 0.00027644920723314493,
"loss": 1.9960405349731445,
"step": 14290
},
{
"epoch": 1.830336,
"grad_norm": 0.6070801019668579,
"learning_rate": 0.0002764166690883473,
"loss": 1.9760414123535157,
"step": 14300
},
{
"epoch": 1.831616,
"grad_norm": 0.6035956740379333,
"learning_rate": 0.00027638411039908973,
"loss": 1.9959955215454102,
"step": 14310
},
{
"epoch": 1.8328959999999999,
"grad_norm": 0.5952864289283752,
"learning_rate": 0.0002763515311706634,
"loss": 1.9982990264892577,
"step": 14320
},
{
"epoch": 1.834176,
"grad_norm": 0.6509754061698914,
"learning_rate": 0.00027631893140836293,
"loss": 1.9895006179809571,
"step": 14330
},
{
"epoch": 1.835456,
"grad_norm": 0.636300265789032,
"learning_rate": 0.0002762863111174862,
"loss": 1.9789936065673828,
"step": 14340
},
{
"epoch": 1.8367360000000001,
"grad_norm": 0.6384751200675964,
"learning_rate": 0.0002762536703033345,
"loss": 2.0002975463867188,
"step": 14350
},
{
"epoch": 1.838016,
"grad_norm": 0.5570773482322693,
"learning_rate": 0.00027622100897121233,
"loss": 1.960428237915039,
"step": 14360
},
{
"epoch": 1.839296,
"grad_norm": 0.7164232730865479,
"learning_rate": 0.0002761883271264277,
"loss": 1.9880035400390625,
"step": 14370
},
{
"epoch": 1.840576,
"grad_norm": 0.5662538409233093,
"learning_rate": 0.0002761556247742918,
"loss": 1.9996320724487304,
"step": 14380
},
{
"epoch": 1.841856,
"grad_norm": 0.5953733921051025,
"learning_rate": 0.0002761229019201193,
"loss": 1.9869873046875,
"step": 14390
},
{
"epoch": 1.8431359999999999,
"grad_norm": 0.6056095361709595,
"learning_rate": 0.00027609015856922795,
"loss": 1.981184959411621,
"step": 14400
},
{
"epoch": 1.844416,
"grad_norm": 0.5695070624351501,
"learning_rate": 0.0002760573947269392,
"loss": 1.9867027282714844,
"step": 14410
},
{
"epoch": 1.845696,
"grad_norm": 0.6501023173332214,
"learning_rate": 0.0002760246103985775,
"loss": 1.9639600753784179,
"step": 14420
},
{
"epoch": 1.846976,
"grad_norm": 0.5985293984413147,
"learning_rate": 0.00027599180558947074,
"loss": 1.9643999099731446,
"step": 14430
},
{
"epoch": 1.8482560000000001,
"grad_norm": 0.7530742883682251,
"learning_rate": 0.00027595898030495026,
"loss": 1.990458869934082,
"step": 14440
},
{
"epoch": 1.849536,
"grad_norm": 0.6245275735855103,
"learning_rate": 0.0002759261345503506,
"loss": 1.9790266036987305,
"step": 14450
},
{
"epoch": 1.850816,
"grad_norm": 0.563879132270813,
"learning_rate": 0.0002758932683310095,
"loss": 1.9952646255493165,
"step": 14460
},
{
"epoch": 1.852096,
"grad_norm": 0.6672703623771667,
"learning_rate": 0.00027586038165226844,
"loss": 1.996124267578125,
"step": 14470
},
{
"epoch": 1.853376,
"grad_norm": 0.5894368290901184,
"learning_rate": 0.00027582747451947176,
"loss": 2.0115827560424804,
"step": 14480
},
{
"epoch": 1.8546559999999999,
"grad_norm": 0.6556432843208313,
"learning_rate": 0.0002757945469379673,
"loss": 2.0157970428466796,
"step": 14490
},
{
"epoch": 1.855936,
"grad_norm": 0.5600073337554932,
"learning_rate": 0.0002757615989131064,
"loss": 1.9951839447021484,
"step": 14500
},
{
"epoch": 1.857216,
"grad_norm": 0.5757896304130554,
"learning_rate": 0.00027572863045024354,
"loss": 1.993458366394043,
"step": 14510
},
{
"epoch": 1.8584960000000001,
"grad_norm": 1.796272873878479,
"learning_rate": 0.0002756956415547365,
"loss": 1.987896728515625,
"step": 14520
},
{
"epoch": 1.859776,
"grad_norm": 0.5895299911499023,
"learning_rate": 0.0002756626322319464,
"loss": 1.9889129638671874,
"step": 14530
},
{
"epoch": 1.861056,
"grad_norm": 0.5607683658599854,
"learning_rate": 0.0002756296024872378,
"loss": 1.976766014099121,
"step": 14540
},
{
"epoch": 1.862336,
"grad_norm": 0.5705486536026001,
"learning_rate": 0.0002755965523259784,
"loss": 2.0038755416870115,
"step": 14550
},
{
"epoch": 1.863616,
"grad_norm": 0.6873815059661865,
"learning_rate": 0.0002755634817535394,
"loss": 2.0010583877563475,
"step": 14560
},
{
"epoch": 1.8648959999999999,
"grad_norm": 0.6385495066642761,
"learning_rate": 0.0002755303907752951,
"loss": 2.004182815551758,
"step": 14570
},
{
"epoch": 1.8661759999999998,
"grad_norm": 0.594531774520874,
"learning_rate": 0.00027549727939662344,
"loss": 1.9861839294433594,
"step": 14580
},
{
"epoch": 1.867456,
"grad_norm": 0.658338725566864,
"learning_rate": 0.00027546414762290533,
"loss": 2.0028461456298827,
"step": 14590
},
{
"epoch": 1.868736,
"grad_norm": 0.6874401569366455,
"learning_rate": 0.00027543099545952515,
"loss": 2.0001737594604494,
"step": 14600
},
{
"epoch": 1.8700160000000001,
"grad_norm": 0.6376484036445618,
"learning_rate": 0.00027539782291187057,
"loss": 1.978806495666504,
"step": 14610
},
{
"epoch": 1.871296,
"grad_norm": 0.5759891271591187,
"learning_rate": 0.00027536462998533273,
"loss": 1.9915178298950196,
"step": 14620
},
{
"epoch": 1.872576,
"grad_norm": 0.5909021496772766,
"learning_rate": 0.00027533141668530575,
"loss": 1.998671531677246,
"step": 14630
},
{
"epoch": 1.873856,
"grad_norm": 0.6341341137886047,
"learning_rate": 0.0002752981830171874,
"loss": 1.986322784423828,
"step": 14640
},
{
"epoch": 1.875136,
"grad_norm": 0.6043877601623535,
"learning_rate": 0.00027526492898637855,
"loss": 1.9800695419311523,
"step": 14650
},
{
"epoch": 1.8764159999999999,
"grad_norm": 0.6429261565208435,
"learning_rate": 0.0002752316545982834,
"loss": 1.988995361328125,
"step": 14660
},
{
"epoch": 1.877696,
"grad_norm": 0.5567938089370728,
"learning_rate": 0.0002751983598583096,
"loss": 1.981565284729004,
"step": 14670
},
{
"epoch": 1.878976,
"grad_norm": 0.6165884733200073,
"learning_rate": 0.00027516504477186787,
"loss": 1.9861871719360351,
"step": 14680
},
{
"epoch": 1.8802560000000001,
"grad_norm": 0.6105359792709351,
"learning_rate": 0.00027513170934437254,
"loss": 1.9892169952392578,
"step": 14690
},
{
"epoch": 1.881536,
"grad_norm": 0.5695372223854065,
"learning_rate": 0.00027509835358124093,
"loss": 2.0050533294677733,
"step": 14700
},
{
"epoch": 1.882816,
"grad_norm": 0.6628261208534241,
"learning_rate": 0.0002750649774878939,
"loss": 1.99969482421875,
"step": 14710
},
{
"epoch": 1.884096,
"grad_norm": 0.6528040170669556,
"learning_rate": 0.00027503158106975553,
"loss": 1.9953853607177734,
"step": 14720
},
{
"epoch": 1.885376,
"grad_norm": 0.5966925621032715,
"learning_rate": 0.00027499816433225314,
"loss": 1.9600650787353515,
"step": 14730
},
{
"epoch": 1.8866559999999999,
"grad_norm": 0.6320626139640808,
"learning_rate": 0.00027496472728081743,
"loss": 1.9932964324951172,
"step": 14740
},
{
"epoch": 1.8879359999999998,
"grad_norm": 0.5818320512771606,
"learning_rate": 0.0002749312699208824,
"loss": 2.0017929077148438,
"step": 14750
},
{
"epoch": 1.889216,
"grad_norm": 0.5997017025947571,
"learning_rate": 0.0002748977922578854,
"loss": 1.9842348098754883,
"step": 14760
},
{
"epoch": 1.890496,
"grad_norm": 0.5786566138267517,
"learning_rate": 0.0002748642942972669,
"loss": 1.9812992095947266,
"step": 14770
},
{
"epoch": 1.8917760000000001,
"grad_norm": 0.5843164324760437,
"learning_rate": 0.0002748307760444708,
"loss": 1.9856365203857422,
"step": 14780
},
{
"epoch": 1.893056,
"grad_norm": 0.6202820539474487,
"learning_rate": 0.00027479723750494433,
"loss": 2.0052860260009764,
"step": 14790
},
{
"epoch": 1.894336,
"grad_norm": 0.6694955825805664,
"learning_rate": 0.000274763678684138,
"loss": 1.974207878112793,
"step": 14800
},
{
"epoch": 1.895616,
"grad_norm": 0.5736132860183716,
"learning_rate": 0.00027473009958750545,
"loss": 1.9973283767700196,
"step": 14810
},
{
"epoch": 1.896896,
"grad_norm": 0.557710587978363,
"learning_rate": 0.0002746965002205039,
"loss": 1.9818191528320312,
"step": 14820
},
{
"epoch": 1.8981759999999999,
"grad_norm": 0.8220569491386414,
"learning_rate": 0.0002746628805885936,
"loss": 1.9967378616333007,
"step": 14830
},
{
"epoch": 1.899456,
"grad_norm": 0.6114009618759155,
"learning_rate": 0.00027462924069723827,
"loss": 1.9965011596679687,
"step": 14840
},
{
"epoch": 1.900736,
"grad_norm": 0.6124056577682495,
"learning_rate": 0.0002745955805519048,
"loss": 2.016282081604004,
"step": 14850
},
{
"epoch": 1.9020160000000002,
"grad_norm": 0.5910442471504211,
"learning_rate": 0.00027456190015806353,
"loss": 1.9879514694213867,
"step": 14860
},
{
"epoch": 1.903296,
"grad_norm": 1.2884783744812012,
"learning_rate": 0.0002745281995211879,
"loss": 2.015022850036621,
"step": 14870
},
{
"epoch": 1.904576,
"grad_norm": 0.5896376371383667,
"learning_rate": 0.00027449447864675474,
"loss": 1.9929832458496093,
"step": 14880
},
{
"epoch": 1.905856,
"grad_norm": 1.4045612812042236,
"learning_rate": 0.0002744607375402442,
"loss": 2.008869934082031,
"step": 14890
},
{
"epoch": 1.907136,
"grad_norm": 1.1903825998306274,
"learning_rate": 0.0002744269762071396,
"loss": 2.0193687438964845,
"step": 14900
},
{
"epoch": 1.908416,
"grad_norm": 0.6184222102165222,
"learning_rate": 0.00027439319465292773,
"loss": 2.0176204681396483,
"step": 14910
},
{
"epoch": 1.9096959999999998,
"grad_norm": 0.6446005702018738,
"learning_rate": 0.0002743593928830985,
"loss": 1.9901769638061524,
"step": 14920
},
{
"epoch": 1.910976,
"grad_norm": 0.5989486575126648,
"learning_rate": 0.0002743255709031452,
"loss": 2.00903263092041,
"step": 14930
},
{
"epoch": 1.912256,
"grad_norm": 0.8651177287101746,
"learning_rate": 0.0002742917287185643,
"loss": 1.9887771606445312,
"step": 14940
},
{
"epoch": 1.9135360000000001,
"grad_norm": 0.5811720490455627,
"learning_rate": 0.00027425786633485566,
"loss": 1.9947555541992188,
"step": 14950
},
{
"epoch": 1.914816,
"grad_norm": 0.6343239545822144,
"learning_rate": 0.00027422398375752247,
"loss": 1.9907865524291992,
"step": 14960
},
{
"epoch": 1.916096,
"grad_norm": 0.6845805048942566,
"learning_rate": 0.00027419008099207096,
"loss": 1.9792387008666992,
"step": 14970
},
{
"epoch": 1.917376,
"grad_norm": 0.6633698344230652,
"learning_rate": 0.00027415615804401094,
"loss": 1.9987592697143555,
"step": 14980
},
{
"epoch": 1.918656,
"grad_norm": 0.5590615272521973,
"learning_rate": 0.0002741222149188553,
"loss": 1.9844167709350586,
"step": 14990
},
{
"epoch": 1.9199359999999999,
"grad_norm": 0.6449314951896667,
"learning_rate": 0.00027408825162212024,
"loss": 1.9796030044555664,
"step": 15000
},
{
"epoch": 1.921216,
"grad_norm": 0.5761886835098267,
"learning_rate": 0.0002740542681593253,
"loss": 2.000011444091797,
"step": 15010
},
{
"epoch": 1.922496,
"grad_norm": 0.6057732701301575,
"learning_rate": 0.0002740202645359933,
"loss": 1.992666244506836,
"step": 15020
},
{
"epoch": 1.9237760000000002,
"grad_norm": 0.6328257322311401,
"learning_rate": 0.0002739862407576502,
"loss": 1.9957653045654298,
"step": 15030
},
{
"epoch": 1.925056,
"grad_norm": 0.5427049398422241,
"learning_rate": 0.0002739521968298254,
"loss": 2.001283645629883,
"step": 15040
},
{
"epoch": 1.926336,
"grad_norm": 0.5683019161224365,
"learning_rate": 0.0002739181327580515,
"loss": 1.97943058013916,
"step": 15050
},
{
"epoch": 1.927616,
"grad_norm": 0.8532147407531738,
"learning_rate": 0.00027388404854786447,
"loss": 1.9755849838256836,
"step": 15060
},
{
"epoch": 1.928896,
"grad_norm": 0.624137818813324,
"learning_rate": 0.0002738499442048033,
"loss": 1.9904840469360352,
"step": 15070
},
{
"epoch": 1.930176,
"grad_norm": 0.7412115931510925,
"learning_rate": 0.0002738158197344105,
"loss": 1.9809625625610352,
"step": 15080
},
{
"epoch": 1.9314559999999998,
"grad_norm": 0.7491364479064941,
"learning_rate": 0.00027378167514223185,
"loss": 1.9927919387817383,
"step": 15090
},
{
"epoch": 1.932736,
"grad_norm": 0.6788904666900635,
"learning_rate": 0.0002737475104338162,
"loss": 1.9927122116088867,
"step": 15100
},
{
"epoch": 1.934016,
"grad_norm": 0.5447447299957275,
"learning_rate": 0.00027371332561471583,
"loss": 1.9873960494995118,
"step": 15110
},
{
"epoch": 1.9352960000000001,
"grad_norm": 0.5972921252250671,
"learning_rate": 0.00027367912069048623,
"loss": 2.003441047668457,
"step": 15120
},
{
"epoch": 1.936576,
"grad_norm": 0.7013078331947327,
"learning_rate": 0.0002736448956666862,
"loss": 1.9882572174072266,
"step": 15130
},
{
"epoch": 1.937856,
"grad_norm": 0.6605074405670166,
"learning_rate": 0.0002736106505488777,
"loss": 1.9947832107543946,
"step": 15140
},
{
"epoch": 1.939136,
"grad_norm": 0.624704897403717,
"learning_rate": 0.00027357638534262623,
"loss": 1.985148048400879,
"step": 15150
},
{
"epoch": 1.940416,
"grad_norm": 0.6804410815238953,
"learning_rate": 0.00027354210005350014,
"loss": 1.9888923645019532,
"step": 15160
},
{
"epoch": 1.9416959999999999,
"grad_norm": 0.678081214427948,
"learning_rate": 0.00027350779468707145,
"loss": 1.9940563201904298,
"step": 15170
},
{
"epoch": 1.942976,
"grad_norm": 0.7491250038146973,
"learning_rate": 0.0002734734692489151,
"loss": 1.9851743698120117,
"step": 15180
},
{
"epoch": 1.944256,
"grad_norm": 0.7810075283050537,
"learning_rate": 0.0002734391237446095,
"loss": 1.9839576721191405,
"step": 15190
},
{
"epoch": 1.9455360000000002,
"grad_norm": 0.6328175663948059,
"learning_rate": 0.0002734047581797364,
"loss": 1.9939220428466797,
"step": 15200
},
{
"epoch": 1.946816,
"grad_norm": 0.5270907282829285,
"learning_rate": 0.00027337037255988045,
"loss": 2.000494956970215,
"step": 15210
},
{
"epoch": 1.948096,
"grad_norm": 0.6894044280052185,
"learning_rate": 0.00027333596689063,
"loss": 2.0118940353393553,
"step": 15220
},
{
"epoch": 1.949376,
"grad_norm": 0.7083652019500732,
"learning_rate": 0.00027330154117757627,
"loss": 2.0064647674560545,
"step": 15230
},
{
"epoch": 1.950656,
"grad_norm": 0.635725200176239,
"learning_rate": 0.00027326709542631404,
"loss": 1.9808626174926758,
"step": 15240
},
{
"epoch": 1.951936,
"grad_norm": 0.7094610333442688,
"learning_rate": 0.0002732326296424412,
"loss": 2.0019426345825195,
"step": 15250
},
{
"epoch": 1.9532159999999998,
"grad_norm": 0.5832314491271973,
"learning_rate": 0.00027319814383155885,
"loss": 1.9864368438720703,
"step": 15260
},
{
"epoch": 1.954496,
"grad_norm": 0.5181657671928406,
"learning_rate": 0.0002731636379992715,
"loss": 1.9723577499389648,
"step": 15270
},
{
"epoch": 1.955776,
"grad_norm": 0.7585449814796448,
"learning_rate": 0.0002731291121511868,
"loss": 2.0052982330322267,
"step": 15280
},
{
"epoch": 1.9570560000000001,
"grad_norm": 0.5650632381439209,
"learning_rate": 0.00027309456629291565,
"loss": 2.011658477783203,
"step": 15290
},
{
"epoch": 1.958336,
"grad_norm": 0.6076073050498962,
"learning_rate": 0.00027306000043007226,
"loss": 1.981151008605957,
"step": 15300
},
{
"epoch": 1.959616,
"grad_norm": 0.6412877440452576,
"learning_rate": 0.000273025414568274,
"loss": 1.9722415924072265,
"step": 15310
},
{
"epoch": 1.960896,
"grad_norm": 0.6715180277824402,
"learning_rate": 0.0002729908087131417,
"loss": 1.9752307891845704,
"step": 15320
},
{
"epoch": 1.962176,
"grad_norm": 0.6363363265991211,
"learning_rate": 0.0002729561828702991,
"loss": 1.9887128829956056,
"step": 15330
},
{
"epoch": 1.9634559999999999,
"grad_norm": 0.5581247210502625,
"learning_rate": 0.00027292153704537353,
"loss": 2.000994110107422,
"step": 15340
},
{
"epoch": 1.964736,
"grad_norm": 0.5792102813720703,
"learning_rate": 0.0002728868712439953,
"loss": 1.9863534927368165,
"step": 15350
},
{
"epoch": 1.966016,
"grad_norm": 0.6057248115539551,
"learning_rate": 0.00027285218547179815,
"loss": 1.9823060989379884,
"step": 15360
},
{
"epoch": 1.9672960000000002,
"grad_norm": 0.6092322468757629,
"learning_rate": 0.00027281747973441903,
"loss": 1.9872922897338867,
"step": 15370
},
{
"epoch": 1.968576,
"grad_norm": 0.5774412751197815,
"learning_rate": 0.0002727827540374981,
"loss": 1.988485336303711,
"step": 15380
},
{
"epoch": 1.969856,
"grad_norm": 0.5677629709243774,
"learning_rate": 0.0002727480083866786,
"loss": 1.998712921142578,
"step": 15390
},
{
"epoch": 1.971136,
"grad_norm": 0.5547299385070801,
"learning_rate": 0.0002727132427876074,
"loss": 1.992000198364258,
"step": 15400
},
{
"epoch": 1.972416,
"grad_norm": 0.5611027479171753,
"learning_rate": 0.0002726784572459343,
"loss": 1.9901214599609376,
"step": 15410
},
{
"epoch": 1.973696,
"grad_norm": 0.6831710338592529,
"learning_rate": 0.0002726436517673124,
"loss": 2.0092220306396484,
"step": 15420
},
{
"epoch": 1.9749759999999998,
"grad_norm": 0.652474045753479,
"learning_rate": 0.00027260882635739807,
"loss": 1.9941566467285157,
"step": 15430
},
{
"epoch": 1.976256,
"grad_norm": 0.6066447496414185,
"learning_rate": 0.000272573981021851,
"loss": 1.9903448104858399,
"step": 15440
},
{
"epoch": 1.977536,
"grad_norm": 0.5975965261459351,
"learning_rate": 0.000272539115766334,
"loss": 1.9858333587646484,
"step": 15450
},
{
"epoch": 1.9788160000000001,
"grad_norm": 0.6692200899124146,
"learning_rate": 0.0002725042305965132,
"loss": 1.9625078201293946,
"step": 15460
},
{
"epoch": 1.980096,
"grad_norm": 0.6145694255828857,
"learning_rate": 0.0002724693255180578,
"loss": 1.987001609802246,
"step": 15470
},
{
"epoch": 1.981376,
"grad_norm": 0.6061978936195374,
"learning_rate": 0.0002724344005366404,
"loss": 1.9908834457397462,
"step": 15480
},
{
"epoch": 1.982656,
"grad_norm": 0.6346415281295776,
"learning_rate": 0.00027239945565793694,
"loss": 1.9886537551879884,
"step": 15490
},
{
"epoch": 1.983936,
"grad_norm": 0.531440019607544,
"learning_rate": 0.00027236449088762627,
"loss": 1.9866472244262696,
"step": 15500
},
{
"epoch": 1.9852159999999999,
"grad_norm": 0.6538944244384766,
"learning_rate": 0.00027232950623139076,
"loss": 1.9970335006713866,
"step": 15510
},
{
"epoch": 1.986496,
"grad_norm": 0.6200321316719055,
"learning_rate": 0.00027229450169491583,
"loss": 1.9915782928466796,
"step": 15520
},
{
"epoch": 1.987776,
"grad_norm": 0.5558231472969055,
"learning_rate": 0.00027225947728389015,
"loss": 1.9931659698486328,
"step": 15530
},
{
"epoch": 1.9890560000000002,
"grad_norm": 0.5900181531906128,
"learning_rate": 0.00027222443300400586,
"loss": 1.9951946258544921,
"step": 15540
},
{
"epoch": 1.990336,
"grad_norm": 0.6925880908966064,
"learning_rate": 0.000272189368860958,
"loss": 2.009657287597656,
"step": 15550
},
{
"epoch": 1.991616,
"grad_norm": 0.7228490710258484,
"learning_rate": 0.00027215428486044495,
"loss": 1.9677371978759766,
"step": 15560
},
{
"epoch": 1.992896,
"grad_norm": 0.6594810485839844,
"learning_rate": 0.0002721191810081685,
"loss": 1.9823518753051759,
"step": 15570
},
{
"epoch": 1.994176,
"grad_norm": 0.6771001815795898,
"learning_rate": 0.0002720840573098334,
"loss": 1.990148162841797,
"step": 15580
},
{
"epoch": 1.995456,
"grad_norm": 0.6207253932952881,
"learning_rate": 0.0002720489137711477,
"loss": 2.003261756896973,
"step": 15590
},
{
"epoch": 1.9967359999999998,
"grad_norm": 0.5417797565460205,
"learning_rate": 0.00027201375039782275,
"loss": 1.9974004745483398,
"step": 15600
},
{
"epoch": 1.998016,
"grad_norm": 0.6458831429481506,
"learning_rate": 0.0002719785671955732,
"loss": 1.9819438934326172,
"step": 15610
},
{
"epoch": 1.999296,
"grad_norm": 0.5755211114883423,
"learning_rate": 0.00027194336417011666,
"loss": 1.9855327606201172,
"step": 15620
},
{
"epoch": 2.0,
"eval_loss": 1.9955754280090332,
"eval_runtime": 18.0366,
"eval_samples_per_second": 554.427,
"eval_steps_per_second": 69.303,
"step": 15626
},
{
"epoch": 2.000512,
"grad_norm": 0.7575382590293884,
"learning_rate": 0.00027190814132717415,
"loss": 1.9999902725219727,
"step": 15630
},
{
"epoch": 2.001792,
"grad_norm": 0.5760068297386169,
"learning_rate": 0.00027187289867246987,
"loss": 1.9975374221801758,
"step": 15640
},
{
"epoch": 2.003072,
"grad_norm": 0.6450005173683167,
"learning_rate": 0.0002718376362117313,
"loss": 1.9883235931396483,
"step": 15650
},
{
"epoch": 2.004352,
"grad_norm": 0.5935179591178894,
"learning_rate": 0.000271802353950689,
"loss": 1.9953784942626953,
"step": 15660
},
{
"epoch": 2.005632,
"grad_norm": 0.639032244682312,
"learning_rate": 0.00027176705189507685,
"loss": 1.9881561279296875,
"step": 15670
},
{
"epoch": 2.006912,
"grad_norm": 0.6289457678794861,
"learning_rate": 0.0002717317300506319,
"loss": 1.9996967315673828,
"step": 15680
},
{
"epoch": 2.008192,
"grad_norm": 0.5696092844009399,
"learning_rate": 0.0002716963884230945,
"loss": 1.9919151306152343,
"step": 15690
},
{
"epoch": 2.009472,
"grad_norm": 0.6254428625106812,
"learning_rate": 0.0002716610270182081,
"loss": 1.976341438293457,
"step": 15700
},
{
"epoch": 2.010752,
"grad_norm": 0.6007362008094788,
"learning_rate": 0.00027162564584171946,
"loss": 1.971974754333496,
"step": 15710
},
{
"epoch": 2.012032,
"grad_norm": 0.7049063444137573,
"learning_rate": 0.0002715902448993785,
"loss": 1.9820518493652344,
"step": 15720
},
{
"epoch": 2.013312,
"grad_norm": 0.5854899287223816,
"learning_rate": 0.0002715548241969383,
"loss": 1.9993438720703125,
"step": 15730
},
{
"epoch": 2.014592,
"grad_norm": 0.6590539216995239,
"learning_rate": 0.00027151938374015533,
"loss": 1.9982528686523438,
"step": 15740
},
{
"epoch": 2.015872,
"grad_norm": 0.6146329045295715,
"learning_rate": 0.00027148392353478905,
"loss": 1.9886005401611329,
"step": 15750
},
{
"epoch": 2.017152,
"grad_norm": 0.5906791090965271,
"learning_rate": 0.0002714484435866023,
"loss": 1.9895116806030273,
"step": 15760
},
{
"epoch": 2.018432,
"grad_norm": 0.6036437153816223,
"learning_rate": 0.00027141294390136105,
"loss": 1.9861625671386718,
"step": 15770
},
{
"epoch": 2.019712,
"grad_norm": 0.6529788374900818,
"learning_rate": 0.00027137742448483444,
"loss": 1.9986656188964844,
"step": 15780
},
{
"epoch": 2.020992,
"grad_norm": 0.6340635418891907,
"learning_rate": 0.00027134188534279493,
"loss": 1.9745838165283203,
"step": 15790
},
{
"epoch": 2.022272,
"grad_norm": 0.7375162243843079,
"learning_rate": 0.00027130632648101813,
"loss": 1.9868717193603516,
"step": 15800
},
{
"epoch": 2.023552,
"grad_norm": 0.6215420365333557,
"learning_rate": 0.00027127074790528284,
"loss": 1.978240966796875,
"step": 15810
},
{
"epoch": 2.024832,
"grad_norm": 0.7992658019065857,
"learning_rate": 0.000271235149621371,
"loss": 2.0204965591430666,
"step": 15820
},
{
"epoch": 2.026112,
"grad_norm": 0.5523297786712646,
"learning_rate": 0.00027119953163506785,
"loss": 1.9970949172973633,
"step": 15830
},
{
"epoch": 2.027392,
"grad_norm": 0.5994831919670105,
"learning_rate": 0.00027116389395216185,
"loss": 2.000766372680664,
"step": 15840
},
{
"epoch": 2.028672,
"grad_norm": 0.553413987159729,
"learning_rate": 0.00027112823657844464,
"loss": 1.9850357055664063,
"step": 15850
},
{
"epoch": 2.029952,
"grad_norm": 0.5261474847793579,
"learning_rate": 0.00027109255951971096,
"loss": 1.9862333297729493,
"step": 15860
},
{
"epoch": 2.031232,
"grad_norm": 0.8071524500846863,
"learning_rate": 0.0002710568627817589,
"loss": 1.9902618408203125,
"step": 15870
},
{
"epoch": 2.032512,
"grad_norm": 0.6147076487541199,
"learning_rate": 0.0002710211463703896,
"loss": 1.982151985168457,
"step": 15880
},
{
"epoch": 2.033792,
"grad_norm": 0.6551554203033447,
"learning_rate": 0.0002709854102914075,
"loss": 1.9985364913940429,
"step": 15890
},
{
"epoch": 2.035072,
"grad_norm": 0.6600733399391174,
"learning_rate": 0.00027094965455062023,
"loss": 1.988072967529297,
"step": 15900
},
{
"epoch": 2.036352,
"grad_norm": 0.592311441898346,
"learning_rate": 0.0002709138791538386,
"loss": 1.9871063232421875,
"step": 15910
},
{
"epoch": 2.037632,
"grad_norm": 0.5898208618164062,
"learning_rate": 0.0002708780841068766,
"loss": 1.9862031936645508,
"step": 15920
},
{
"epoch": 2.038912,
"grad_norm": 0.6081909537315369,
"learning_rate": 0.0002708422694155513,
"loss": 1.9805521011352538,
"step": 15930
},
{
"epoch": 2.040192,
"grad_norm": 0.6301562786102295,
"learning_rate": 0.0002708064350856833,
"loss": 1.9769163131713867,
"step": 15940
},
{
"epoch": 2.041472,
"grad_norm": 0.6689660549163818,
"learning_rate": 0.000270770581123096,
"loss": 1.9893808364868164,
"step": 15950
},
{
"epoch": 2.042752,
"grad_norm": 0.5094631314277649,
"learning_rate": 0.0002707347075336163,
"loss": 1.9901708602905273,
"step": 15960
},
{
"epoch": 2.044032,
"grad_norm": 0.6027694344520569,
"learning_rate": 0.000270698814323074,
"loss": 1.987225341796875,
"step": 15970
},
{
"epoch": 2.045312,
"grad_norm": 0.6419055461883545,
"learning_rate": 0.00027066290149730244,
"loss": 1.9939151763916017,
"step": 15980
},
{
"epoch": 2.046592,
"grad_norm": 0.6224138140678406,
"learning_rate": 0.00027062696906213777,
"loss": 1.9852516174316406,
"step": 15990
},
{
"epoch": 2.047872,
"grad_norm": 0.6197025179862976,
"learning_rate": 0.00027059101702341955,
"loss": 2.0067655563354494,
"step": 16000
},
{
"epoch": 2.049152,
"grad_norm": 0.5472074747085571,
"learning_rate": 0.0002705550453869906,
"loss": 1.981334114074707,
"step": 16010
},
{
"epoch": 2.050432,
"grad_norm": 0.6448740363121033,
"learning_rate": 0.00027051905415869667,
"loss": 1.9770565032958984,
"step": 16020
},
{
"epoch": 2.051712,
"grad_norm": 0.6080293655395508,
"learning_rate": 0.0002704830433443869,
"loss": 1.9848661422729492,
"step": 16030
},
{
"epoch": 2.052992,
"grad_norm": 0.6401488184928894,
"learning_rate": 0.00027044701294991354,
"loss": 2.008693885803223,
"step": 16040
},
{
"epoch": 2.054272,
"grad_norm": 0.5457377433776855,
"learning_rate": 0.000270410962981132,
"loss": 1.9782979965209961,
"step": 16050
},
{
"epoch": 2.055552,
"grad_norm": 0.5170462131500244,
"learning_rate": 0.000270374893443901,
"loss": 1.9964645385742188,
"step": 16060
},
{
"epoch": 2.056832,
"grad_norm": 0.6594231724739075,
"learning_rate": 0.0002703388043440822,
"loss": 2.0046707153320313,
"step": 16070
},
{
"epoch": 2.058112,
"grad_norm": 0.628869891166687,
"learning_rate": 0.0002703026956875406,
"loss": 1.9893329620361329,
"step": 16080
},
{
"epoch": 2.059392,
"grad_norm": 0.6653295755386353,
"learning_rate": 0.00027026656748014444,
"loss": 1.9940963745117188,
"step": 16090
},
{
"epoch": 2.060672,
"grad_norm": 0.5196301937103271,
"learning_rate": 0.000270230419727765,
"loss": 1.9959924697875977,
"step": 16100
},
{
"epoch": 2.061952,
"grad_norm": 0.6060832738876343,
"learning_rate": 0.0002701942524362768,
"loss": 1.9904598236083983,
"step": 16110
},
{
"epoch": 2.063232,
"grad_norm": 0.6250558495521545,
"learning_rate": 0.0002701580656115576,
"loss": 2.0013706207275392,
"step": 16120
},
{
"epoch": 2.064512,
"grad_norm": 0.5976873636245728,
"learning_rate": 0.0002701218592594881,
"loss": 1.9950780868530273,
"step": 16130
},
{
"epoch": 2.065792,
"grad_norm": 0.6311057806015015,
"learning_rate": 0.00027008563338595245,
"loss": 1.99168701171875,
"step": 16140
},
{
"epoch": 2.067072,
"grad_norm": 0.6298433542251587,
"learning_rate": 0.0002700493879968378,
"loss": 1.9721656799316407,
"step": 16150
},
{
"epoch": 2.068352,
"grad_norm": 0.558314323425293,
"learning_rate": 0.0002700131230980346,
"loss": 1.9936748504638673,
"step": 16160
},
{
"epoch": 2.069632,
"grad_norm": 0.5926997661590576,
"learning_rate": 0.0002699768386954363,
"loss": 1.9939022064208984,
"step": 16170
},
{
"epoch": 2.070912,
"grad_norm": 0.5976966619491577,
"learning_rate": 0.0002699405347949397,
"loss": 1.98126163482666,
"step": 16180
},
{
"epoch": 2.072192,
"grad_norm": 0.5566416382789612,
"learning_rate": 0.00026990421140244466,
"loss": 1.9798095703125,
"step": 16190
},
{
"epoch": 2.073472,
"grad_norm": 0.6109269261360168,
"learning_rate": 0.00026986786852385426,
"loss": 2.0042726516723635,
"step": 16200
},
{
"epoch": 2.074752,
"grad_norm": 0.6241195201873779,
"learning_rate": 0.00026983150616507464,
"loss": 2.007033920288086,
"step": 16210
},
{
"epoch": 2.076032,
"grad_norm": 0.6290252804756165,
"learning_rate": 0.0002697951243320153,
"loss": 1.986985206604004,
"step": 16220
},
{
"epoch": 2.077312,
"grad_norm": 0.6204255223274231,
"learning_rate": 0.0002697587230305887,
"loss": 1.988245391845703,
"step": 16230
},
{
"epoch": 2.078592,
"grad_norm": 0.6347700953483582,
"learning_rate": 0.0002697223022667106,
"loss": 1.9791933059692384,
"step": 16240
},
{
"epoch": 2.079872,
"grad_norm": 0.6098595857620239,
"learning_rate": 0.0002696858620462999,
"loss": 1.9958024978637696,
"step": 16250
},
{
"epoch": 2.081152,
"grad_norm": 0.6507900357246399,
"learning_rate": 0.00026964940237527856,
"loss": 1.9857753753662108,
"step": 16260
},
{
"epoch": 2.082432,
"grad_norm": 0.7905516028404236,
"learning_rate": 0.0002696129232595718,
"loss": 1.9972883224487306,
"step": 16270
},
{
"epoch": 2.083712,
"grad_norm": 0.5942310690879822,
"learning_rate": 0.00026957642470510805,
"loss": 2.0049812316894533,
"step": 16280
},
{
"epoch": 2.084992,
"grad_norm": 0.7992523312568665,
"learning_rate": 0.00026953990671781884,
"loss": 1.9872356414794923,
"step": 16290
},
{
"epoch": 2.086272,
"grad_norm": 0.603329598903656,
"learning_rate": 0.0002695033693036387,
"loss": 1.9872465133666992,
"step": 16300
},
{
"epoch": 2.087552,
"grad_norm": 0.6421152949333191,
"learning_rate": 0.0002694668124685056,
"loss": 1.9916072845458985,
"step": 16310
},
{
"epoch": 2.088832,
"grad_norm": 0.6409503221511841,
"learning_rate": 0.00026943023621836055,
"loss": 1.9827692031860351,
"step": 16320
},
{
"epoch": 2.090112,
"grad_norm": 0.5767059922218323,
"learning_rate": 0.00026939364055914765,
"loss": 2.0160579681396484,
"step": 16330
},
{
"epoch": 2.091392,
"grad_norm": 0.6523997783660889,
"learning_rate": 0.00026935702549681415,
"loss": 1.997859573364258,
"step": 16340
},
{
"epoch": 2.092672,
"grad_norm": 0.6683000922203064,
"learning_rate": 0.0002693203910373106,
"loss": 1.993924331665039,
"step": 16350
},
{
"epoch": 2.093952,
"grad_norm": 0.6003236770629883,
"learning_rate": 0.00026928373718659046,
"loss": 2.013311767578125,
"step": 16360
},
{
"epoch": 2.095232,
"grad_norm": 0.7068638205528259,
"learning_rate": 0.00026924706395061067,
"loss": 2.0059343338012696,
"step": 16370
},
{
"epoch": 2.096512,
"grad_norm": 0.6155956387519836,
"learning_rate": 0.00026921037133533106,
"loss": 2.001392364501953,
"step": 16380
},
{
"epoch": 2.097792,
"grad_norm": 0.5663672685623169,
"learning_rate": 0.0002691736593467147,
"loss": 1.9822683334350586,
"step": 16390
},
{
"epoch": 2.099072,
"grad_norm": 0.6799958348274231,
"learning_rate": 0.0002691369279907277,
"loss": 1.975177001953125,
"step": 16400
},
{
"epoch": 2.100352,
"grad_norm": 0.5932645797729492,
"learning_rate": 0.0002691001772733396,
"loss": 1.984330940246582,
"step": 16410
},
{
"epoch": 2.101632,
"grad_norm": 0.5412223935127258,
"learning_rate": 0.0002690634072005227,
"loss": 1.9948488235473634,
"step": 16420
},
{
"epoch": 2.102912,
"grad_norm": 0.6713813543319702,
"learning_rate": 0.00026902661777825277,
"loss": 1.9833383560180664,
"step": 16430
},
{
"epoch": 2.104192,
"grad_norm": 0.7013354897499084,
"learning_rate": 0.00026898980901250865,
"loss": 1.9929718017578124,
"step": 16440
},
{
"epoch": 2.105472,
"grad_norm": 0.6265549659729004,
"learning_rate": 0.00026895298090927214,
"loss": 2.0051721572875976,
"step": 16450
},
{
"epoch": 2.106752,
"grad_norm": 0.5798888802528381,
"learning_rate": 0.0002689161334745283,
"loss": 1.994221305847168,
"step": 16460
},
{
"epoch": 2.108032,
"grad_norm": 0.6222522854804993,
"learning_rate": 0.00026887926671426555,
"loss": 1.9974424362182617,
"step": 16470
},
{
"epoch": 2.109312,
"grad_norm": 0.6265146136283875,
"learning_rate": 0.0002688423806344751,
"loss": 2.002239227294922,
"step": 16480
},
{
"epoch": 2.110592,
"grad_norm": 0.6058418154716492,
"learning_rate": 0.00026880547524115144,
"loss": 1.985641098022461,
"step": 16490
},
{
"epoch": 2.111872,
"grad_norm": 0.5120436549186707,
"learning_rate": 0.00026876855054029224,
"loss": 2.00388126373291,
"step": 16500
},
{
"epoch": 2.113152,
"grad_norm": 0.665870726108551,
"learning_rate": 0.00026873160653789833,
"loss": 1.9777925491333008,
"step": 16510
},
{
"epoch": 2.114432,
"grad_norm": 0.6199498176574707,
"learning_rate": 0.0002686946432399735,
"loss": 1.990678596496582,
"step": 16520
},
{
"epoch": 2.115712,
"grad_norm": 0.6568735241889954,
"learning_rate": 0.0002686576606525249,
"loss": 1.9832338333129882,
"step": 16530
},
{
"epoch": 2.116992,
"grad_norm": 0.6233317255973816,
"learning_rate": 0.0002686206587815627,
"loss": 1.9852951049804688,
"step": 16540
},
{
"epoch": 2.118272,
"grad_norm": 0.5914753675460815,
"learning_rate": 0.0002685836376331001,
"loss": 2.0060564041137696,
"step": 16550
},
{
"epoch": 2.119552,
"grad_norm": 0.590316653251648,
"learning_rate": 0.0002685465972131537,
"loss": 1.999317741394043,
"step": 16560
},
{
"epoch": 2.120832,
"grad_norm": 0.7263337969779968,
"learning_rate": 0.000268509537527743,
"loss": 1.974721336364746,
"step": 16570
},
{
"epoch": 2.122112,
"grad_norm": 0.5347442626953125,
"learning_rate": 0.0002684724585828908,
"loss": 1.981308937072754,
"step": 16580
},
{
"epoch": 2.123392,
"grad_norm": 0.5929756164550781,
"learning_rate": 0.0002684353603846229,
"loss": 2.009960174560547,
"step": 16590
},
{
"epoch": 2.124672,
"grad_norm": 0.709740400314331,
"learning_rate": 0.0002683982429389682,
"loss": 2.0006996154785157,
"step": 16600
},
{
"epoch": 2.125952,
"grad_norm": 0.7660552859306335,
"learning_rate": 0.00026836110625195884,
"loss": 1.990781021118164,
"step": 16610
},
{
"epoch": 2.1272320000000002,
"grad_norm": 0.6065070629119873,
"learning_rate": 0.0002683239503296301,
"loss": 1.9956268310546874,
"step": 16620
},
{
"epoch": 2.128512,
"grad_norm": 0.5575572848320007,
"learning_rate": 0.0002682867751780202,
"loss": 1.987329864501953,
"step": 16630
},
{
"epoch": 2.129792,
"grad_norm": 0.6174092888832092,
"learning_rate": 0.0002682495808031708,
"loss": 2.0006046295166016,
"step": 16640
},
{
"epoch": 2.131072,
"grad_norm": 0.586053729057312,
"learning_rate": 0.00026821236721112635,
"loss": 1.9746065139770508,
"step": 16650
},
{
"epoch": 2.132352,
"grad_norm": 0.5903230905532837,
"learning_rate": 0.00026817513440793465,
"loss": 2.0078052520751952,
"step": 16660
},
{
"epoch": 2.133632,
"grad_norm": 0.5094022750854492,
"learning_rate": 0.00026813788239964657,
"loss": 1.998386764526367,
"step": 16670
},
{
"epoch": 2.134912,
"grad_norm": 0.6100565195083618,
"learning_rate": 0.000268100611192316,
"loss": 1.9805191040039063,
"step": 16680
},
{
"epoch": 2.136192,
"grad_norm": 0.6118277311325073,
"learning_rate": 0.00026806332079200003,
"loss": 2.000060272216797,
"step": 16690
},
{
"epoch": 2.137472,
"grad_norm": 0.5466392636299133,
"learning_rate": 0.00026802601120475895,
"loss": 1.9962371826171874,
"step": 16700
},
{
"epoch": 2.138752,
"grad_norm": 0.6108764410018921,
"learning_rate": 0.000267988682436656,
"loss": 1.987067985534668,
"step": 16710
},
{
"epoch": 2.140032,
"grad_norm": 0.577490508556366,
"learning_rate": 0.00026795133449375767,
"loss": 1.9899030685424806,
"step": 16720
},
{
"epoch": 2.141312,
"grad_norm": 0.6618934869766235,
"learning_rate": 0.0002679139673821335,
"loss": 2.004457473754883,
"step": 16730
},
{
"epoch": 2.142592,
"grad_norm": 0.5606252551078796,
"learning_rate": 0.0002678765811078561,
"loss": 1.998589324951172,
"step": 16740
},
{
"epoch": 2.143872,
"grad_norm": 0.5495691895484924,
"learning_rate": 0.00026783917567700145,
"loss": 1.9854915618896485,
"step": 16750
},
{
"epoch": 2.145152,
"grad_norm": 0.685238778591156,
"learning_rate": 0.00026780175109564826,
"loss": 1.9930635452270509,
"step": 16760
},
{
"epoch": 2.146432,
"grad_norm": 0.6507768630981445,
"learning_rate": 0.00026776430736987866,
"loss": 1.9720846176147462,
"step": 16770
},
{
"epoch": 2.147712,
"grad_norm": 0.5852428674697876,
"learning_rate": 0.0002677268445057776,
"loss": 1.9960498809814453,
"step": 16780
},
{
"epoch": 2.148992,
"grad_norm": 0.5501282215118408,
"learning_rate": 0.00026768936250943354,
"loss": 1.98053035736084,
"step": 16790
},
{
"epoch": 2.150272,
"grad_norm": 0.5651091933250427,
"learning_rate": 0.00026765186138693773,
"loss": 1.9873149871826172,
"step": 16800
},
{
"epoch": 2.151552,
"grad_norm": 0.7594240307807922,
"learning_rate": 0.00026761434114438455,
"loss": 1.9697820663452148,
"step": 16810
},
{
"epoch": 2.152832,
"grad_norm": 0.5956317782402039,
"learning_rate": 0.00026757680178787167,
"loss": 1.9906524658203124,
"step": 16820
},
{
"epoch": 2.154112,
"grad_norm": 0.6509799361228943,
"learning_rate": 0.0002675392433234997,
"loss": 1.9799036026000976,
"step": 16830
},
{
"epoch": 2.155392,
"grad_norm": 0.6479411125183105,
"learning_rate": 0.0002675016657573724,
"loss": 2.0049615859985352,
"step": 16840
},
{
"epoch": 2.156672,
"grad_norm": 0.7550004720687866,
"learning_rate": 0.0002674640690955966,
"loss": 1.9783910751342773,
"step": 16850
},
{
"epoch": 2.157952,
"grad_norm": 0.5814908742904663,
"learning_rate": 0.0002674264533442825,
"loss": 1.986196517944336,
"step": 16860
},
{
"epoch": 2.159232,
"grad_norm": 0.6661468148231506,
"learning_rate": 0.0002673888185095429,
"loss": 1.9800241470336915,
"step": 16870
},
{
"epoch": 2.1605119999999998,
"grad_norm": 0.611038088798523,
"learning_rate": 0.00026735116459749414,
"loss": 1.9995229721069336,
"step": 16880
},
{
"epoch": 2.161792,
"grad_norm": 0.6498743891716003,
"learning_rate": 0.00026731349161425547,
"loss": 1.9997146606445313,
"step": 16890
},
{
"epoch": 2.163072,
"grad_norm": 0.5541350841522217,
"learning_rate": 0.0002672757995659493,
"loss": 1.981035041809082,
"step": 16900
},
{
"epoch": 2.164352,
"grad_norm": 0.6435770988464355,
"learning_rate": 0.000267238088458701,
"loss": 2.0012491226196287,
"step": 16910
},
{
"epoch": 2.165632,
"grad_norm": 0.6115776896476746,
"learning_rate": 0.0002672003582986393,
"loss": 2.002513313293457,
"step": 16920
},
{
"epoch": 2.166912,
"grad_norm": 0.6045442819595337,
"learning_rate": 0.0002671626090918957,
"loss": 1.9888334274291992,
"step": 16930
},
{
"epoch": 2.168192,
"grad_norm": 0.5707836151123047,
"learning_rate": 0.0002671248408446052,
"loss": 2.0129619598388673,
"step": 16940
},
{
"epoch": 2.169472,
"grad_norm": 0.5378072261810303,
"learning_rate": 0.0002670870535629054,
"loss": 1.9875873565673827,
"step": 16950
},
{
"epoch": 2.170752,
"grad_norm": 0.6076937913894653,
"learning_rate": 0.00026704924725293747,
"loss": 1.9787542343139648,
"step": 16960
},
{
"epoch": 2.172032,
"grad_norm": 0.7067393660545349,
"learning_rate": 0.00026701142192084537,
"loss": 1.9859958648681642,
"step": 16970
},
{
"epoch": 2.173312,
"grad_norm": 0.5844358205795288,
"learning_rate": 0.0002669735775727762,
"loss": 1.9807163238525392,
"step": 16980
},
{
"epoch": 2.174592,
"grad_norm": 0.6154071688652039,
"learning_rate": 0.0002669357142148803,
"loss": 1.9953973770141602,
"step": 16990
},
{
"epoch": 2.175872,
"grad_norm": 0.5516527891159058,
"learning_rate": 0.0002668978318533109,
"loss": 1.9940542221069335,
"step": 17000
},
{
"epoch": 2.177152,
"grad_norm": 0.5863473415374756,
"learning_rate": 0.0002668599304942244,
"loss": 1.981297492980957,
"step": 17010
},
{
"epoch": 2.178432,
"grad_norm": 0.6265770196914673,
"learning_rate": 0.0002668220101437804,
"loss": 1.978145217895508,
"step": 17020
},
{
"epoch": 2.179712,
"grad_norm": 0.6085952520370483,
"learning_rate": 0.0002667840708081414,
"loss": 2.0016925811767576,
"step": 17030
},
{
"epoch": 2.180992,
"grad_norm": 0.6539912819862366,
"learning_rate": 0.0002667461124934731,
"loss": 1.990496826171875,
"step": 17040
},
{
"epoch": 2.182272,
"grad_norm": 0.5354396104812622,
"learning_rate": 0.0002667081352059442,
"loss": 1.9734737396240234,
"step": 17050
},
{
"epoch": 2.183552,
"grad_norm": 0.5724342465400696,
"learning_rate": 0.0002666701389517266,
"loss": 1.9981573104858399,
"step": 17060
},
{
"epoch": 2.184832,
"grad_norm": 0.6858850121498108,
"learning_rate": 0.00026663212373699514,
"loss": 1.9815401077270507,
"step": 17070
},
{
"epoch": 2.186112,
"grad_norm": 0.6323288679122925,
"learning_rate": 0.0002665940895679279,
"loss": 1.9894081115722657,
"step": 17080
},
{
"epoch": 2.187392,
"grad_norm": 0.6740830540657043,
"learning_rate": 0.0002665560364507059,
"loss": 2.002239990234375,
"step": 17090
},
{
"epoch": 2.188672,
"grad_norm": 0.5833739638328552,
"learning_rate": 0.00026651796439151334,
"loss": 2.007330894470215,
"step": 17100
},
{
"epoch": 2.189952,
"grad_norm": 0.5850293040275574,
"learning_rate": 0.0002664798733965374,
"loss": 1.9780082702636719,
"step": 17110
},
{
"epoch": 2.191232,
"grad_norm": 0.6229511499404907,
"learning_rate": 0.00026644176347196855,
"loss": 2.005515289306641,
"step": 17120
},
{
"epoch": 2.192512,
"grad_norm": 0.6311277151107788,
"learning_rate": 0.00026640363462399996,
"loss": 2.0076393127441405,
"step": 17130
},
{
"epoch": 2.193792,
"grad_norm": 0.5841075778007507,
"learning_rate": 0.0002663654868588282,
"loss": 1.9827613830566406,
"step": 17140
},
{
"epoch": 2.195072,
"grad_norm": 0.5864089131355286,
"learning_rate": 0.0002663273201826529,
"loss": 1.9886590957641601,
"step": 17150
},
{
"epoch": 2.196352,
"grad_norm": 0.5810127258300781,
"learning_rate": 0.00026628913460167655,
"loss": 1.9926353454589845,
"step": 17160
},
{
"epoch": 2.197632,
"grad_norm": 0.5428857803344727,
"learning_rate": 0.00026625093012210487,
"loss": 1.9673389434814452,
"step": 17170
},
{
"epoch": 2.198912,
"grad_norm": 0.6185862421989441,
"learning_rate": 0.0002662127067501466,
"loss": 1.9891109466552734,
"step": 17180
},
{
"epoch": 2.200192,
"grad_norm": 0.5858584046363831,
"learning_rate": 0.00026617446449201364,
"loss": 2.01077823638916,
"step": 17190
},
{
"epoch": 2.201472,
"grad_norm": 0.6249011754989624,
"learning_rate": 0.00026613620335392077,
"loss": 2.001113700866699,
"step": 17200
},
{
"epoch": 2.202752,
"grad_norm": 0.5884816646575928,
"learning_rate": 0.00026609792334208603,
"loss": 1.9786155700683594,
"step": 17210
},
{
"epoch": 2.2040319999999998,
"grad_norm": 0.6134967803955078,
"learning_rate": 0.0002660596244627305,
"loss": 1.987736129760742,
"step": 17220
},
{
"epoch": 2.205312,
"grad_norm": 0.5670050382614136,
"learning_rate": 0.00026602130672207815,
"loss": 1.9764320373535156,
"step": 17230
},
{
"epoch": 2.206592,
"grad_norm": 0.5781689286231995,
"learning_rate": 0.00026598297012635626,
"loss": 2.0029085159301756,
"step": 17240
},
{
"epoch": 2.207872,
"grad_norm": 0.5889942646026611,
"learning_rate": 0.000265944614681795,
"loss": 1.9878557205200196,
"step": 17250
},
{
"epoch": 2.209152,
"grad_norm": 0.6128296852111816,
"learning_rate": 0.00026590624039462765,
"loss": 1.9860177993774415,
"step": 17260
},
{
"epoch": 2.210432,
"grad_norm": 0.6180892586708069,
"learning_rate": 0.0002658678472710906,
"loss": 1.9996210098266602,
"step": 17270
},
{
"epoch": 2.211712,
"grad_norm": 0.6520475745201111,
"learning_rate": 0.00026582943531742325,
"loss": 2.0114347457885744,
"step": 17280
},
{
"epoch": 2.212992,
"grad_norm": 0.6385353803634644,
"learning_rate": 0.0002657910045398681,
"loss": 1.9796951293945313,
"step": 17290
},
{
"epoch": 2.214272,
"grad_norm": 0.6996184587478638,
"learning_rate": 0.00026575255494467063,
"loss": 1.9854059219360352,
"step": 17300
},
{
"epoch": 2.215552,
"grad_norm": 0.6439588665962219,
"learning_rate": 0.0002657140865380795,
"loss": 2.0004249572753907,
"step": 17310
},
{
"epoch": 2.216832,
"grad_norm": 0.6236804127693176,
"learning_rate": 0.00026567559932634625,
"loss": 1.9976974487304688,
"step": 17320
},
{
"epoch": 2.218112,
"grad_norm": 0.6050711274147034,
"learning_rate": 0.0002656370933157257,
"loss": 1.9854665756225587,
"step": 17330
},
{
"epoch": 2.219392,
"grad_norm": 0.609224796295166,
"learning_rate": 0.00026559856851247557,
"loss": 1.9931524276733399,
"step": 17340
},
{
"epoch": 2.220672,
"grad_norm": 0.6614769697189331,
"learning_rate": 0.0002655600249228566,
"loss": 1.9923330307006837,
"step": 17350
},
{
"epoch": 2.221952,
"grad_norm": 0.6758286356925964,
"learning_rate": 0.0002655214625531328,
"loss": 2.006420135498047,
"step": 17360
},
{
"epoch": 2.223232,
"grad_norm": 0.7730108499526978,
"learning_rate": 0.00026548288140957097,
"loss": 1.985184097290039,
"step": 17370
},
{
"epoch": 2.224512,
"grad_norm": 0.6342745423316956,
"learning_rate": 0.0002654442814984412,
"loss": 2.0148239135742188,
"step": 17380
},
{
"epoch": 2.225792,
"grad_norm": 0.9029709100723267,
"learning_rate": 0.0002654056628260164,
"loss": 1.9754899978637694,
"step": 17390
},
{
"epoch": 2.227072,
"grad_norm": 0.6500838398933411,
"learning_rate": 0.00026536702539857264,
"loss": 1.9933507919311524,
"step": 17400
},
{
"epoch": 2.228352,
"grad_norm": 0.5569091439247131,
"learning_rate": 0.0002653283692223891,
"loss": 2.0048309326171876,
"step": 17410
},
{
"epoch": 2.229632,
"grad_norm": 0.6215612292289734,
"learning_rate": 0.0002652896943037479,
"loss": 1.9965944290161133,
"step": 17420
},
{
"epoch": 2.230912,
"grad_norm": 0.6349321603775024,
"learning_rate": 0.00026525100064893423,
"loss": 2.0106012344360353,
"step": 17430
},
{
"epoch": 2.232192,
"grad_norm": 0.6146599650382996,
"learning_rate": 0.00026521228826423645,
"loss": 1.9990884780883789,
"step": 17440
},
{
"epoch": 2.233472,
"grad_norm": 0.5484580397605896,
"learning_rate": 0.0002651735571559458,
"loss": 1.9951343536376953,
"step": 17450
},
{
"epoch": 2.234752,
"grad_norm": 0.6171528100967407,
"learning_rate": 0.0002651348073303566,
"loss": 1.989231491088867,
"step": 17460
},
{
"epoch": 2.236032,
"grad_norm": 0.5962356925010681,
"learning_rate": 0.0002650960387937662,
"loss": 1.9707433700561523,
"step": 17470
},
{
"epoch": 2.237312,
"grad_norm": 0.5511232614517212,
"learning_rate": 0.0002650572515524751,
"loss": 1.9919361114501952,
"step": 17480
},
{
"epoch": 2.238592,
"grad_norm": 0.5594847798347473,
"learning_rate": 0.00026501844561278674,
"loss": 2.0052515029907227,
"step": 17490
},
{
"epoch": 2.239872,
"grad_norm": 0.6095628142356873,
"learning_rate": 0.0002649796209810076,
"loss": 1.982847023010254,
"step": 17500
},
{
"epoch": 2.241152,
"grad_norm": 0.6263659000396729,
"learning_rate": 0.0002649407776634472,
"loss": 1.980853271484375,
"step": 17510
},
{
"epoch": 2.242432,
"grad_norm": 0.5929980278015137,
"learning_rate": 0.0002649019156664182,
"loss": 1.9783164978027343,
"step": 17520
},
{
"epoch": 2.243712,
"grad_norm": 0.5372905731201172,
"learning_rate": 0.00026486303499623616,
"loss": 1.9970552444458007,
"step": 17530
},
{
"epoch": 2.244992,
"grad_norm": 0.6050425171852112,
"learning_rate": 0.0002648241356592198,
"loss": 1.9909782409667969,
"step": 17540
},
{
"epoch": 2.246272,
"grad_norm": 0.6140843033790588,
"learning_rate": 0.00026478521766169067,
"loss": 1.9854942321777345,
"step": 17550
},
{
"epoch": 2.2475519999999998,
"grad_norm": 0.7113184928894043,
"learning_rate": 0.00026474628100997354,
"loss": 1.9969526290893556,
"step": 17560
},
{
"epoch": 2.248832,
"grad_norm": 0.5417931079864502,
"learning_rate": 0.0002647073257103962,
"loss": 1.9742570877075196,
"step": 17570
},
{
"epoch": 2.250112,
"grad_norm": 0.6141971349716187,
"learning_rate": 0.00026466835176928934,
"loss": 1.9803632736206054,
"step": 17580
},
{
"epoch": 2.251392,
"grad_norm": 0.6388177871704102,
"learning_rate": 0.0002646293591929869,
"loss": 1.9904024124145507,
"step": 17590
},
{
"epoch": 2.252672,
"grad_norm": 0.6556872725486755,
"learning_rate": 0.00026459034798782555,
"loss": 1.9844205856323243,
"step": 17600
},
{
"epoch": 2.253952,
"grad_norm": 0.5495221018791199,
"learning_rate": 0.0002645513181601453,
"loss": 2.020542335510254,
"step": 17610
},
{
"epoch": 2.255232,
"grad_norm": 0.5736384391784668,
"learning_rate": 0.00026451226971628886,
"loss": 1.9936111450195313,
"step": 17620
},
{
"epoch": 2.256512,
"grad_norm": 0.564436137676239,
"learning_rate": 0.00026447320266260234,
"loss": 1.980379295349121,
"step": 17630
},
{
"epoch": 2.2577920000000002,
"grad_norm": 0.6160185933113098,
"learning_rate": 0.00026443411700543455,
"loss": 1.977853775024414,
"step": 17640
},
{
"epoch": 2.259072,
"grad_norm": 0.5299120545387268,
"learning_rate": 0.0002643950127511375,
"loss": 1.9687599182128905,
"step": 17650
},
{
"epoch": 2.260352,
"grad_norm": 0.5919175148010254,
"learning_rate": 0.0002643558899060662,
"loss": 1.9940338134765625,
"step": 17660
},
{
"epoch": 2.261632,
"grad_norm": 0.5085598230361938,
"learning_rate": 0.0002643167484765786,
"loss": 2.0114133834838865,
"step": 17670
},
{
"epoch": 2.262912,
"grad_norm": 0.5511118769645691,
"learning_rate": 0.00026427758846903564,
"loss": 1.990182113647461,
"step": 17680
},
{
"epoch": 2.264192,
"grad_norm": 0.694582998752594,
"learning_rate": 0.0002642384098898016,
"loss": 1.9865318298339845,
"step": 17690
},
{
"epoch": 2.265472,
"grad_norm": 0.589838445186615,
"learning_rate": 0.00026419921274524336,
"loss": 1.9712915420532227,
"step": 17700
},
{
"epoch": 2.266752,
"grad_norm": 0.6161808371543884,
"learning_rate": 0.00026415999704173105,
"loss": 1.9968269348144532,
"step": 17710
},
{
"epoch": 2.268032,
"grad_norm": 0.6550986766815186,
"learning_rate": 0.0002641207627856377,
"loss": 1.9906673431396484,
"step": 17720
},
{
"epoch": 2.269312,
"grad_norm": 0.5769799947738647,
"learning_rate": 0.00026408150998333957,
"loss": 1.9803895950317383,
"step": 17730
},
{
"epoch": 2.270592,
"grad_norm": 0.5785390734672546,
"learning_rate": 0.00026404223864121565,
"loss": 1.9883649826049805,
"step": 17740
},
{
"epoch": 2.271872,
"grad_norm": 0.6457592844963074,
"learning_rate": 0.0002640029487656482,
"loss": 1.9991167068481446,
"step": 17750
},
{
"epoch": 2.273152,
"grad_norm": 0.63456130027771,
"learning_rate": 0.0002639636403630222,
"loss": 1.9905855178833007,
"step": 17760
},
{
"epoch": 2.274432,
"grad_norm": 0.5794703960418701,
"learning_rate": 0.00026392431343972596,
"loss": 1.9707206726074218,
"step": 17770
},
{
"epoch": 2.275712,
"grad_norm": 0.647574245929718,
"learning_rate": 0.00026388496800215053,
"loss": 1.9875864028930663,
"step": 17780
},
{
"epoch": 2.276992,
"grad_norm": 0.602533757686615,
"learning_rate": 0.0002638456040566902,
"loss": 1.9891407012939453,
"step": 17790
},
{
"epoch": 2.278272,
"grad_norm": 0.5761497616767883,
"learning_rate": 0.0002638062216097421,
"loss": 1.9913896560668944,
"step": 17800
},
{
"epoch": 2.279552,
"grad_norm": 0.5868620872497559,
"learning_rate": 0.0002637668206677064,
"loss": 1.9959062576293944,
"step": 17810
},
{
"epoch": 2.280832,
"grad_norm": 0.6054340600967407,
"learning_rate": 0.00026372740123698635,
"loss": 2.005269432067871,
"step": 17820
},
{
"epoch": 2.282112,
"grad_norm": 0.5764527916908264,
"learning_rate": 0.00026368796332398815,
"loss": 1.9795032501220704,
"step": 17830
},
{
"epoch": 2.283392,
"grad_norm": 0.5876204371452332,
"learning_rate": 0.000263648506935121,
"loss": 1.9995332717895509,
"step": 17840
},
{
"epoch": 2.284672,
"grad_norm": 0.6628866791725159,
"learning_rate": 0.00026360903207679705,
"loss": 1.9832889556884765,
"step": 17850
},
{
"epoch": 2.285952,
"grad_norm": 0.5570622086524963,
"learning_rate": 0.0002635695387554316,
"loss": 1.9629539489746093,
"step": 17860
},
{
"epoch": 2.287232,
"grad_norm": 2.845510244369507,
"learning_rate": 0.00026353002697744273,
"loss": 1.9968286514282227,
"step": 17870
},
{
"epoch": 2.288512,
"grad_norm": 0.6645617485046387,
"learning_rate": 0.0002634904967492519,
"loss": 1.9903200149536133,
"step": 17880
},
{
"epoch": 2.289792,
"grad_norm": 0.6028800010681152,
"learning_rate": 0.00026345094807728306,
"loss": 2.018163871765137,
"step": 17890
},
{
"epoch": 2.2910719999999998,
"grad_norm": 0.6261778473854065,
"learning_rate": 0.0002634113809679635,
"loss": 1.9931440353393555,
"step": 17900
},
{
"epoch": 2.292352,
"grad_norm": 0.6389691829681396,
"learning_rate": 0.0002633717954277235,
"loss": 1.9589393615722657,
"step": 17910
},
{
"epoch": 2.293632,
"grad_norm": 0.6498133540153503,
"learning_rate": 0.0002633321914629962,
"loss": 2.0118669509887694,
"step": 17920
},
{
"epoch": 2.294912,
"grad_norm": 0.6024123430252075,
"learning_rate": 0.0002632925690802178,
"loss": 1.9996503829956054,
"step": 17930
},
{
"epoch": 2.296192,
"grad_norm": 0.6111660003662109,
"learning_rate": 0.00026325292828582744,
"loss": 1.9967226028442382,
"step": 17940
},
{
"epoch": 2.297472,
"grad_norm": 0.6169027090072632,
"learning_rate": 0.00026321326908626736,
"loss": 1.994611930847168,
"step": 17950
},
{
"epoch": 2.298752,
"grad_norm": 0.6254280805587769,
"learning_rate": 0.00026317359148798273,
"loss": 1.9748584747314453,
"step": 17960
},
{
"epoch": 2.300032,
"grad_norm": 0.7401241660118103,
"learning_rate": 0.0002631338954974217,
"loss": 1.9849956512451172,
"step": 17970
},
{
"epoch": 2.3013120000000002,
"grad_norm": 0.6513580083847046,
"learning_rate": 0.0002630941811210354,
"loss": 1.984991455078125,
"step": 17980
},
{
"epoch": 2.302592,
"grad_norm": 0.5755980610847473,
"learning_rate": 0.00026305444836527794,
"loss": 1.9999402999877929,
"step": 17990
},
{
"epoch": 2.303872,
"grad_norm": 0.6190004348754883,
"learning_rate": 0.00026301469723660645,
"loss": 1.986185646057129,
"step": 18000
},
{
"epoch": 2.305152,
"grad_norm": 0.8319825530052185,
"learning_rate": 0.0002629749277414811,
"loss": 1.98746337890625,
"step": 18010
},
{
"epoch": 2.306432,
"grad_norm": 0.5444178581237793,
"learning_rate": 0.00026293513988636497,
"loss": 1.9850433349609375,
"step": 18020
},
{
"epoch": 2.307712,
"grad_norm": 0.7057374715805054,
"learning_rate": 0.0002628953336777241,
"loss": 1.98632755279541,
"step": 18030
},
{
"epoch": 2.308992,
"grad_norm": 0.5286999940872192,
"learning_rate": 0.00026285550912202757,
"loss": 2.009767532348633,
"step": 18040
},
{
"epoch": 2.310272,
"grad_norm": 0.5687797665596008,
"learning_rate": 0.00026281566622574743,
"loss": 1.985262680053711,
"step": 18050
},
{
"epoch": 2.311552,
"grad_norm": 0.5449957847595215,
"learning_rate": 0.0002627758049953587,
"loss": 1.9883285522460938,
"step": 18060
},
{
"epoch": 2.312832,
"grad_norm": 0.5505937933921814,
"learning_rate": 0.00026273592543733933,
"loss": 1.9691699981689452,
"step": 18070
},
{
"epoch": 2.314112,
"grad_norm": 0.6437209844589233,
"learning_rate": 0.00026269602755817035,
"loss": 1.9835197448730468,
"step": 18080
},
{
"epoch": 2.315392,
"grad_norm": 0.5695046782493591,
"learning_rate": 0.0002626561113643358,
"loss": 1.9965295791625977,
"step": 18090
},
{
"epoch": 2.316672,
"grad_norm": 0.6105008125305176,
"learning_rate": 0.0002626161768623224,
"loss": 1.9894573211669921,
"step": 18100
},
{
"epoch": 2.317952,
"grad_norm": 0.49766212701797485,
"learning_rate": 0.0002625762240586203,
"loss": 1.9858783721923827,
"step": 18110
},
{
"epoch": 2.319232,
"grad_norm": 0.6884617805480957,
"learning_rate": 0.0002625362529597222,
"loss": 1.9937780380249024,
"step": 18120
},
{
"epoch": 2.320512,
"grad_norm": 0.5463731288909912,
"learning_rate": 0.000262496263572124,
"loss": 1.987080955505371,
"step": 18130
},
{
"epoch": 2.321792,
"grad_norm": 0.6046035289764404,
"learning_rate": 0.0002624562559023246,
"loss": 1.9951005935668946,
"step": 18140
},
{
"epoch": 2.323072,
"grad_norm": 0.6887285709381104,
"learning_rate": 0.0002624162299568258,
"loss": 1.9866437911987305,
"step": 18150
},
{
"epoch": 2.324352,
"grad_norm": 0.6155261993408203,
"learning_rate": 0.00026237618574213227,
"loss": 1.9840744018554688,
"step": 18160
},
{
"epoch": 2.325632,
"grad_norm": 0.5990715622901917,
"learning_rate": 0.0002623361232647518,
"loss": 1.9707279205322266,
"step": 18170
},
{
"epoch": 2.326912,
"grad_norm": 0.6326669454574585,
"learning_rate": 0.00026229604253119517,
"loss": 1.9886531829833984,
"step": 18180
},
{
"epoch": 2.328192,
"grad_norm": 0.4987259805202484,
"learning_rate": 0.000262255943547976,
"loss": 1.9729671478271484,
"step": 18190
},
{
"epoch": 2.329472,
"grad_norm": 0.5996296405792236,
"learning_rate": 0.00026221582632161085,
"loss": 1.9805360794067384,
"step": 18200
},
{
"epoch": 2.330752,
"grad_norm": 0.5726966261863708,
"learning_rate": 0.0002621756908586194,
"loss": 1.9721216201782226,
"step": 18210
},
{
"epoch": 2.332032,
"grad_norm": 0.5757941603660583,
"learning_rate": 0.00026213553716552423,
"loss": 1.9805669784545898,
"step": 18220
},
{
"epoch": 2.333312,
"grad_norm": 0.6958896517753601,
"learning_rate": 0.00026209536524885086,
"loss": 2.0002296447753904,
"step": 18230
},
{
"epoch": 2.334592,
"grad_norm": 0.5439685583114624,
"learning_rate": 0.00026205517511512777,
"loss": 1.9935373306274413,
"step": 18240
},
{
"epoch": 2.335872,
"grad_norm": 0.7447227239608765,
"learning_rate": 0.00026201496677088647,
"loss": 1.9828699111938477,
"step": 18250
},
{
"epoch": 2.337152,
"grad_norm": 0.6051545739173889,
"learning_rate": 0.0002619747402226613,
"loss": 1.9827491760253906,
"step": 18260
},
{
"epoch": 2.338432,
"grad_norm": 0.5751016736030579,
"learning_rate": 0.00026193449547698964,
"loss": 1.978916549682617,
"step": 18270
},
{
"epoch": 2.339712,
"grad_norm": 0.6401804089546204,
"learning_rate": 0.0002618942325404119,
"loss": 1.978558349609375,
"step": 18280
},
{
"epoch": 2.340992,
"grad_norm": 0.6589930653572083,
"learning_rate": 0.0002618539514194712,
"loss": 1.9918073654174804,
"step": 18290
},
{
"epoch": 2.342272,
"grad_norm": 0.6129372119903564,
"learning_rate": 0.00026181365212071394,
"loss": 1.9875959396362304,
"step": 18300
},
{
"epoch": 2.343552,
"grad_norm": 0.5950047969818115,
"learning_rate": 0.00026177333465068924,
"loss": 1.9785568237304687,
"step": 18310
},
{
"epoch": 2.3448320000000002,
"grad_norm": 0.6365408897399902,
"learning_rate": 0.00026173299901594926,
"loss": 1.9915584564208983,
"step": 18320
},
{
"epoch": 2.3461119999999998,
"grad_norm": 0.6290407776832581,
"learning_rate": 0.0002616926452230491,
"loss": 1.981247901916504,
"step": 18330
},
{
"epoch": 2.347392,
"grad_norm": 0.5886977314949036,
"learning_rate": 0.0002616522732785468,
"loss": 1.9987859725952148,
"step": 18340
},
{
"epoch": 2.348672,
"grad_norm": 0.6366518139839172,
"learning_rate": 0.0002616118831890034,
"loss": 1.9884737014770508,
"step": 18350
},
{
"epoch": 2.349952,
"grad_norm": 0.6193708777427673,
"learning_rate": 0.0002615714749609828,
"loss": 1.9755212783813476,
"step": 18360
},
{
"epoch": 2.351232,
"grad_norm": 0.5880627036094666,
"learning_rate": 0.00026153104860105196,
"loss": 1.995524024963379,
"step": 18370
},
{
"epoch": 2.352512,
"grad_norm": 0.5569005608558655,
"learning_rate": 0.00026149060411578064,
"loss": 1.9932077407836915,
"step": 18380
},
{
"epoch": 2.353792,
"grad_norm": 0.6529499292373657,
"learning_rate": 0.0002614501415117417,
"loss": 1.9880897521972656,
"step": 18390
},
{
"epoch": 2.355072,
"grad_norm": 0.5713687539100647,
"learning_rate": 0.0002614096607955108,
"loss": 2.005253219604492,
"step": 18400
},
{
"epoch": 2.3563520000000002,
"grad_norm": 0.6287975311279297,
"learning_rate": 0.0002613691619736667,
"loss": 2.0002485275268556,
"step": 18410
},
{
"epoch": 2.357632,
"grad_norm": 0.5959773659706116,
"learning_rate": 0.000261328645052791,
"loss": 1.9881851196289062,
"step": 18420
},
{
"epoch": 2.358912,
"grad_norm": 0.6311219334602356,
"learning_rate": 0.0002612881100394682,
"loss": 1.995595932006836,
"step": 18430
},
{
"epoch": 2.360192,
"grad_norm": 0.5564364194869995,
"learning_rate": 0.00026124755694028587,
"loss": 1.9897619247436524,
"step": 18440
},
{
"epoch": 2.361472,
"grad_norm": 0.665963888168335,
"learning_rate": 0.0002612069857618344,
"loss": 2.002791976928711,
"step": 18450
},
{
"epoch": 2.362752,
"grad_norm": 0.5972951650619507,
"learning_rate": 0.0002611663965107072,
"loss": 1.9918428421020509,
"step": 18460
},
{
"epoch": 2.364032,
"grad_norm": 0.5791471600532532,
"learning_rate": 0.00026112578919350055,
"loss": 1.9771808624267577,
"step": 18470
},
{
"epoch": 2.365312,
"grad_norm": 0.5478102564811707,
"learning_rate": 0.00026108516381681383,
"loss": 1.9878179550170898,
"step": 18480
},
{
"epoch": 2.366592,
"grad_norm": 0.5556607246398926,
"learning_rate": 0.000261044520387249,
"loss": 2.0111122131347656,
"step": 18490
},
{
"epoch": 2.367872,
"grad_norm": 0.6312317252159119,
"learning_rate": 0.00026100385891141134,
"loss": 1.981797981262207,
"step": 18500
},
{
"epoch": 2.369152,
"grad_norm": 0.6411010026931763,
"learning_rate": 0.00026096317939590895,
"loss": 1.9774667739868164,
"step": 18510
},
{
"epoch": 2.370432,
"grad_norm": 0.6606515049934387,
"learning_rate": 0.0002609224818473527,
"loss": 1.9800556182861329,
"step": 18520
},
{
"epoch": 2.371712,
"grad_norm": 0.7043580412864685,
"learning_rate": 0.00026088176627235646,
"loss": 1.9968446731567382,
"step": 18530
},
{
"epoch": 2.372992,
"grad_norm": 0.6475520133972168,
"learning_rate": 0.00026084103267753726,
"loss": 1.9720874786376954,
"step": 18540
},
{
"epoch": 2.374272,
"grad_norm": 0.5602666139602661,
"learning_rate": 0.0002608002810695147,
"loss": 1.9932565689086914,
"step": 18550
},
{
"epoch": 2.375552,
"grad_norm": 0.6393943428993225,
"learning_rate": 0.0002607595114549116,
"loss": 1.981712532043457,
"step": 18560
},
{
"epoch": 2.376832,
"grad_norm": 0.597075343132019,
"learning_rate": 0.0002607187238403535,
"loss": 2.000041961669922,
"step": 18570
},
{
"epoch": 2.378112,
"grad_norm": 0.5137078166007996,
"learning_rate": 0.00026067791823246897,
"loss": 2.00323371887207,
"step": 18580
},
{
"epoch": 2.379392,
"grad_norm": 0.6078038811683655,
"learning_rate": 0.0002606370946378895,
"loss": 1.9918882369995117,
"step": 18590
},
{
"epoch": 2.380672,
"grad_norm": 0.6824135780334473,
"learning_rate": 0.00026059625306324954,
"loss": 1.9980682373046874,
"step": 18600
},
{
"epoch": 2.381952,
"grad_norm": 0.6160506010055542,
"learning_rate": 0.00026055539351518635,
"loss": 1.9969316482543946,
"step": 18610
},
{
"epoch": 2.383232,
"grad_norm": 0.6948800683021545,
"learning_rate": 0.0002605145160003402,
"loss": 2.0089136123657227,
"step": 18620
},
{
"epoch": 2.384512,
"grad_norm": 0.5942563414573669,
"learning_rate": 0.00026047362052535424,
"loss": 1.987285041809082,
"step": 18630
},
{
"epoch": 2.385792,
"grad_norm": 0.6004053950309753,
"learning_rate": 0.00026043270709687455,
"loss": 1.9770875930786134,
"step": 18640
},
{
"epoch": 2.387072,
"grad_norm": 0.542837917804718,
"learning_rate": 0.0002603917757215501,
"loss": 1.9975923538208007,
"step": 18650
},
{
"epoch": 2.388352,
"grad_norm": 0.7407999634742737,
"learning_rate": 0.00026035082640603287,
"loss": 1.9821928024291993,
"step": 18660
},
{
"epoch": 2.3896319999999998,
"grad_norm": 0.5443654656410217,
"learning_rate": 0.0002603098591569776,
"loss": 2.0052337646484375,
"step": 18670
},
{
"epoch": 2.390912,
"grad_norm": 0.5436390042304993,
"learning_rate": 0.0002602688739810422,
"loss": 1.989246940612793,
"step": 18680
},
{
"epoch": 2.392192,
"grad_norm": 0.573955774307251,
"learning_rate": 0.00026022787088488713,
"loss": 1.9641033172607423,
"step": 18690
},
{
"epoch": 2.393472,
"grad_norm": 0.6994761228561401,
"learning_rate": 0.0002601868498751761,
"loss": 1.9879415512084961,
"step": 18700
},
{
"epoch": 2.394752,
"grad_norm": 0.7131639719009399,
"learning_rate": 0.00026014581095857556,
"loss": 2.016462516784668,
"step": 18710
},
{
"epoch": 2.396032,
"grad_norm": 0.6583504676818848,
"learning_rate": 0.00026010475414175484,
"loss": 2.0001264572143556,
"step": 18720
},
{
"epoch": 2.397312,
"grad_norm": 0.6237305998802185,
"learning_rate": 0.00026006367943138636,
"loss": 2.004298973083496,
"step": 18730
},
{
"epoch": 2.398592,
"grad_norm": 0.5682983994483948,
"learning_rate": 0.0002600225868341452,
"loss": 1.993436050415039,
"step": 18740
},
{
"epoch": 2.3998720000000002,
"grad_norm": 0.6071258783340454,
"learning_rate": 0.00025998147635670956,
"loss": 1.9881019592285156,
"step": 18750
},
{
"epoch": 2.401152,
"grad_norm": 0.7686408758163452,
"learning_rate": 0.0002599403480057604,
"loss": 2.004251480102539,
"step": 18760
},
{
"epoch": 2.402432,
"grad_norm": 0.7473837733268738,
"learning_rate": 0.00025989920178798174,
"loss": 2.0021596908569337,
"step": 18770
},
{
"epoch": 2.403712,
"grad_norm": 0.5940958857536316,
"learning_rate": 0.0002598580377100603,
"loss": 1.9837324142456054,
"step": 18780
},
{
"epoch": 2.404992,
"grad_norm": 0.5570366382598877,
"learning_rate": 0.0002598168557786859,
"loss": 1.9943326950073241,
"step": 18790
},
{
"epoch": 2.406272,
"grad_norm": 0.6208986639976501,
"learning_rate": 0.00025977565600055115,
"loss": 1.9745706558227538,
"step": 18800
},
{
"epoch": 2.407552,
"grad_norm": 0.559415340423584,
"learning_rate": 0.00025973443838235154,
"loss": 2.0017213821411133,
"step": 18810
},
{
"epoch": 2.408832,
"grad_norm": 0.5845942497253418,
"learning_rate": 0.00025969320293078553,
"loss": 1.9871971130371093,
"step": 18820
},
{
"epoch": 2.410112,
"grad_norm": 0.6346386075019836,
"learning_rate": 0.0002596519496525546,
"loss": 2.006735610961914,
"step": 18830
},
{
"epoch": 2.411392,
"grad_norm": 0.5966348648071289,
"learning_rate": 0.00025961067855436267,
"loss": 1.9854211807250977,
"step": 18840
},
{
"epoch": 2.412672,
"grad_norm": 0.5762596726417542,
"learning_rate": 0.0002595693896429171,
"loss": 1.9633712768554688,
"step": 18850
},
{
"epoch": 2.413952,
"grad_norm": 0.5814616084098816,
"learning_rate": 0.00025952808292492786,
"loss": 1.9804222106933593,
"step": 18860
},
{
"epoch": 2.415232,
"grad_norm": 0.6572591662406921,
"learning_rate": 0.00025948675840710786,
"loss": 1.993008804321289,
"step": 18870
},
{
"epoch": 2.416512,
"grad_norm": 0.726348340511322,
"learning_rate": 0.0002594454160961729,
"loss": 1.980824089050293,
"step": 18880
},
{
"epoch": 2.417792,
"grad_norm": 0.573469877243042,
"learning_rate": 0.00025940405599884166,
"loss": 2.003045654296875,
"step": 18890
},
{
"epoch": 2.419072,
"grad_norm": 0.7677147388458252,
"learning_rate": 0.0002593626781218357,
"loss": 1.9893501281738282,
"step": 18900
},
{
"epoch": 2.420352,
"grad_norm": 0.5860797166824341,
"learning_rate": 0.0002593212824718796,
"loss": 1.978499984741211,
"step": 18910
},
{
"epoch": 2.421632,
"grad_norm": 0.6168810129165649,
"learning_rate": 0.0002592798690557007,
"loss": 2.0058513641357423,
"step": 18920
},
{
"epoch": 2.422912,
"grad_norm": 0.5480222105979919,
"learning_rate": 0.0002592384378800292,
"loss": 1.9898273468017578,
"step": 18930
},
{
"epoch": 2.424192,
"grad_norm": 0.6359663605690002,
"learning_rate": 0.0002591969889515982,
"loss": 1.9913637161254882,
"step": 18940
},
{
"epoch": 2.425472,
"grad_norm": 0.5736200213432312,
"learning_rate": 0.00025915552227714385,
"loss": 1.9791936874389648,
"step": 18950
},
{
"epoch": 2.426752,
"grad_norm": 0.6981081366539001,
"learning_rate": 0.00025911403786340497,
"loss": 1.9981678009033204,
"step": 18960
},
{
"epoch": 2.428032,
"grad_norm": 0.6156948208808899,
"learning_rate": 0.0002590725357171234,
"loss": 1.9793056488037108,
"step": 18970
},
{
"epoch": 2.429312,
"grad_norm": 0.5631027221679688,
"learning_rate": 0.00025903101584504376,
"loss": 1.9992835998535157,
"step": 18980
},
{
"epoch": 2.430592,
"grad_norm": 0.6569755673408508,
"learning_rate": 0.0002589894782539137,
"loss": 1.9871839523315429,
"step": 18990
},
{
"epoch": 2.431872,
"grad_norm": 0.687342643737793,
"learning_rate": 0.0002589479229504835,
"loss": 1.9880578994750977,
"step": 19000
},
{
"epoch": 2.4331519999999998,
"grad_norm": 0.5785744190216064,
"learning_rate": 0.00025890634994150664,
"loss": 1.9778207778930663,
"step": 19010
},
{
"epoch": 2.434432,
"grad_norm": 0.6440554857254028,
"learning_rate": 0.0002588647592337392,
"loss": 1.9771774291992188,
"step": 19020
},
{
"epoch": 2.435712,
"grad_norm": 0.605967104434967,
"learning_rate": 0.0002588231508339403,
"loss": 1.9786693572998046,
"step": 19030
},
{
"epoch": 2.436992,
"grad_norm": 0.553110659122467,
"learning_rate": 0.0002587815247488718,
"loss": 1.978370475769043,
"step": 19040
},
{
"epoch": 2.438272,
"grad_norm": 0.6702958345413208,
"learning_rate": 0.0002587398809852987,
"loss": 1.9785800933837892,
"step": 19050
},
{
"epoch": 2.439552,
"grad_norm": 0.6458420753479004,
"learning_rate": 0.0002586982195499885,
"loss": 1.978866195678711,
"step": 19060
},
{
"epoch": 2.440832,
"grad_norm": 0.5337917804718018,
"learning_rate": 0.0002586565404497118,
"loss": 2.002280044555664,
"step": 19070
},
{
"epoch": 2.442112,
"grad_norm": 0.6380541920661926,
"learning_rate": 0.0002586148436912421,
"loss": 1.99906005859375,
"step": 19080
},
{
"epoch": 2.4433920000000002,
"grad_norm": 0.6328800320625305,
"learning_rate": 0.0002585731292813556,
"loss": 2.0017967224121094,
"step": 19090
},
{
"epoch": 2.444672,
"grad_norm": 0.6171122193336487,
"learning_rate": 0.0002585313972268316,
"loss": 1.995723342895508,
"step": 19100
},
{
"epoch": 2.445952,
"grad_norm": 0.5575553178787231,
"learning_rate": 0.000258489647534452,
"loss": 1.9684785842895507,
"step": 19110
},
{
"epoch": 2.447232,
"grad_norm": 0.5836408734321594,
"learning_rate": 0.0002584478802110018,
"loss": 1.9995527267456055,
"step": 19120
},
{
"epoch": 2.448512,
"grad_norm": 0.5745341181755066,
"learning_rate": 0.00025840609526326877,
"loss": 1.989507293701172,
"step": 19130
},
{
"epoch": 2.449792,
"grad_norm": 0.6977150440216064,
"learning_rate": 0.00025836429269804345,
"loss": 1.9933107376098633,
"step": 19140
},
{
"epoch": 2.451072,
"grad_norm": 0.6025040149688721,
"learning_rate": 0.0002583224725221194,
"loss": 2.012924385070801,
"step": 19150
},
{
"epoch": 2.452352,
"grad_norm": 0.6754240989685059,
"learning_rate": 0.000258280634742293,
"loss": 1.9989892959594726,
"step": 19160
},
{
"epoch": 2.453632,
"grad_norm": 0.5747501254081726,
"learning_rate": 0.0002582387793653634,
"loss": 1.987969970703125,
"step": 19170
},
{
"epoch": 2.454912,
"grad_norm": 0.6488444209098816,
"learning_rate": 0.0002581969063981328,
"loss": 2.004003143310547,
"step": 19180
},
{
"epoch": 2.456192,
"grad_norm": 0.5848543047904968,
"learning_rate": 0.000258155015847406,
"loss": 2.0106645584106446,
"step": 19190
},
{
"epoch": 2.457472,
"grad_norm": 0.6598523855209351,
"learning_rate": 0.0002581131077199909,
"loss": 1.9829477310180663,
"step": 19200
},
{
"epoch": 2.458752,
"grad_norm": 0.6288194060325623,
"learning_rate": 0.000258071182022698,
"loss": 1.982452964782715,
"step": 19210
},
{
"epoch": 2.460032,
"grad_norm": 0.678694486618042,
"learning_rate": 0.000258029238762341,
"loss": 1.9718639373779296,
"step": 19220
},
{
"epoch": 2.461312,
"grad_norm": 0.6441311240196228,
"learning_rate": 0.00025798727794573614,
"loss": 1.9897354125976563,
"step": 19230
},
{
"epoch": 2.462592,
"grad_norm": 0.6032792925834656,
"learning_rate": 0.0002579452995797027,
"loss": 1.9928312301635742,
"step": 19240
},
{
"epoch": 2.463872,
"grad_norm": 0.6283438801765442,
"learning_rate": 0.00025790330367106266,
"loss": 1.976766014099121,
"step": 19250
},
{
"epoch": 2.465152,
"grad_norm": 0.5702046155929565,
"learning_rate": 0.000257861290226641,
"loss": 1.9974056243896485,
"step": 19260
},
{
"epoch": 2.466432,
"grad_norm": 0.566746175289154,
"learning_rate": 0.00025781925925326545,
"loss": 1.9723855972290039,
"step": 19270
},
{
"epoch": 2.467712,
"grad_norm": 0.5223069787025452,
"learning_rate": 0.0002577772107577667,
"loss": 1.9898614883422852,
"step": 19280
},
{
"epoch": 2.468992,
"grad_norm": 0.5777895450592041,
"learning_rate": 0.0002577351447469782,
"loss": 1.9875097274780273,
"step": 19290
},
{
"epoch": 2.470272,
"grad_norm": 0.5831905603408813,
"learning_rate": 0.0002576930612277361,
"loss": 1.9867504119873047,
"step": 19300
},
{
"epoch": 2.471552,
"grad_norm": 0.6571018099784851,
"learning_rate": 0.0002576509602068798,
"loss": 1.992510986328125,
"step": 19310
},
{
"epoch": 2.472832,
"grad_norm": 0.6134997606277466,
"learning_rate": 0.00025760884169125114,
"loss": 1.9807838439941405,
"step": 19320
},
{
"epoch": 2.474112,
"grad_norm": 0.5618544816970825,
"learning_rate": 0.000257566705687695,
"loss": 1.969550895690918,
"step": 19330
},
{
"epoch": 2.475392,
"grad_norm": 0.5581712126731873,
"learning_rate": 0.000257524552203059,
"loss": 1.976296615600586,
"step": 19340
},
{
"epoch": 2.4766719999999998,
"grad_norm": 0.6056370139122009,
"learning_rate": 0.00025748238124419383,
"loss": 1.9826347351074218,
"step": 19350
},
{
"epoch": 2.477952,
"grad_norm": 0.5810211300849915,
"learning_rate": 0.0002574401928179527,
"loss": 1.9865314483642578,
"step": 19360
},
{
"epoch": 2.479232,
"grad_norm": 0.5448335409164429,
"learning_rate": 0.00025739798693119186,
"loss": 1.9740015029907227,
"step": 19370
},
{
"epoch": 2.480512,
"grad_norm": 0.5299369096755981,
"learning_rate": 0.0002573557635907704,
"loss": 1.9704181671142578,
"step": 19380
},
{
"epoch": 2.481792,
"grad_norm": 0.6057597398757935,
"learning_rate": 0.0002573135228035501,
"loss": 1.9929649353027343,
"step": 19390
},
{
"epoch": 2.483072,
"grad_norm": 0.5530161261558533,
"learning_rate": 0.00025727126457639573,
"loss": 1.9847925186157227,
"step": 19400
},
{
"epoch": 2.484352,
"grad_norm": 0.598807156085968,
"learning_rate": 0.0002572289889161749,
"loss": 1.9861415863037108,
"step": 19410
},
{
"epoch": 2.485632,
"grad_norm": 0.5776075720787048,
"learning_rate": 0.00025718669582975787,
"loss": 1.981983757019043,
"step": 19420
},
{
"epoch": 2.4869120000000002,
"grad_norm": 0.6371303796768188,
"learning_rate": 0.0002571443853240179,
"loss": 1.9868539810180663,
"step": 19430
},
{
"epoch": 2.488192,
"grad_norm": 0.5989758968353271,
"learning_rate": 0.00025710205740583106,
"loss": 1.9826787948608398,
"step": 19440
},
{
"epoch": 2.489472,
"grad_norm": 0.5794610977172852,
"learning_rate": 0.00025705971208207627,
"loss": 1.997962760925293,
"step": 19450
},
{
"epoch": 2.490752,
"grad_norm": 0.6710638999938965,
"learning_rate": 0.0002570173493596351,
"loss": 1.9764627456665038,
"step": 19460
},
{
"epoch": 2.492032,
"grad_norm": 0.6291629076004028,
"learning_rate": 0.00025697496924539215,
"loss": 1.9890480041503906,
"step": 19470
},
{
"epoch": 2.493312,
"grad_norm": 0.7084724307060242,
"learning_rate": 0.00025693257174623475,
"loss": 1.971998405456543,
"step": 19480
},
{
"epoch": 2.494592,
"grad_norm": 0.6851497888565063,
"learning_rate": 0.00025689015686905314,
"loss": 1.9731935501098632,
"step": 19490
},
{
"epoch": 2.495872,
"grad_norm": 0.6015073657035828,
"learning_rate": 0.0002568477246207403,
"loss": 1.9931316375732422,
"step": 19500
},
{
"epoch": 2.497152,
"grad_norm": 0.5399225950241089,
"learning_rate": 0.00025680527500819203,
"loss": 1.9871559143066406,
"step": 19510
},
{
"epoch": 2.498432,
"grad_norm": 0.6576582789421082,
"learning_rate": 0.000256762808038307,
"loss": 1.9710853576660157,
"step": 19520
},
{
"epoch": 2.499712,
"grad_norm": 0.733462929725647,
"learning_rate": 0.0002567203237179867,
"loss": 2.0100780487060548,
"step": 19530
},
{
"epoch": 2.500992,
"grad_norm": 0.7081230282783508,
"learning_rate": 0.00025667782205413534,
"loss": 1.9903432846069335,
"step": 19540
},
{
"epoch": 2.502272,
"grad_norm": 0.5553060173988342,
"learning_rate": 0.00025663530305366014,
"loss": 1.9692209243774415,
"step": 19550
},
{
"epoch": 2.503552,
"grad_norm": 0.558704137802124,
"learning_rate": 0.00025659276672347094,
"loss": 1.989079475402832,
"step": 19560
},
{
"epoch": 2.504832,
"grad_norm": 0.6334920525550842,
"learning_rate": 0.00025655021307048056,
"loss": 2.000078582763672,
"step": 19570
},
{
"epoch": 2.506112,
"grad_norm": 0.6278864741325378,
"learning_rate": 0.00025650764210160453,
"loss": 2.009459686279297,
"step": 19580
},
{
"epoch": 2.507392,
"grad_norm": 0.5404990911483765,
"learning_rate": 0.00025646505382376114,
"loss": 1.974098014831543,
"step": 19590
},
{
"epoch": 2.508672,
"grad_norm": 0.5697225332260132,
"learning_rate": 0.00025642244824387167,
"loss": 2.0033769607543945,
"step": 19600
},
{
"epoch": 2.509952,
"grad_norm": 0.5483403205871582,
"learning_rate": 0.00025637982536886006,
"loss": 2.0001211166381836,
"step": 19610
},
{
"epoch": 2.511232,
"grad_norm": 0.6175240874290466,
"learning_rate": 0.0002563371852056532,
"loss": 1.9716852188110352,
"step": 19620
},
{
"epoch": 2.512512,
"grad_norm": 0.6626465320587158,
"learning_rate": 0.0002562945277611806,
"loss": 1.9929630279541015,
"step": 19630
},
{
"epoch": 2.513792,
"grad_norm": 0.5586512684822083,
"learning_rate": 0.0002562518530423748,
"loss": 1.9667703628540039,
"step": 19640
},
{
"epoch": 2.515072,
"grad_norm": 0.5869081616401672,
"learning_rate": 0.0002562091610561709,
"loss": 1.9819976806640625,
"step": 19650
},
{
"epoch": 2.516352,
"grad_norm": 0.6683209538459778,
"learning_rate": 0.00025616645180950707,
"loss": 1.985342025756836,
"step": 19660
},
{
"epoch": 2.517632,
"grad_norm": 0.573935329914093,
"learning_rate": 0.0002561237253093241,
"loss": 1.9735506057739258,
"step": 19670
},
{
"epoch": 2.5189120000000003,
"grad_norm": 0.5255028009414673,
"learning_rate": 0.0002560809815625656,
"loss": 1.9975545883178711,
"step": 19680
},
{
"epoch": 2.5201919999999998,
"grad_norm": 0.6108468770980835,
"learning_rate": 0.0002560382205761781,
"loss": 1.971180534362793,
"step": 19690
},
{
"epoch": 2.521472,
"grad_norm": 0.5645589232444763,
"learning_rate": 0.00025599544235711076,
"loss": 1.9998807907104492,
"step": 19700
},
{
"epoch": 2.522752,
"grad_norm": 0.5751380324363708,
"learning_rate": 0.0002559526469123157,
"loss": 1.9857597351074219,
"step": 19710
},
{
"epoch": 2.524032,
"grad_norm": 0.712150514125824,
"learning_rate": 0.00025590983424874774,
"loss": 2.009260559082031,
"step": 19720
},
{
"epoch": 2.525312,
"grad_norm": 0.7002987861633301,
"learning_rate": 0.0002558670043733646,
"loss": 1.990275764465332,
"step": 19730
},
{
"epoch": 2.526592,
"grad_norm": 0.573772668838501,
"learning_rate": 0.0002558241572931266,
"loss": 1.9961820602416993,
"step": 19740
},
{
"epoch": 2.527872,
"grad_norm": 0.5856409668922424,
"learning_rate": 0.00025578129301499704,
"loss": 1.9695327758789063,
"step": 19750
},
{
"epoch": 2.529152,
"grad_norm": 0.6549233198165894,
"learning_rate": 0.00025573841154594206,
"loss": 1.9791854858398437,
"step": 19760
},
{
"epoch": 2.5304320000000002,
"grad_norm": 0.6636223793029785,
"learning_rate": 0.00025569551289293034,
"loss": 2.000459671020508,
"step": 19770
},
{
"epoch": 2.5317119999999997,
"grad_norm": 0.5736064314842224,
"learning_rate": 0.00025565259706293364,
"loss": 1.9912738800048828,
"step": 19780
},
{
"epoch": 2.532992,
"grad_norm": 0.6241442561149597,
"learning_rate": 0.0002556096640629263,
"loss": 1.9991056442260742,
"step": 19790
},
{
"epoch": 2.534272,
"grad_norm": 0.586201012134552,
"learning_rate": 0.0002555667138998855,
"loss": 1.9797237396240235,
"step": 19800
},
{
"epoch": 2.535552,
"grad_norm": 0.7390239834785461,
"learning_rate": 0.0002555237465807913,
"loss": 1.983125877380371,
"step": 19810
},
{
"epoch": 2.536832,
"grad_norm": 0.6870375871658325,
"learning_rate": 0.0002554807621126264,
"loss": 2.0042842864990233,
"step": 19820
},
{
"epoch": 2.538112,
"grad_norm": 0.5499140024185181,
"learning_rate": 0.0002554377605023765,
"loss": 1.9860500335693358,
"step": 19830
},
{
"epoch": 2.539392,
"grad_norm": 0.6080779433250427,
"learning_rate": 0.00025539474175702986,
"loss": 1.9869606018066406,
"step": 19840
},
{
"epoch": 2.540672,
"grad_norm": 0.578097403049469,
"learning_rate": 0.0002553517058835777,
"loss": 2.0069324493408205,
"step": 19850
},
{
"epoch": 2.541952,
"grad_norm": 0.5416150689125061,
"learning_rate": 0.0002553086528890138,
"loss": 1.991615104675293,
"step": 19860
},
{
"epoch": 2.543232,
"grad_norm": 0.6478689908981323,
"learning_rate": 0.0002552655827803351,
"loss": 2.0078989028930665,
"step": 19870
},
{
"epoch": 2.544512,
"grad_norm": 0.5298005938529968,
"learning_rate": 0.00025522249556454084,
"loss": 2.0122814178466797,
"step": 19880
},
{
"epoch": 2.545792,
"grad_norm": 0.602726936340332,
"learning_rate": 0.0002551793912486334,
"loss": 1.9970083236694336,
"step": 19890
},
{
"epoch": 2.547072,
"grad_norm": 0.5833805203437805,
"learning_rate": 0.00025513626983961787,
"loss": 1.9784685134887696,
"step": 19900
},
{
"epoch": 2.548352,
"grad_norm": 0.5292240381240845,
"learning_rate": 0.000255093131344502,
"loss": 1.973914337158203,
"step": 19910
},
{
"epoch": 2.549632,
"grad_norm": 0.6284692287445068,
"learning_rate": 0.00025504997577029646,
"loss": 1.994587516784668,
"step": 19920
},
{
"epoch": 2.550912,
"grad_norm": 0.6330600380897522,
"learning_rate": 0.00025500680312401467,
"loss": 2.0039180755615233,
"step": 19930
},
{
"epoch": 2.552192,
"grad_norm": 0.6104211211204529,
"learning_rate": 0.00025496361341267263,
"loss": 1.9674293518066406,
"step": 19940
},
{
"epoch": 2.553472,
"grad_norm": 0.6590460538864136,
"learning_rate": 0.0002549204066432894,
"loss": 1.9785411834716797,
"step": 19950
},
{
"epoch": 2.554752,
"grad_norm": 0.5464001297950745,
"learning_rate": 0.0002548771828228866,
"loss": 1.9918848037719727,
"step": 19960
},
{
"epoch": 2.556032,
"grad_norm": 0.5842373371124268,
"learning_rate": 0.00025483394195848863,
"loss": 1.9752182006835937,
"step": 19970
},
{
"epoch": 2.557312,
"grad_norm": 0.5500530004501343,
"learning_rate": 0.000254790684057123,
"loss": 1.995145606994629,
"step": 19980
},
{
"epoch": 2.558592,
"grad_norm": 0.6013555526733398,
"learning_rate": 0.0002547474091258194,
"loss": 1.9743261337280273,
"step": 19990
},
{
"epoch": 2.559872,
"grad_norm": 0.6047675013542175,
"learning_rate": 0.0002547041171716108,
"loss": 1.986902618408203,
"step": 20000
},
{
"epoch": 2.561152,
"grad_norm": 0.6075738072395325,
"learning_rate": 0.00025466080820153275,
"loss": 2.0010066986083985,
"step": 20010
},
{
"epoch": 2.5624320000000003,
"grad_norm": 0.6177447438240051,
"learning_rate": 0.00025461748222262344,
"loss": 2.0025201797485352,
"step": 20020
},
{
"epoch": 2.5637119999999998,
"grad_norm": 0.619729220867157,
"learning_rate": 0.000254574139241924,
"loss": 2.003738594055176,
"step": 20030
},
{
"epoch": 2.564992,
"grad_norm": 0.5726514458656311,
"learning_rate": 0.00025453077926647827,
"loss": 1.9913459777832032,
"step": 20040
},
{
"epoch": 2.566272,
"grad_norm": 0.6998363137245178,
"learning_rate": 0.00025448740230333286,
"loss": 1.997663688659668,
"step": 20050
},
{
"epoch": 2.567552,
"grad_norm": 0.6182289719581604,
"learning_rate": 0.0002544440083595371,
"loss": 1.9665864944458007,
"step": 20060
},
{
"epoch": 2.568832,
"grad_norm": 0.594597578048706,
"learning_rate": 0.0002544005974421432,
"loss": 2.0070768356323243,
"step": 20070
},
{
"epoch": 2.570112,
"grad_norm": 0.5241022706031799,
"learning_rate": 0.00025435716955820583,
"loss": 1.9789278030395507,
"step": 20080
},
{
"epoch": 2.571392,
"grad_norm": 0.5931116938591003,
"learning_rate": 0.0002543137247147828,
"loss": 1.9810028076171875,
"step": 20090
},
{
"epoch": 2.572672,
"grad_norm": 0.5686262249946594,
"learning_rate": 0.0002542702629189344,
"loss": 1.990350341796875,
"step": 20100
},
{
"epoch": 2.5739520000000002,
"grad_norm": 0.5527195930480957,
"learning_rate": 0.00025422678417772387,
"loss": 1.959096336364746,
"step": 20110
},
{
"epoch": 2.5752319999999997,
"grad_norm": 0.5645495653152466,
"learning_rate": 0.00025418328849821707,
"loss": 1.983707809448242,
"step": 20120
},
{
"epoch": 2.576512,
"grad_norm": 0.5288122296333313,
"learning_rate": 0.00025413977588748264,
"loss": 1.979289436340332,
"step": 20130
},
{
"epoch": 2.577792,
"grad_norm": 0.6246774196624756,
"learning_rate": 0.0002540962463525919,
"loss": 1.9880058288574218,
"step": 20140
},
{
"epoch": 2.579072,
"grad_norm": 0.5489709973335266,
"learning_rate": 0.0002540526999006191,
"loss": 1.9917680740356445,
"step": 20150
},
{
"epoch": 2.580352,
"grad_norm": 0.6247199177742004,
"learning_rate": 0.0002540091365386412,
"loss": 1.9764829635620118,
"step": 20160
},
{
"epoch": 2.581632,
"grad_norm": 0.6039682030677795,
"learning_rate": 0.00025396555627373767,
"loss": 1.9900489807128907,
"step": 20170
},
{
"epoch": 2.582912,
"grad_norm": 0.6036326885223389,
"learning_rate": 0.0002539219591129911,
"loss": 1.9871999740600585,
"step": 20180
},
{
"epoch": 2.584192,
"grad_norm": 0.6709945201873779,
"learning_rate": 0.0002538783450634865,
"loss": 1.9915632247924804,
"step": 20190
},
{
"epoch": 2.585472,
"grad_norm": 0.6652891635894775,
"learning_rate": 0.00025383471413231186,
"loss": 1.9761783599853515,
"step": 20200
},
{
"epoch": 2.586752,
"grad_norm": 0.5688115358352661,
"learning_rate": 0.0002537910663265577,
"loss": 1.9812030792236328,
"step": 20210
},
{
"epoch": 2.588032,
"grad_norm": 0.6178849339485168,
"learning_rate": 0.0002537474016533174,
"loss": 1.975015640258789,
"step": 20220
},
{
"epoch": 2.589312,
"grad_norm": 0.5673245787620544,
"learning_rate": 0.00025370372011968717,
"loss": 1.9930919647216796,
"step": 20230
},
{
"epoch": 2.590592,
"grad_norm": 0.6779986023902893,
"learning_rate": 0.0002536600217327658,
"loss": 1.9958126068115234,
"step": 20240
},
{
"epoch": 2.591872,
"grad_norm": 0.6642085909843445,
"learning_rate": 0.00025361630649965494,
"loss": 1.9800233840942383,
"step": 20250
},
{
"epoch": 2.593152,
"grad_norm": 0.5161119103431702,
"learning_rate": 0.0002535725744274588,
"loss": 1.9976259231567384,
"step": 20260
},
{
"epoch": 2.594432,
"grad_norm": 0.6207406520843506,
"learning_rate": 0.00025352882552328454,
"loss": 1.973170852661133,
"step": 20270
},
{
"epoch": 2.595712,
"grad_norm": 0.6226537227630615,
"learning_rate": 0.000253485059794242,
"loss": 1.9873443603515626,
"step": 20280
},
{
"epoch": 2.596992,
"grad_norm": 0.5566911101341248,
"learning_rate": 0.0002534412772474437,
"loss": 1.992578125,
"step": 20290
},
{
"epoch": 2.598272,
"grad_norm": 0.6286208629608154,
"learning_rate": 0.00025339747789000474,
"loss": 1.9946355819702148,
"step": 20300
},
{
"epoch": 2.599552,
"grad_norm": 0.6312817335128784,
"learning_rate": 0.0002533536617290433,
"loss": 1.9855249404907227,
"step": 20310
},
{
"epoch": 2.600832,
"grad_norm": 0.6072302460670471,
"learning_rate": 0.0002533098287716801,
"loss": 2.0041105270385744,
"step": 20320
},
{
"epoch": 2.602112,
"grad_norm": 0.5875717401504517,
"learning_rate": 0.0002532659790250386,
"loss": 1.9714643478393554,
"step": 20330
},
{
"epoch": 2.603392,
"grad_norm": 0.5450795292854309,
"learning_rate": 0.0002532221124962449,
"loss": 2.0028911590576173,
"step": 20340
},
{
"epoch": 2.604672,
"grad_norm": 0.604469358921051,
"learning_rate": 0.000253178229192428,
"loss": 1.9884471893310547,
"step": 20350
},
{
"epoch": 2.6059520000000003,
"grad_norm": 0.5368990898132324,
"learning_rate": 0.0002531343291207195,
"loss": 1.9929889678955077,
"step": 20360
},
{
"epoch": 2.6072319999999998,
"grad_norm": 0.6074661612510681,
"learning_rate": 0.00025309041228825386,
"loss": 1.965897560119629,
"step": 20370
},
{
"epoch": 2.608512,
"grad_norm": 0.7553812265396118,
"learning_rate": 0.0002530464787021681,
"loss": 1.9846771240234375,
"step": 20380
},
{
"epoch": 2.609792,
"grad_norm": 0.6679823398590088,
"learning_rate": 0.00025300252836960203,
"loss": 1.9871477127075194,
"step": 20390
},
{
"epoch": 2.611072,
"grad_norm": 0.6761512160301208,
"learning_rate": 0.0002529585612976982,
"loss": 1.9801698684692384,
"step": 20400
},
{
"epoch": 2.612352,
"grad_norm": 0.6821287274360657,
"learning_rate": 0.00025291457749360195,
"loss": 1.9769842147827148,
"step": 20410
},
{
"epoch": 2.613632,
"grad_norm": 0.6064489483833313,
"learning_rate": 0.00025287057696446105,
"loss": 1.982829475402832,
"step": 20420
},
{
"epoch": 2.614912,
"grad_norm": 0.5214614868164062,
"learning_rate": 0.0002528265597174264,
"loss": 1.9859834671020509,
"step": 20430
},
{
"epoch": 2.616192,
"grad_norm": 0.5939595103263855,
"learning_rate": 0.00025278252575965137,
"loss": 1.9625492095947266,
"step": 20440
},
{
"epoch": 2.6174720000000002,
"grad_norm": 0.6118872761726379,
"learning_rate": 0.00025273847509829194,
"loss": 2.005154037475586,
"step": 20450
},
{
"epoch": 2.6187519999999997,
"grad_norm": 0.5511441826820374,
"learning_rate": 0.0002526944077405072,
"loss": 1.9900339126586915,
"step": 20460
},
{
"epoch": 2.620032,
"grad_norm": 0.570794403553009,
"learning_rate": 0.00025265032369345853,
"loss": 1.9650077819824219,
"step": 20470
},
{
"epoch": 2.621312,
"grad_norm": 0.6029294729232788,
"learning_rate": 0.0002526062229643102,
"loss": 1.9891155242919922,
"step": 20480
},
{
"epoch": 2.622592,
"grad_norm": 0.5444130897521973,
"learning_rate": 0.0002525621055602293,
"loss": 2.0081520080566406,
"step": 20490
},
{
"epoch": 2.623872,
"grad_norm": 0.597112238407135,
"learning_rate": 0.0002525179714883854,
"loss": 2.004727363586426,
"step": 20500
},
{
"epoch": 2.625152,
"grad_norm": 0.7724485993385315,
"learning_rate": 0.00025247382075595096,
"loss": 1.9853918075561523,
"step": 20510
},
{
"epoch": 2.626432,
"grad_norm": 0.5809451937675476,
"learning_rate": 0.0002524296533701011,
"loss": 1.9954465866088866,
"step": 20520
},
{
"epoch": 2.627712,
"grad_norm": 0.6128025650978088,
"learning_rate": 0.0002523854693380136,
"loss": 1.992453384399414,
"step": 20530
},
{
"epoch": 2.628992,
"grad_norm": 0.7289356589317322,
"learning_rate": 0.000252341268666869,
"loss": 1.9801334381103515,
"step": 20540
},
{
"epoch": 2.630272,
"grad_norm": 0.5952761173248291,
"learning_rate": 0.00025229705136385047,
"loss": 1.9824865341186524,
"step": 20550
},
{
"epoch": 2.631552,
"grad_norm": 0.8996570110321045,
"learning_rate": 0.00025225281743614396,
"loss": 1.9769027709960938,
"step": 20560
},
{
"epoch": 2.632832,
"grad_norm": 0.6465076208114624,
"learning_rate": 0.00025220856689093813,
"loss": 2.0001235961914063,
"step": 20570
},
{
"epoch": 2.634112,
"grad_norm": 0.5432006120681763,
"learning_rate": 0.00025216429973542434,
"loss": 2.004485321044922,
"step": 20580
},
{
"epoch": 2.635392,
"grad_norm": 0.6515646576881409,
"learning_rate": 0.0002521200159767965,
"loss": 1.987103271484375,
"step": 20590
},
{
"epoch": 2.636672,
"grad_norm": 0.5826529264450073,
"learning_rate": 0.00025207571562225144,
"loss": 1.9816333770751953,
"step": 20600
},
{
"epoch": 2.637952,
"grad_norm": 0.5632296204566956,
"learning_rate": 0.0002520313986789885,
"loss": 1.9864912033081055,
"step": 20610
},
{
"epoch": 2.639232,
"grad_norm": 0.6872408390045166,
"learning_rate": 0.0002519870651542098,
"loss": 1.9885236740112304,
"step": 20620
},
{
"epoch": 2.640512,
"grad_norm": 0.5792223215103149,
"learning_rate": 0.0002519427150551202,
"loss": 1.9865676879882812,
"step": 20630
},
{
"epoch": 2.641792,
"grad_norm": 0.8562381863594055,
"learning_rate": 0.00025189834838892726,
"loss": 2.0032394409179686,
"step": 20640
},
{
"epoch": 2.643072,
"grad_norm": 0.5885404944419861,
"learning_rate": 0.00025185396516284105,
"loss": 1.9671829223632813,
"step": 20650
},
{
"epoch": 2.644352,
"grad_norm": 0.8070769309997559,
"learning_rate": 0.0002518095653840745,
"loss": 1.9936813354492187,
"step": 20660
},
{
"epoch": 2.645632,
"grad_norm": 0.591904878616333,
"learning_rate": 0.0002517651490598433,
"loss": 1.9989818572998046,
"step": 20670
},
{
"epoch": 2.646912,
"grad_norm": 0.6247791647911072,
"learning_rate": 0.0002517207161973655,
"loss": 2.0011875152587892,
"step": 20680
},
{
"epoch": 2.648192,
"grad_norm": 0.6229888200759888,
"learning_rate": 0.00025167626680386225,
"loss": 1.9874843597412108,
"step": 20690
},
{
"epoch": 2.6494720000000003,
"grad_norm": 0.6456174254417419,
"learning_rate": 0.0002516318008865571,
"loss": 2.0068422317504884,
"step": 20700
},
{
"epoch": 2.6507519999999998,
"grad_norm": 0.7064493894577026,
"learning_rate": 0.0002515873184526764,
"loss": 1.9966400146484375,
"step": 20710
},
{
"epoch": 2.652032,
"grad_norm": 0.5477823615074158,
"learning_rate": 0.0002515428195094492,
"loss": 1.9846370697021485,
"step": 20720
},
{
"epoch": 2.653312,
"grad_norm": 0.6381616592407227,
"learning_rate": 0.00025149830406410713,
"loss": 2.004497528076172,
"step": 20730
},
{
"epoch": 2.654592,
"grad_norm": 0.608198881149292,
"learning_rate": 0.0002514537721238846,
"loss": 2.005198860168457,
"step": 20740
},
{
"epoch": 2.655872,
"grad_norm": 0.5920321345329285,
"learning_rate": 0.00025140922369601865,
"loss": 1.9849420547485352,
"step": 20750
},
{
"epoch": 2.657152,
"grad_norm": 0.5282834768295288,
"learning_rate": 0.000251364658787749,
"loss": 1.9829280853271485,
"step": 20760
},
{
"epoch": 2.658432,
"grad_norm": 0.6273874640464783,
"learning_rate": 0.0002513200774063181,
"loss": 1.9876821517944336,
"step": 20770
},
{
"epoch": 2.659712,
"grad_norm": 0.5633535385131836,
"learning_rate": 0.00025127547955897106,
"loss": 1.9835145950317383,
"step": 20780
},
{
"epoch": 2.6609920000000002,
"grad_norm": 0.5835238695144653,
"learning_rate": 0.0002512308652529556,
"loss": 1.9885004043579102,
"step": 20790
},
{
"epoch": 2.6622719999999997,
"grad_norm": 0.5682324767112732,
"learning_rate": 0.0002511862344955222,
"loss": 1.9821138381958008,
"step": 20800
},
{
"epoch": 2.663552,
"grad_norm": 0.5757207870483398,
"learning_rate": 0.00025114158729392397,
"loss": 1.9952198028564454,
"step": 20810
},
{
"epoch": 2.664832,
"grad_norm": 0.6029953956604004,
"learning_rate": 0.0002510969236554167,
"loss": 1.9975730895996093,
"step": 20820
},
{
"epoch": 2.666112,
"grad_norm": 0.6032052040100098,
"learning_rate": 0.0002510522435872588,
"loss": 1.9650405883789062,
"step": 20830
},
{
"epoch": 2.667392,
"grad_norm": 0.6359588503837585,
"learning_rate": 0.00025100754709671144,
"loss": 1.9740261077880858,
"step": 20840
},
{
"epoch": 2.668672,
"grad_norm": 0.5589191317558289,
"learning_rate": 0.00025096283419103843,
"loss": 1.9808784484863282,
"step": 20850
},
{
"epoch": 2.669952,
"grad_norm": 0.5811957716941833,
"learning_rate": 0.00025091810487750624,
"loss": 1.9942428588867187,
"step": 20860
},
{
"epoch": 2.671232,
"grad_norm": 0.5617012977600098,
"learning_rate": 0.00025087335916338397,
"loss": 1.9816143035888671,
"step": 20870
},
{
"epoch": 2.672512,
"grad_norm": 0.5664853453636169,
"learning_rate": 0.00025082859705594345,
"loss": 1.9817010879516601,
"step": 20880
},
{
"epoch": 2.6737919999999997,
"grad_norm": 0.6202268600463867,
"learning_rate": 0.00025078381856245916,
"loss": 1.9885351181030273,
"step": 20890
},
{
"epoch": 2.675072,
"grad_norm": 0.6660192608833313,
"learning_rate": 0.00025073902369020815,
"loss": 1.9655139923095704,
"step": 20900
},
{
"epoch": 2.676352,
"grad_norm": 0.5392022132873535,
"learning_rate": 0.00025069421244647023,
"loss": 1.9887245178222657,
"step": 20910
},
{
"epoch": 2.677632,
"grad_norm": 0.6066935658454895,
"learning_rate": 0.00025064938483852794,
"loss": 1.9802143096923828,
"step": 20920
},
{
"epoch": 2.678912,
"grad_norm": 0.6226590871810913,
"learning_rate": 0.00025060454087366625,
"loss": 1.9787609100341796,
"step": 20930
},
{
"epoch": 2.680192,
"grad_norm": 0.672764241695404,
"learning_rate": 0.00025055968055917306,
"loss": 1.9923643112182616,
"step": 20940
},
{
"epoch": 2.681472,
"grad_norm": 0.5383614897727966,
"learning_rate": 0.00025051480390233865,
"loss": 1.9703481674194336,
"step": 20950
},
{
"epoch": 2.682752,
"grad_norm": 0.5925970673561096,
"learning_rate": 0.00025046991091045624,
"loss": 1.9890668869018555,
"step": 20960
},
{
"epoch": 2.684032,
"grad_norm": 0.6486144661903381,
"learning_rate": 0.00025042500159082147,
"loss": 1.9898494720458983,
"step": 20970
},
{
"epoch": 2.685312,
"grad_norm": 0.8324211835861206,
"learning_rate": 0.00025038007595073276,
"loss": 1.9953990936279298,
"step": 20980
},
{
"epoch": 2.686592,
"grad_norm": 0.645918607711792,
"learning_rate": 0.0002503351339974911,
"loss": 1.9979892730712892,
"step": 20990
},
{
"epoch": 2.687872,
"grad_norm": 0.5677788257598877,
"learning_rate": 0.00025029017573840025,
"loss": 2.006285285949707,
"step": 21000
},
{
"epoch": 2.689152,
"grad_norm": 0.6017535924911499,
"learning_rate": 0.00025024520118076656,
"loss": 1.979229736328125,
"step": 21010
},
{
"epoch": 2.690432,
"grad_norm": 0.582040011882782,
"learning_rate": 0.0002502002103318989,
"loss": 2.004988670349121,
"step": 21020
},
{
"epoch": 2.691712,
"grad_norm": 0.6636188626289368,
"learning_rate": 0.000250155203199109,
"loss": 2.0048398971557617,
"step": 21030
},
{
"epoch": 2.6929920000000003,
"grad_norm": 0.5977628231048584,
"learning_rate": 0.00025011017978971117,
"loss": 1.9760339736938477,
"step": 21040
},
{
"epoch": 2.694272,
"grad_norm": 0.5863640904426575,
"learning_rate": 0.0002500651401110223,
"loss": 2.0042751312255858,
"step": 21050
},
{
"epoch": 2.695552,
"grad_norm": 0.5521929264068604,
"learning_rate": 0.00025002008417036197,
"loss": 1.9981359481811523,
"step": 21060
},
{
"epoch": 2.696832,
"grad_norm": 0.5774256587028503,
"learning_rate": 0.00024997501197505234,
"loss": 2.002712059020996,
"step": 21070
},
{
"epoch": 2.698112,
"grad_norm": 0.6083162426948547,
"learning_rate": 0.00024992992353241837,
"loss": 1.99127254486084,
"step": 21080
},
{
"epoch": 2.699392,
"grad_norm": 0.5739796161651611,
"learning_rate": 0.0002498848188497874,
"loss": 1.9638822555541993,
"step": 21090
},
{
"epoch": 2.700672,
"grad_norm": 0.5196062326431274,
"learning_rate": 0.00024983969793448973,
"loss": 1.9926578521728515,
"step": 21100
},
{
"epoch": 2.701952,
"grad_norm": 0.5688360929489136,
"learning_rate": 0.0002497945607938581,
"loss": 1.9990943908691405,
"step": 21110
},
{
"epoch": 2.703232,
"grad_norm": 0.570843517780304,
"learning_rate": 0.0002497494074352278,
"loss": 1.9999536514282226,
"step": 21120
},
{
"epoch": 2.7045120000000002,
"grad_norm": 0.6709941029548645,
"learning_rate": 0.000249704237865937,
"loss": 1.978362464904785,
"step": 21130
},
{
"epoch": 2.7057919999999998,
"grad_norm": 0.6102050542831421,
"learning_rate": 0.00024965905209332635,
"loss": 1.9959636688232423,
"step": 21140
},
{
"epoch": 2.707072,
"grad_norm": 0.5855663418769836,
"learning_rate": 0.00024961385012473914,
"loss": 1.9774696350097656,
"step": 21150
},
{
"epoch": 2.708352,
"grad_norm": 0.5771845579147339,
"learning_rate": 0.00024956863196752137,
"loss": 1.9720075607299805,
"step": 21160
},
{
"epoch": 2.709632,
"grad_norm": 0.5946242809295654,
"learning_rate": 0.00024952339762902154,
"loss": 1.9740718841552733,
"step": 21170
},
{
"epoch": 2.710912,
"grad_norm": 0.5482192635536194,
"learning_rate": 0.0002494781471165909,
"loss": 1.9831159591674805,
"step": 21180
},
{
"epoch": 2.712192,
"grad_norm": 0.5820874571800232,
"learning_rate": 0.00024943288043758336,
"loss": 2.0058475494384767,
"step": 21190
},
{
"epoch": 2.713472,
"grad_norm": 0.5787574648857117,
"learning_rate": 0.0002493875975993552,
"loss": 1.9697856903076172,
"step": 21200
},
{
"epoch": 2.714752,
"grad_norm": 0.5676509141921997,
"learning_rate": 0.0002493422986092656,
"loss": 1.9971519470214845,
"step": 21210
},
{
"epoch": 2.7160320000000002,
"grad_norm": 0.5169781446456909,
"learning_rate": 0.00024929698347467635,
"loss": 1.9816612243652343,
"step": 21220
},
{
"epoch": 2.7173119999999997,
"grad_norm": 0.7070567011833191,
"learning_rate": 0.0002492516522029518,
"loss": 1.9721418380737306,
"step": 21230
},
{
"epoch": 2.718592,
"grad_norm": 0.5570013523101807,
"learning_rate": 0.0002492063048014587,
"loss": 1.9668373107910155,
"step": 21240
},
{
"epoch": 2.719872,
"grad_norm": 0.5987828969955444,
"learning_rate": 0.0002491609412775668,
"loss": 1.983280563354492,
"step": 21250
},
{
"epoch": 2.721152,
"grad_norm": 0.5788903832435608,
"learning_rate": 0.0002491155616386483,
"loss": 1.989954376220703,
"step": 21260
},
{
"epoch": 2.722432,
"grad_norm": 0.6754418015480042,
"learning_rate": 0.00024907016589207796,
"loss": 1.994033432006836,
"step": 21270
},
{
"epoch": 2.723712,
"grad_norm": 0.6426966786384583,
"learning_rate": 0.0002490247540452332,
"loss": 2.0031221389770506,
"step": 21280
},
{
"epoch": 2.724992,
"grad_norm": 0.6188598871231079,
"learning_rate": 0.00024897932610549415,
"loss": 1.9844345092773437,
"step": 21290
},
{
"epoch": 2.726272,
"grad_norm": 0.598725438117981,
"learning_rate": 0.0002489338820802435,
"loss": 1.9899524688720702,
"step": 21300
},
{
"epoch": 2.727552,
"grad_norm": 0.5706913471221924,
"learning_rate": 0.0002488884219768664,
"loss": 2.0209716796875,
"step": 21310
},
{
"epoch": 2.728832,
"grad_norm": 0.614439070224762,
"learning_rate": 0.0002488429458027509,
"loss": 1.9777687072753907,
"step": 21320
},
{
"epoch": 2.730112,
"grad_norm": 0.6483980417251587,
"learning_rate": 0.00024879745356528736,
"loss": 2.0050498962402346,
"step": 21330
},
{
"epoch": 2.731392,
"grad_norm": 0.6259683966636658,
"learning_rate": 0.000248751945271869,
"loss": 1.958902931213379,
"step": 21340
},
{
"epoch": 2.732672,
"grad_norm": 0.5778680443763733,
"learning_rate": 0.00024870642092989154,
"loss": 1.9971967697143556,
"step": 21350
},
{
"epoch": 2.733952,
"grad_norm": 0.5903841257095337,
"learning_rate": 0.0002486608805467533,
"loss": 1.9903545379638672,
"step": 21360
},
{
"epoch": 2.735232,
"grad_norm": 0.6269921064376831,
"learning_rate": 0.0002486153241298552,
"loss": 1.9959684371948243,
"step": 21370
},
{
"epoch": 2.736512,
"grad_norm": 0.5614582300186157,
"learning_rate": 0.0002485697516866008,
"loss": 1.9801317214965821,
"step": 21380
},
{
"epoch": 2.737792,
"grad_norm": 0.5842646956443787,
"learning_rate": 0.0002485241632243963,
"loss": 1.9919420242309571,
"step": 21390
},
{
"epoch": 2.739072,
"grad_norm": 0.6948349475860596,
"learning_rate": 0.0002484785587506504,
"loss": 1.9789737701416015,
"step": 21400
},
{
"epoch": 2.740352,
"grad_norm": 0.5703866481781006,
"learning_rate": 0.0002484329382727745,
"loss": 2.0018083572387697,
"step": 21410
},
{
"epoch": 2.741632,
"grad_norm": 0.60796719789505,
"learning_rate": 0.0002483873017981825,
"loss": 1.9723936080932618,
"step": 21420
},
{
"epoch": 2.742912,
"grad_norm": 0.5989227890968323,
"learning_rate": 0.00024834164933429105,
"loss": 1.9768259048461914,
"step": 21430
},
{
"epoch": 2.744192,
"grad_norm": 0.5481821894645691,
"learning_rate": 0.0002482959808885192,
"loss": 1.9955833435058594,
"step": 21440
},
{
"epoch": 2.745472,
"grad_norm": 0.5941957235336304,
"learning_rate": 0.0002482502964682888,
"loss": 1.9914913177490234,
"step": 21450
},
{
"epoch": 2.746752,
"grad_norm": 0.6133338809013367,
"learning_rate": 0.00024820459608102414,
"loss": 2.0105052947998048,
"step": 21460
},
{
"epoch": 2.7480320000000003,
"grad_norm": 0.6438705921173096,
"learning_rate": 0.00024815887973415223,
"loss": 2.0010025024414064,
"step": 21470
},
{
"epoch": 2.7493119999999998,
"grad_norm": 0.5801704525947571,
"learning_rate": 0.0002481131474351025,
"loss": 1.9993968963623048,
"step": 21480
},
{
"epoch": 2.750592,
"grad_norm": 0.640665590763092,
"learning_rate": 0.0002480673991913072,
"loss": 1.9930713653564454,
"step": 21490
},
{
"epoch": 2.751872,
"grad_norm": 0.6179438829421997,
"learning_rate": 0.000248021635010201,
"loss": 1.9839179992675782,
"step": 21500
},
{
"epoch": 2.753152,
"grad_norm": 0.5531506538391113,
"learning_rate": 0.0002479758548992212,
"loss": 1.987672996520996,
"step": 21510
},
{
"epoch": 2.754432,
"grad_norm": 0.6056403517723083,
"learning_rate": 0.00024793005886580773,
"loss": 1.9669315338134765,
"step": 21520
},
{
"epoch": 2.755712,
"grad_norm": 0.5558757781982422,
"learning_rate": 0.000247884246917403,
"loss": 1.9910009384155274,
"step": 21530
},
{
"epoch": 2.756992,
"grad_norm": 0.5062780976295471,
"learning_rate": 0.00024783841906145227,
"loss": 1.9710603713989259,
"step": 21540
},
{
"epoch": 2.758272,
"grad_norm": 0.5504339933395386,
"learning_rate": 0.000247792575305403,
"loss": 1.9982128143310547,
"step": 21550
},
{
"epoch": 2.7595520000000002,
"grad_norm": 0.5725201964378357,
"learning_rate": 0.0002477467156567055,
"loss": 1.978424644470215,
"step": 21560
},
{
"epoch": 2.7608319999999997,
"grad_norm": 0.6802664399147034,
"learning_rate": 0.00024770084012281274,
"loss": 1.9947689056396485,
"step": 21570
},
{
"epoch": 2.762112,
"grad_norm": 0.5704305768013,
"learning_rate": 0.0002476549487111799,
"loss": 1.9845502853393555,
"step": 21580
},
{
"epoch": 2.763392,
"grad_norm": 0.558707594871521,
"learning_rate": 0.00024760904142926516,
"loss": 1.9857442855834961,
"step": 21590
},
{
"epoch": 2.764672,
"grad_norm": 0.6300362944602966,
"learning_rate": 0.000247563118284529,
"loss": 1.9798593521118164,
"step": 21600
},
{
"epoch": 2.765952,
"grad_norm": 0.5548649430274963,
"learning_rate": 0.0002475171792844345,
"loss": 1.9844810485839843,
"step": 21610
},
{
"epoch": 2.767232,
"grad_norm": 0.595108151435852,
"learning_rate": 0.00024747122443644754,
"loss": 1.9949005126953125,
"step": 21620
},
{
"epoch": 2.768512,
"grad_norm": 0.6572628617286682,
"learning_rate": 0.00024742525374803627,
"loss": 1.9751409530639648,
"step": 21630
},
{
"epoch": 2.769792,
"grad_norm": 0.5465295910835266,
"learning_rate": 0.0002473792672266717,
"loss": 1.9829030990600587,
"step": 21640
},
{
"epoch": 2.771072,
"grad_norm": 0.643868088722229,
"learning_rate": 0.00024733326487982714,
"loss": 1.9697589874267578,
"step": 21650
},
{
"epoch": 2.772352,
"grad_norm": 0.6807072162628174,
"learning_rate": 0.00024728724671497875,
"loss": 1.9979555130004882,
"step": 21660
},
{
"epoch": 2.773632,
"grad_norm": 0.538885772228241,
"learning_rate": 0.000247241212739605,
"loss": 2.0033884048461914,
"step": 21670
},
{
"epoch": 2.774912,
"grad_norm": 0.6614050269126892,
"learning_rate": 0.0002471951629611871,
"loss": 1.9835277557373048,
"step": 21680
},
{
"epoch": 2.776192,
"grad_norm": 0.5746902823448181,
"learning_rate": 0.0002471490973872088,
"loss": 1.9843767166137696,
"step": 21690
},
{
"epoch": 2.777472,
"grad_norm": 0.643732488155365,
"learning_rate": 0.0002471030160251564,
"loss": 1.9837514877319335,
"step": 21700
},
{
"epoch": 2.778752,
"grad_norm": 0.6497176885604858,
"learning_rate": 0.0002470569188825186,
"loss": 1.9753828048706055,
"step": 21710
},
{
"epoch": 2.780032,
"grad_norm": 0.6284911632537842,
"learning_rate": 0.00024701080596678705,
"loss": 2.0010009765625,
"step": 21720
},
{
"epoch": 2.781312,
"grad_norm": 0.646207332611084,
"learning_rate": 0.00024696467728545565,
"loss": 2.000373649597168,
"step": 21730
},
{
"epoch": 2.782592,
"grad_norm": 0.8075646758079529,
"learning_rate": 0.0002469185328460209,
"loss": 1.9939256668090821,
"step": 21740
},
{
"epoch": 2.783872,
"grad_norm": 0.6705852150917053,
"learning_rate": 0.0002468723726559819,
"loss": 1.9859939575195313,
"step": 21750
},
{
"epoch": 2.785152,
"grad_norm": 0.5609583258628845,
"learning_rate": 0.00024682619672284045,
"loss": 1.970254898071289,
"step": 21760
},
{
"epoch": 2.786432,
"grad_norm": 0.6125978231430054,
"learning_rate": 0.00024678000505410074,
"loss": 1.9793630599975587,
"step": 21770
},
{
"epoch": 2.787712,
"grad_norm": 0.7464768886566162,
"learning_rate": 0.0002467337976572694,
"loss": 1.9855781555175782,
"step": 21780
},
{
"epoch": 2.788992,
"grad_norm": 0.6230854392051697,
"learning_rate": 0.00024668757453985595,
"loss": 1.9919401168823243,
"step": 21790
},
{
"epoch": 2.790272,
"grad_norm": 0.6750345230102539,
"learning_rate": 0.0002466413357093722,
"loss": 2.001689910888672,
"step": 21800
},
{
"epoch": 2.7915520000000003,
"grad_norm": 0.5662205219268799,
"learning_rate": 0.0002465950811733326,
"loss": 1.9804395675659179,
"step": 21810
},
{
"epoch": 2.7928319999999998,
"grad_norm": 0.5635572075843811,
"learning_rate": 0.00024654881093925425,
"loss": 1.9903745651245117,
"step": 21820
},
{
"epoch": 2.794112,
"grad_norm": 0.596569299697876,
"learning_rate": 0.0002465025250146566,
"loss": 1.9859878540039062,
"step": 21830
},
{
"epoch": 2.795392,
"grad_norm": 0.5884014964103699,
"learning_rate": 0.00024645622340706177,
"loss": 1.9819086074829102,
"step": 21840
},
{
"epoch": 2.796672,
"grad_norm": 0.6158134937286377,
"learning_rate": 0.0002464099061239944,
"loss": 1.9932640075683594,
"step": 21850
},
{
"epoch": 2.797952,
"grad_norm": 0.5887981057167053,
"learning_rate": 0.0002463635731729818,
"loss": 1.9920537948608399,
"step": 21860
},
{
"epoch": 2.799232,
"grad_norm": 0.5190674662590027,
"learning_rate": 0.00024631722456155356,
"loss": 2.004105567932129,
"step": 21870
},
{
"epoch": 2.800512,
"grad_norm": 0.6308513283729553,
"learning_rate": 0.00024627086029724207,
"loss": 1.9859233856201173,
"step": 21880
},
{
"epoch": 2.801792,
"grad_norm": 0.5891023874282837,
"learning_rate": 0.00024622448038758216,
"loss": 1.977734375,
"step": 21890
},
{
"epoch": 2.8030720000000002,
"grad_norm": 0.5666506290435791,
"learning_rate": 0.0002461780848401112,
"loss": 1.977214241027832,
"step": 21900
},
{
"epoch": 2.8043519999999997,
"grad_norm": 0.5679246187210083,
"learning_rate": 0.000246131673662369,
"loss": 1.9786552429199218,
"step": 21910
},
{
"epoch": 2.805632,
"grad_norm": 0.6503104567527771,
"learning_rate": 0.0002460852468618982,
"loss": 1.9844184875488282,
"step": 21920
},
{
"epoch": 2.806912,
"grad_norm": 0.582848072052002,
"learning_rate": 0.00024603880444624374,
"loss": 2.0020204544067384,
"step": 21930
},
{
"epoch": 2.808192,
"grad_norm": 0.5356834530830383,
"learning_rate": 0.0002459923464229531,
"loss": 1.9909164428710937,
"step": 21940
},
{
"epoch": 2.809472,
"grad_norm": 0.6107462644577026,
"learning_rate": 0.00024594587279957634,
"loss": 1.9916419982910156,
"step": 21950
},
{
"epoch": 2.810752,
"grad_norm": 0.5901324152946472,
"learning_rate": 0.0002458993835836662,
"loss": 1.9786615371704102,
"step": 21960
},
{
"epoch": 2.812032,
"grad_norm": 0.5983129143714905,
"learning_rate": 0.00024585287878277763,
"loss": 1.9897510528564453,
"step": 21970
},
{
"epoch": 2.813312,
"grad_norm": 0.5816282629966736,
"learning_rate": 0.0002458063584044685,
"loss": 1.981691551208496,
"step": 21980
},
{
"epoch": 2.814592,
"grad_norm": 0.5935829877853394,
"learning_rate": 0.00024575982245629883,
"loss": 1.981523323059082,
"step": 21990
},
{
"epoch": 2.815872,
"grad_norm": 0.5594017505645752,
"learning_rate": 0.00024571327094583147,
"loss": 1.9830625534057618,
"step": 22000
},
{
"epoch": 2.817152,
"grad_norm": 0.5698060393333435,
"learning_rate": 0.00024566670388063175,
"loss": 1.9858173370361327,
"step": 22010
},
{
"epoch": 2.818432,
"grad_norm": 0.5697092413902283,
"learning_rate": 0.0002456201212682673,
"loss": 1.9921161651611328,
"step": 22020
},
{
"epoch": 2.819712,
"grad_norm": 0.6146735548973083,
"learning_rate": 0.00024557352311630853,
"loss": 2.009439468383789,
"step": 22030
},
{
"epoch": 2.820992,
"grad_norm": 0.6026655435562134,
"learning_rate": 0.0002455269094323283,
"loss": 1.9936819076538086,
"step": 22040
},
{
"epoch": 2.822272,
"grad_norm": 0.604644238948822,
"learning_rate": 0.00024548028022390193,
"loss": 1.999912643432617,
"step": 22050
},
{
"epoch": 2.823552,
"grad_norm": 0.6419292688369751,
"learning_rate": 0.00024543363549860733,
"loss": 1.9958415985107423,
"step": 22060
},
{
"epoch": 2.824832,
"grad_norm": 0.506291389465332,
"learning_rate": 0.0002453869752640249,
"loss": 1.9692180633544922,
"step": 22070
},
{
"epoch": 2.826112,
"grad_norm": 0.5913361310958862,
"learning_rate": 0.00024534029952773766,
"loss": 2.0046966552734373,
"step": 22080
},
{
"epoch": 2.827392,
"grad_norm": 0.5525562763214111,
"learning_rate": 0.00024529360829733096,
"loss": 1.9856571197509765,
"step": 22090
},
{
"epoch": 2.828672,
"grad_norm": 0.7101224064826965,
"learning_rate": 0.00024524690158039277,
"loss": 2.0003843307495117,
"step": 22100
},
{
"epoch": 2.829952,
"grad_norm": 0.5780585408210754,
"learning_rate": 0.00024520017938451364,
"loss": 1.9761104583740234,
"step": 22110
},
{
"epoch": 2.831232,
"grad_norm": 0.6270367503166199,
"learning_rate": 0.00024515344171728657,
"loss": 1.9795106887817382,
"step": 22120
},
{
"epoch": 2.832512,
"grad_norm": 0.6195119619369507,
"learning_rate": 0.0002451066885863071,
"loss": 1.9822874069213867,
"step": 22130
},
{
"epoch": 2.833792,
"grad_norm": 0.5710272789001465,
"learning_rate": 0.0002450599199991732,
"loss": 2.003024864196777,
"step": 22140
},
{
"epoch": 2.8350720000000003,
"grad_norm": 0.5935205221176147,
"learning_rate": 0.00024501313596348535,
"loss": 1.9804107666015625,
"step": 22150
},
{
"epoch": 2.8363519999999998,
"grad_norm": 0.6654118895530701,
"learning_rate": 0.00024496633648684674,
"loss": 2.0020801544189455,
"step": 22160
},
{
"epoch": 2.837632,
"grad_norm": 0.6343352198600769,
"learning_rate": 0.00024491952157686293,
"loss": 1.9739692687988282,
"step": 22170
},
{
"epoch": 2.838912,
"grad_norm": 0.6836406588554382,
"learning_rate": 0.00024487269124114184,
"loss": 1.968768310546875,
"step": 22180
},
{
"epoch": 2.840192,
"grad_norm": 0.6519709825515747,
"learning_rate": 0.0002448258454872943,
"loss": 1.9951820373535156,
"step": 22190
},
{
"epoch": 2.841472,
"grad_norm": 0.6881632804870605,
"learning_rate": 0.0002447789843229331,
"loss": 1.9956289291381837,
"step": 22200
},
{
"epoch": 2.842752,
"grad_norm": 0.5613313913345337,
"learning_rate": 0.000244732107755674,
"loss": 1.9837509155273438,
"step": 22210
},
{
"epoch": 2.844032,
"grad_norm": 0.553723156452179,
"learning_rate": 0.0002446852157931351,
"loss": 1.974177932739258,
"step": 22220
},
{
"epoch": 2.845312,
"grad_norm": 0.6348955631256104,
"learning_rate": 0.000244638308442937,
"loss": 1.9742591857910157,
"step": 22230
},
{
"epoch": 2.8465920000000002,
"grad_norm": 0.5995672941207886,
"learning_rate": 0.0002445913857127027,
"loss": 1.9930797576904298,
"step": 22240
},
{
"epoch": 2.8478719999999997,
"grad_norm": 0.5643102526664734,
"learning_rate": 0.0002445444476100579,
"loss": 1.980391502380371,
"step": 22250
},
{
"epoch": 2.849152,
"grad_norm": 0.5699091553688049,
"learning_rate": 0.00024449749414263056,
"loss": 1.9759817123413086,
"step": 22260
},
{
"epoch": 2.850432,
"grad_norm": 0.6020393967628479,
"learning_rate": 0.0002444505253180514,
"loss": 1.9716245651245117,
"step": 22270
},
{
"epoch": 2.851712,
"grad_norm": 0.5212053060531616,
"learning_rate": 0.00024440354114395335,
"loss": 1.9582582473754884,
"step": 22280
},
{
"epoch": 2.852992,
"grad_norm": 0.6842744946479797,
"learning_rate": 0.0002443565416279722,
"loss": 1.9844417572021484,
"step": 22290
},
{
"epoch": 2.854272,
"grad_norm": 0.7084447145462036,
"learning_rate": 0.00024430952677774594,
"loss": 1.9883953094482423,
"step": 22300
},
{
"epoch": 2.855552,
"grad_norm": 0.5122093558311462,
"learning_rate": 0.00024426249660091504,
"loss": 1.9889739990234374,
"step": 22310
},
{
"epoch": 2.856832,
"grad_norm": 0.6394441723823547,
"learning_rate": 0.0002442154511051226,
"loss": 1.9776674270629884,
"step": 22320
},
{
"epoch": 2.858112,
"grad_norm": 0.5593616366386414,
"learning_rate": 0.00024416839029801424,
"loss": 1.9840169906616212,
"step": 22330
},
{
"epoch": 2.859392,
"grad_norm": 0.6495934128761292,
"learning_rate": 0.00024412131418723794,
"loss": 1.992612075805664,
"step": 22340
},
{
"epoch": 2.860672,
"grad_norm": 0.6087314486503601,
"learning_rate": 0.0002440742227804442,
"loss": 1.9710994720458985,
"step": 22350
},
{
"epoch": 2.861952,
"grad_norm": 0.5750599503517151,
"learning_rate": 0.00024402711608528604,
"loss": 1.9617900848388672,
"step": 22360
},
{
"epoch": 2.863232,
"grad_norm": 0.5630548596382141,
"learning_rate": 0.00024397999410941899,
"loss": 1.9829639434814452,
"step": 22370
},
{
"epoch": 2.864512,
"grad_norm": 0.5493993163108826,
"learning_rate": 0.00024393285686050096,
"loss": 1.9779756546020508,
"step": 22380
},
{
"epoch": 2.865792,
"grad_norm": 0.6558434963226318,
"learning_rate": 0.00024388570434619243,
"loss": 1.989352035522461,
"step": 22390
},
{
"epoch": 2.867072,
"grad_norm": 0.6214169263839722,
"learning_rate": 0.00024383853657415638,
"loss": 2.0066959381103517,
"step": 22400
},
{
"epoch": 2.868352,
"grad_norm": 0.6326498985290527,
"learning_rate": 0.00024379135355205822,
"loss": 1.9828102111816406,
"step": 22410
},
{
"epoch": 2.869632,
"grad_norm": 0.5637285113334656,
"learning_rate": 0.00024374415528756574,
"loss": 2.001946449279785,
"step": 22420
},
{
"epoch": 2.870912,
"grad_norm": 0.5488138198852539,
"learning_rate": 0.0002436969417883494,
"loss": 1.990394401550293,
"step": 22430
},
{
"epoch": 2.872192,
"grad_norm": 0.5735793709754944,
"learning_rate": 0.0002436497130620821,
"loss": 1.9791196823120116,
"step": 22440
},
{
"epoch": 2.873472,
"grad_norm": 0.5670044422149658,
"learning_rate": 0.00024360246911643904,
"loss": 1.9941360473632812,
"step": 22450
},
{
"epoch": 2.874752,
"grad_norm": 0.6228307485580444,
"learning_rate": 0.00024355520995909808,
"loss": 1.981285858154297,
"step": 22460
},
{
"epoch": 2.876032,
"grad_norm": 0.515998125076294,
"learning_rate": 0.00024350793559773946,
"loss": 1.9688381195068358,
"step": 22470
},
{
"epoch": 2.877312,
"grad_norm": 0.6016044616699219,
"learning_rate": 0.000243460646040046,
"loss": 1.997141456604004,
"step": 22480
},
{
"epoch": 2.8785920000000003,
"grad_norm": 0.5613240003585815,
"learning_rate": 0.0002434133412937028,
"loss": 2.011920928955078,
"step": 22490
},
{
"epoch": 2.8798719999999998,
"grad_norm": 0.6312174797058105,
"learning_rate": 0.00024336602136639757,
"loss": 1.9861148834228515,
"step": 22500
},
{
"epoch": 2.881152,
"grad_norm": 0.6346191763877869,
"learning_rate": 0.00024331868626582053,
"loss": 1.9831401824951171,
"step": 22510
},
{
"epoch": 2.882432,
"grad_norm": 0.568792998790741,
"learning_rate": 0.0002432713359996642,
"loss": 1.9808233261108399,
"step": 22520
},
{
"epoch": 2.883712,
"grad_norm": 0.6010581851005554,
"learning_rate": 0.00024322397057562365,
"loss": 1.9755977630615233,
"step": 22530
},
{
"epoch": 2.884992,
"grad_norm": 0.6198073625564575,
"learning_rate": 0.00024317659000139648,
"loss": 1.9872373580932616,
"step": 22540
},
{
"epoch": 2.886272,
"grad_norm": 0.5780988335609436,
"learning_rate": 0.00024312919428468266,
"loss": 1.994683074951172,
"step": 22550
},
{
"epoch": 2.887552,
"grad_norm": 0.6072776913642883,
"learning_rate": 0.00024308178343318464,
"loss": 1.9857389450073242,
"step": 22560
},
{
"epoch": 2.888832,
"grad_norm": 0.5802062749862671,
"learning_rate": 0.00024303435745460733,
"loss": 1.989776611328125,
"step": 22570
},
{
"epoch": 2.8901120000000002,
"grad_norm": 0.5251072645187378,
"learning_rate": 0.00024298691635665816,
"loss": 1.9787878036499023,
"step": 22580
},
{
"epoch": 2.8913919999999997,
"grad_norm": 0.5801512002944946,
"learning_rate": 0.00024293946014704692,
"loss": 1.9932672500610351,
"step": 22590
},
{
"epoch": 2.892672,
"grad_norm": 0.5265316963195801,
"learning_rate": 0.0002428919888334859,
"loss": 1.97564640045166,
"step": 22600
},
{
"epoch": 2.893952,
"grad_norm": 0.615715503692627,
"learning_rate": 0.0002428445024236899,
"loss": 1.9956838607788085,
"step": 22610
},
{
"epoch": 2.895232,
"grad_norm": 0.6207094788551331,
"learning_rate": 0.00024279700092537604,
"loss": 1.9927707672119142,
"step": 22620
},
{
"epoch": 2.896512,
"grad_norm": 0.5834910869598389,
"learning_rate": 0.00024274948434626405,
"loss": 1.9957403182983398,
"step": 22630
},
{
"epoch": 2.897792,
"grad_norm": 0.6218145489692688,
"learning_rate": 0.00024270195269407592,
"loss": 2.000038719177246,
"step": 22640
},
{
"epoch": 2.899072,
"grad_norm": 0.5531935691833496,
"learning_rate": 0.00024265440597653638,
"loss": 1.9788873672485352,
"step": 22650
},
{
"epoch": 2.900352,
"grad_norm": 0.6910541653633118,
"learning_rate": 0.00024260684420137227,
"loss": 1.9910291671752929,
"step": 22660
},
{
"epoch": 2.901632,
"grad_norm": 0.5196993947029114,
"learning_rate": 0.00024255926737631309,
"loss": 1.9908140182495118,
"step": 22670
},
{
"epoch": 2.902912,
"grad_norm": 0.5265659093856812,
"learning_rate": 0.0002425116755090908,
"loss": 1.9998836517333984,
"step": 22680
},
{
"epoch": 2.904192,
"grad_norm": 0.6247841119766235,
"learning_rate": 0.00024246406860743957,
"loss": 1.9943998336791993,
"step": 22690
},
{
"epoch": 2.905472,
"grad_norm": 0.5924451947212219,
"learning_rate": 0.00024241644667909634,
"loss": 2.0031448364257813,
"step": 22700
},
{
"epoch": 2.906752,
"grad_norm": 0.6125649809837341,
"learning_rate": 0.0002423688097318003,
"loss": 1.9700164794921875,
"step": 22710
},
{
"epoch": 2.908032,
"grad_norm": 0.5865424275398254,
"learning_rate": 0.00024232115777329303,
"loss": 1.99542179107666,
"step": 22720
},
{
"epoch": 2.909312,
"grad_norm": 0.6035805344581604,
"learning_rate": 0.00024227349081131874,
"loss": 1.9953996658325195,
"step": 22730
},
{
"epoch": 2.910592,
"grad_norm": 0.6540617346763611,
"learning_rate": 0.00024222580885362394,
"loss": 2.001887321472168,
"step": 22740
},
{
"epoch": 2.911872,
"grad_norm": 0.5476991534233093,
"learning_rate": 0.00024217811190795758,
"loss": 1.9748291015625,
"step": 22750
},
{
"epoch": 2.913152,
"grad_norm": 0.6497393846511841,
"learning_rate": 0.00024213039998207108,
"loss": 1.9617677688598634,
"step": 22760
},
{
"epoch": 2.914432,
"grad_norm": 0.60370272397995,
"learning_rate": 0.00024208267308371834,
"loss": 1.9807193756103516,
"step": 22770
},
{
"epoch": 2.915712,
"grad_norm": 0.5640918016433716,
"learning_rate": 0.0002420349312206556,
"loss": 1.9930435180664063,
"step": 22780
},
{
"epoch": 2.916992,
"grad_norm": 0.5358017086982727,
"learning_rate": 0.00024198717440064148,
"loss": 2.001434326171875,
"step": 22790
},
{
"epoch": 2.918272,
"grad_norm": 0.5495386123657227,
"learning_rate": 0.0002419394026314373,
"loss": 1.9928983688354491,
"step": 22800
},
{
"epoch": 2.919552,
"grad_norm": 0.5684537291526794,
"learning_rate": 0.00024189161592080654,
"loss": 1.9941551208496093,
"step": 22810
},
{
"epoch": 2.920832,
"grad_norm": 0.6296496391296387,
"learning_rate": 0.00024184381427651522,
"loss": 2.0040258407592773,
"step": 22820
},
{
"epoch": 2.9221120000000003,
"grad_norm": 0.5415356755256653,
"learning_rate": 0.00024179599770633177,
"loss": 1.987640380859375,
"step": 22830
},
{
"epoch": 2.9233919999999998,
"grad_norm": 0.5680559277534485,
"learning_rate": 0.00024174816621802707,
"loss": 1.9798744201660157,
"step": 22840
},
{
"epoch": 2.924672,
"grad_norm": 0.6271728873252869,
"learning_rate": 0.00024170031981937432,
"loss": 2.0287246704101562,
"step": 22850
},
{
"epoch": 2.925952,
"grad_norm": 0.6384287476539612,
"learning_rate": 0.00024165245851814933,
"loss": 2.0148162841796875,
"step": 22860
},
{
"epoch": 2.927232,
"grad_norm": 0.5833884477615356,
"learning_rate": 0.00024160458232213013,
"loss": 1.9851333618164062,
"step": 22870
},
{
"epoch": 2.928512,
"grad_norm": 0.6439580321311951,
"learning_rate": 0.0002415566912390973,
"loss": 1.988656997680664,
"step": 22880
},
{
"epoch": 2.929792,
"grad_norm": 0.6024416089057922,
"learning_rate": 0.00024150878527683383,
"loss": 1.9716609954833983,
"step": 22890
},
{
"epoch": 2.931072,
"grad_norm": 0.5979756116867065,
"learning_rate": 0.00024146086444312505,
"loss": 1.998563575744629,
"step": 22900
},
{
"epoch": 2.932352,
"grad_norm": 0.6395220756530762,
"learning_rate": 0.0002414129287457588,
"loss": 1.9841201782226563,
"step": 22910
},
{
"epoch": 2.9336320000000002,
"grad_norm": 0.5890491604804993,
"learning_rate": 0.00024136497819252531,
"loss": 1.9798524856567383,
"step": 22920
},
{
"epoch": 2.9349119999999997,
"grad_norm": 0.5945888161659241,
"learning_rate": 0.00024131701279121718,
"loss": 1.984349250793457,
"step": 22930
},
{
"epoch": 2.936192,
"grad_norm": 0.6907007098197937,
"learning_rate": 0.00024126903254962942,
"loss": 1.9607254028320313,
"step": 22940
},
{
"epoch": 2.937472,
"grad_norm": 0.5948235392570496,
"learning_rate": 0.0002412210374755595,
"loss": 2.0012508392333985,
"step": 22950
},
{
"epoch": 2.938752,
"grad_norm": 0.662884533405304,
"learning_rate": 0.0002411730275768074,
"loss": 1.9821485519409179,
"step": 22960
},
{
"epoch": 2.940032,
"grad_norm": 0.5196213126182556,
"learning_rate": 0.00024112500286117524,
"loss": 2.0012046813964846,
"step": 22970
},
{
"epoch": 2.941312,
"grad_norm": 0.5713062286376953,
"learning_rate": 0.0002410769633364678,
"loss": 1.9920314788818358,
"step": 22980
},
{
"epoch": 2.942592,
"grad_norm": 0.6510788798332214,
"learning_rate": 0.00024102890901049208,
"loss": 1.9986385345458983,
"step": 22990
},
{
"epoch": 2.943872,
"grad_norm": 0.5226752161979675,
"learning_rate": 0.00024098083989105762,
"loss": 2.0013992309570314,
"step": 23000
},
{
"epoch": 2.945152,
"grad_norm": 0.5960683226585388,
"learning_rate": 0.0002409327559859763,
"loss": 2.0002574920654297,
"step": 23010
},
{
"epoch": 2.946432,
"grad_norm": 0.6040253043174744,
"learning_rate": 0.0002408846573030625,
"loss": 2.007054901123047,
"step": 23020
},
{
"epoch": 2.947712,
"grad_norm": 0.6040448546409607,
"learning_rate": 0.00024083654385013284,
"loss": 1.9935993194580077,
"step": 23030
},
{
"epoch": 2.948992,
"grad_norm": 0.6118789315223694,
"learning_rate": 0.00024078841563500645,
"loss": 2.0116539001464844,
"step": 23040
},
{
"epoch": 2.950272,
"grad_norm": 0.5757555961608887,
"learning_rate": 0.00024074027266550481,
"loss": 1.9783689498901367,
"step": 23050
},
{
"epoch": 2.951552,
"grad_norm": 0.5779050588607788,
"learning_rate": 0.00024069211494945185,
"loss": 1.9797739028930663,
"step": 23060
},
{
"epoch": 2.952832,
"grad_norm": 0.6035221815109253,
"learning_rate": 0.00024064394249467385,
"loss": 2.0258762359619142,
"step": 23070
},
{
"epoch": 2.954112,
"grad_norm": 0.7022536396980286,
"learning_rate": 0.0002405957553089995,
"loss": 1.989544677734375,
"step": 23080
},
{
"epoch": 2.955392,
"grad_norm": 0.5598657727241516,
"learning_rate": 0.0002405475534002599,
"loss": 1.9677785873413085,
"step": 23090
},
{
"epoch": 2.956672,
"grad_norm": 0.5750641822814941,
"learning_rate": 0.0002404993367762885,
"loss": 1.972306251525879,
"step": 23100
},
{
"epoch": 2.957952,
"grad_norm": 0.6550465822219849,
"learning_rate": 0.00024045110544492117,
"loss": 2.0062767028808595,
"step": 23110
},
{
"epoch": 2.959232,
"grad_norm": 0.5702496767044067,
"learning_rate": 0.00024040285941399623,
"loss": 1.9841428756713868,
"step": 23120
},
{
"epoch": 2.960512,
"grad_norm": 0.6065656542778015,
"learning_rate": 0.00024035459869135425,
"loss": 1.989351463317871,
"step": 23130
},
{
"epoch": 2.961792,
"grad_norm": 0.5480714440345764,
"learning_rate": 0.0002403063232848383,
"loss": 2.003216361999512,
"step": 23140
},
{
"epoch": 2.963072,
"grad_norm": 0.5692582130432129,
"learning_rate": 0.00024025803320229384,
"loss": 1.9748903274536134,
"step": 23150
},
{
"epoch": 2.964352,
"grad_norm": 0.513146698474884,
"learning_rate": 0.0002402097284515686,
"loss": 1.976513671875,
"step": 23160
},
{
"epoch": 2.9656320000000003,
"grad_norm": 0.5778273940086365,
"learning_rate": 0.00024016140904051285,
"loss": 1.9823671340942384,
"step": 23170
},
{
"epoch": 2.9669119999999998,
"grad_norm": 0.5786550641059875,
"learning_rate": 0.00024011307497697902,
"loss": 1.9943340301513672,
"step": 23180
},
{
"epoch": 2.968192,
"grad_norm": 0.5391371846199036,
"learning_rate": 0.0002400647262688223,
"loss": 1.9849088668823243,
"step": 23190
},
{
"epoch": 2.969472,
"grad_norm": 0.5713873505592346,
"learning_rate": 0.00024001636292389982,
"loss": 1.984029197692871,
"step": 23200
},
{
"epoch": 2.970752,
"grad_norm": 0.6697408556938171,
"learning_rate": 0.00023996798495007137,
"loss": 1.9858768463134766,
"step": 23210
},
{
"epoch": 2.972032,
"grad_norm": 0.5721818804740906,
"learning_rate": 0.00023991959235519908,
"loss": 1.9833866119384767,
"step": 23220
},
{
"epoch": 2.973312,
"grad_norm": 0.5981219410896301,
"learning_rate": 0.00023987118514714732,
"loss": 1.9875328063964843,
"step": 23230
},
{
"epoch": 2.974592,
"grad_norm": 0.5732999444007874,
"learning_rate": 0.00023982276333378303,
"loss": 1.9882524490356446,
"step": 23240
},
{
"epoch": 2.975872,
"grad_norm": 0.582938551902771,
"learning_rate": 0.0002397743269229754,
"loss": 1.9852949142456056,
"step": 23250
},
{
"epoch": 2.9771520000000002,
"grad_norm": 0.5819177627563477,
"learning_rate": 0.00023972587592259594,
"loss": 1.978376579284668,
"step": 23260
},
{
"epoch": 2.9784319999999997,
"grad_norm": 0.5837857723236084,
"learning_rate": 0.00023967741034051871,
"loss": 2.002117156982422,
"step": 23270
},
{
"epoch": 2.979712,
"grad_norm": 0.5471209287643433,
"learning_rate": 0.00023962893018461998,
"loss": 1.9862586975097656,
"step": 23280
},
{
"epoch": 2.980992,
"grad_norm": 0.6043733954429626,
"learning_rate": 0.00023958043546277847,
"loss": 2.011483001708984,
"step": 23290
},
{
"epoch": 2.982272,
"grad_norm": 0.549296498298645,
"learning_rate": 0.00023953192618287523,
"loss": 1.9886940002441407,
"step": 23300
},
{
"epoch": 2.983552,
"grad_norm": 0.6022505760192871,
"learning_rate": 0.00023948340235279374,
"loss": 1.9712902069091798,
"step": 23310
},
{
"epoch": 2.984832,
"grad_norm": 0.5907219648361206,
"learning_rate": 0.00023943486398041972,
"loss": 1.9826078414916992,
"step": 23320
},
{
"epoch": 2.986112,
"grad_norm": 0.6209723353385925,
"learning_rate": 0.00023938631107364133,
"loss": 1.982292366027832,
"step": 23330
},
{
"epoch": 2.987392,
"grad_norm": 0.5748019814491272,
"learning_rate": 0.0002393377436403492,
"loss": 1.9795719146728517,
"step": 23340
},
{
"epoch": 2.988672,
"grad_norm": 0.6154806017875671,
"learning_rate": 0.0002392891616884361,
"loss": 1.990694808959961,
"step": 23350
},
{
"epoch": 2.989952,
"grad_norm": 0.6091994643211365,
"learning_rate": 0.00023924056522579723,
"loss": 1.995009422302246,
"step": 23360
},
{
"epoch": 2.991232,
"grad_norm": 0.547141432762146,
"learning_rate": 0.00023919195426033038,
"loss": 1.966740608215332,
"step": 23370
},
{
"epoch": 2.992512,
"grad_norm": 0.5385535359382629,
"learning_rate": 0.00023914332879993531,
"loss": 2.0034152984619142,
"step": 23380
},
{
"epoch": 2.993792,
"grad_norm": 0.596025824546814,
"learning_rate": 0.00023909468885251442,
"loss": 2.002064323425293,
"step": 23390
},
{
"epoch": 2.995072,
"grad_norm": 0.6087286472320557,
"learning_rate": 0.00023904603442597238,
"loss": 1.9820171356201173,
"step": 23400
},
{
"epoch": 2.996352,
"grad_norm": 0.6268278360366821,
"learning_rate": 0.0002389973655282162,
"loss": 1.9938814163208007,
"step": 23410
},
{
"epoch": 2.997632,
"grad_norm": 0.5441094636917114,
"learning_rate": 0.0002389486821671552,
"loss": 1.9856067657470704,
"step": 23420
},
{
"epoch": 2.998912,
"grad_norm": 0.5376409888267517,
"learning_rate": 0.00023889998435070112,
"loss": 1.994309425354004,
"step": 23430
},
{
"epoch": 3.0,
"eval_loss": 1.9939625263214111,
"eval_runtime": 18.2654,
"eval_samples_per_second": 547.483,
"eval_steps_per_second": 68.435,
"step": 23439
},
{
"epoch": 3.000128,
"grad_norm": 0.5639065504074097,
"learning_rate": 0.00023885127208676812,
"loss": 1.9813180923461915,
"step": 23440
},
{
"epoch": 3.001408,
"grad_norm": 0.6670764088630676,
"learning_rate": 0.00023880254538327251,
"loss": 1.9976205825805664,
"step": 23450
},
{
"epoch": 3.002688,
"grad_norm": 0.6705673933029175,
"learning_rate": 0.00023875380424813307,
"loss": 1.9890865325927733,
"step": 23460
},
{
"epoch": 3.003968,
"grad_norm": 0.5052220821380615,
"learning_rate": 0.00023870504868927098,
"loss": 1.990891456604004,
"step": 23470
},
{
"epoch": 3.005248,
"grad_norm": 0.5517461895942688,
"learning_rate": 0.00023865627871460965,
"loss": 1.9883235931396483,
"step": 23480
},
{
"epoch": 3.006528,
"grad_norm": 0.5847997069358826,
"learning_rate": 0.0002386074943320748,
"loss": 1.9913190841674804,
"step": 23490
},
{
"epoch": 3.007808,
"grad_norm": 0.5394625067710876,
"learning_rate": 0.00023855869554959472,
"loss": 2.0069372177124025,
"step": 23500
},
{
"epoch": 3.009088,
"grad_norm": 0.5694800615310669,
"learning_rate": 0.0002385098823750998,
"loss": 1.987819290161133,
"step": 23510
},
{
"epoch": 3.010368,
"grad_norm": 0.6342251896858215,
"learning_rate": 0.00023846105481652283,
"loss": 1.9966646194458009,
"step": 23520
},
{
"epoch": 3.011648,
"grad_norm": 0.5453896522521973,
"learning_rate": 0.00023841221288179904,
"loss": 1.973453903198242,
"step": 23530
},
{
"epoch": 3.012928,
"grad_norm": 0.6461560726165771,
"learning_rate": 0.00023836335657886583,
"loss": 1.9855379104614257,
"step": 23540
},
{
"epoch": 3.014208,
"grad_norm": 0.6052563190460205,
"learning_rate": 0.00023831448591566312,
"loss": 1.9872686386108398,
"step": 23550
},
{
"epoch": 3.015488,
"grad_norm": 0.614098846912384,
"learning_rate": 0.00023826560090013305,
"loss": 1.9906736373901368,
"step": 23560
},
{
"epoch": 3.016768,
"grad_norm": 0.5693179368972778,
"learning_rate": 0.00023821670154022,
"loss": 1.9824440002441406,
"step": 23570
},
{
"epoch": 3.018048,
"grad_norm": 0.5466355085372925,
"learning_rate": 0.00023816778784387094,
"loss": 1.974078369140625,
"step": 23580
},
{
"epoch": 3.019328,
"grad_norm": 0.5999950170516968,
"learning_rate": 0.00023811885981903495,
"loss": 2.0020471572875977,
"step": 23590
},
{
"epoch": 3.020608,
"grad_norm": 0.538802444934845,
"learning_rate": 0.00023806991747366357,
"loss": 1.9932767868041992,
"step": 23600
},
{
"epoch": 3.021888,
"grad_norm": 0.5645008087158203,
"learning_rate": 0.00023802096081571055,
"loss": 1.9707088470458984,
"step": 23610
},
{
"epoch": 3.023168,
"grad_norm": 0.5845574736595154,
"learning_rate": 0.000237971989853132,
"loss": 1.9786319732666016,
"step": 23620
},
{
"epoch": 3.024448,
"grad_norm": 0.6093648076057434,
"learning_rate": 0.0002379230045938865,
"loss": 1.9724607467651367,
"step": 23630
},
{
"epoch": 3.025728,
"grad_norm": 0.5240519046783447,
"learning_rate": 0.00023787400504593465,
"loss": 1.9794605255126954,
"step": 23640
},
{
"epoch": 3.027008,
"grad_norm": 0.5123478770256042,
"learning_rate": 0.00023782499121723975,
"loss": 1.9799571990966798,
"step": 23650
},
{
"epoch": 3.028288,
"grad_norm": 0.6042332649230957,
"learning_rate": 0.00023777596311576712,
"loss": 1.9845085144042969,
"step": 23660
},
{
"epoch": 3.029568,
"grad_norm": 0.582607626914978,
"learning_rate": 0.00023772692074948448,
"loss": 1.9775306701660156,
"step": 23670
},
{
"epoch": 3.030848,
"grad_norm": 0.650549590587616,
"learning_rate": 0.00023767786412636196,
"loss": 1.9762508392333984,
"step": 23680
},
{
"epoch": 3.032128,
"grad_norm": 0.6040688753128052,
"learning_rate": 0.00023762879325437197,
"loss": 1.971165657043457,
"step": 23690
},
{
"epoch": 3.033408,
"grad_norm": 0.5430186986923218,
"learning_rate": 0.0002375797081414891,
"loss": 1.9843374252319337,
"step": 23700
},
{
"epoch": 3.034688,
"grad_norm": 0.5402069687843323,
"learning_rate": 0.00023753060879569038,
"loss": 1.9717254638671875,
"step": 23710
},
{
"epoch": 3.035968,
"grad_norm": 0.602627694606781,
"learning_rate": 0.00023748149522495523,
"loss": 1.986790084838867,
"step": 23720
},
{
"epoch": 3.037248,
"grad_norm": 0.6632218956947327,
"learning_rate": 0.0002374323674372652,
"loss": 1.9858266830444335,
"step": 23730
},
{
"epoch": 3.038528,
"grad_norm": 0.538617730140686,
"learning_rate": 0.00023738322544060428,
"loss": 1.9925521850585937,
"step": 23740
},
{
"epoch": 3.039808,
"grad_norm": 0.5587111115455627,
"learning_rate": 0.00023733406924295869,
"loss": 1.9831218719482422,
"step": 23750
},
{
"epoch": 3.041088,
"grad_norm": 0.6544376611709595,
"learning_rate": 0.00023728489885231703,
"loss": 2.004766654968262,
"step": 23760
},
{
"epoch": 3.042368,
"grad_norm": 0.5886899828910828,
"learning_rate": 0.00023723571427667007,
"loss": 1.9740234375,
"step": 23770
},
{
"epoch": 3.043648,
"grad_norm": 0.6062396168708801,
"learning_rate": 0.00023718651552401116,
"loss": 2.0035427093505858,
"step": 23780
},
{
"epoch": 3.044928,
"grad_norm": 0.5452648997306824,
"learning_rate": 0.0002371373026023356,
"loss": 1.9931446075439454,
"step": 23790
},
{
"epoch": 3.046208,
"grad_norm": 0.5492119193077087,
"learning_rate": 0.0002370880755196413,
"loss": 1.9826221466064453,
"step": 23800
},
{
"epoch": 3.047488,
"grad_norm": 0.5913605690002441,
"learning_rate": 0.0002370388342839283,
"loss": 1.9869071960449218,
"step": 23810
},
{
"epoch": 3.048768,
"grad_norm": 0.6001777052879333,
"learning_rate": 0.00023698957890319895,
"loss": 1.999561882019043,
"step": 23820
},
{
"epoch": 3.050048,
"grad_norm": 0.5186832547187805,
"learning_rate": 0.00023694030938545797,
"loss": 1.9632522583007812,
"step": 23830
},
{
"epoch": 3.051328,
"grad_norm": 0.6168757081031799,
"learning_rate": 0.00023689102573871238,
"loss": 2.013888359069824,
"step": 23840
},
{
"epoch": 3.052608,
"grad_norm": 0.6054990291595459,
"learning_rate": 0.00023684172797097135,
"loss": 1.9938718795776367,
"step": 23850
},
{
"epoch": 3.053888,
"grad_norm": 0.6404632925987244,
"learning_rate": 0.00023679241609024656,
"loss": 1.983924102783203,
"step": 23860
},
{
"epoch": 3.055168,
"grad_norm": 0.5868401527404785,
"learning_rate": 0.00023674309010455176,
"loss": 1.9687318801879883,
"step": 23870
},
{
"epoch": 3.056448,
"grad_norm": 0.6216502785682678,
"learning_rate": 0.00023669375002190315,
"loss": 1.9979162216186523,
"step": 23880
},
{
"epoch": 3.057728,
"grad_norm": 0.5381942987442017,
"learning_rate": 0.00023664439585031926,
"loss": 1.9786861419677735,
"step": 23890
},
{
"epoch": 3.059008,
"grad_norm": 0.6260485649108887,
"learning_rate": 0.00023659502759782074,
"loss": 1.9935455322265625,
"step": 23900
},
{
"epoch": 3.060288,
"grad_norm": 0.5902127623558044,
"learning_rate": 0.00023654564527243063,
"loss": 1.986068344116211,
"step": 23910
},
{
"epoch": 3.061568,
"grad_norm": 0.6114174723625183,
"learning_rate": 0.00023649624888217422,
"loss": 1.9879646301269531,
"step": 23920
},
{
"epoch": 3.062848,
"grad_norm": 0.5158154368400574,
"learning_rate": 0.00023644683843507912,
"loss": 1.9839748382568358,
"step": 23930
},
{
"epoch": 3.064128,
"grad_norm": 0.6757631301879883,
"learning_rate": 0.00023639741393917532,
"loss": 1.9806655883789062,
"step": 23940
},
{
"epoch": 3.065408,
"grad_norm": 0.5897718071937561,
"learning_rate": 0.00023634797540249478,
"loss": 1.994769287109375,
"step": 23950
},
{
"epoch": 3.066688,
"grad_norm": 0.6237477660179138,
"learning_rate": 0.00023629852283307213,
"loss": 1.9915149688720704,
"step": 23960
},
{
"epoch": 3.067968,
"grad_norm": 0.5773525834083557,
"learning_rate": 0.000236249056238944,
"loss": 1.9801607131958008,
"step": 23970
},
{
"epoch": 3.069248,
"grad_norm": 0.6370024681091309,
"learning_rate": 0.00023619957562814943,
"loss": 1.9894296646118164,
"step": 23980
},
{
"epoch": 3.070528,
"grad_norm": 0.6042509078979492,
"learning_rate": 0.00023615008100872967,
"loss": 1.9984472274780274,
"step": 23990
},
{
"epoch": 3.071808,
"grad_norm": 0.6538687944412231,
"learning_rate": 0.0002361005723887283,
"loss": 2.0000247955322266,
"step": 24000
},
{
"epoch": 3.073088,
"grad_norm": 0.6268402934074402,
"learning_rate": 0.0002360510497761912,
"loss": 1.9860471725463866,
"step": 24010
},
{
"epoch": 3.074368,
"grad_norm": 0.5438446402549744,
"learning_rate": 0.00023600151317916646,
"loss": 2.000538444519043,
"step": 24020
},
{
"epoch": 3.075648,
"grad_norm": 0.6090270280838013,
"learning_rate": 0.00023595196260570438,
"loss": 2.001275634765625,
"step": 24030
},
{
"epoch": 3.076928,
"grad_norm": 0.5382629036903381,
"learning_rate": 0.0002359023980638578,
"loss": 1.9940172195434571,
"step": 24040
},
{
"epoch": 3.078208,
"grad_norm": 0.7335079312324524,
"learning_rate": 0.00023585281956168145,
"loss": 2.0011875152587892,
"step": 24050
},
{
"epoch": 3.079488,
"grad_norm": 0.5846731066703796,
"learning_rate": 0.00023580322710723258,
"loss": 2.0010887145996095,
"step": 24060
},
{
"epoch": 3.080768,
"grad_norm": 0.6610224843025208,
"learning_rate": 0.00023575362070857073,
"loss": 1.9665187835693358,
"step": 24070
},
{
"epoch": 3.082048,
"grad_norm": 0.5683338642120361,
"learning_rate": 0.00023570400037375752,
"loss": 1.9613128662109376,
"step": 24080
},
{
"epoch": 3.083328,
"grad_norm": 0.5672542452812195,
"learning_rate": 0.000235654366110857,
"loss": 2.000151252746582,
"step": 24090
},
{
"epoch": 3.084608,
"grad_norm": 0.5498451590538025,
"learning_rate": 0.00023560471792793546,
"loss": 1.9828907012939454,
"step": 24100
},
{
"epoch": 3.085888,
"grad_norm": 0.615347683429718,
"learning_rate": 0.00023555505583306132,
"loss": 1.9809429168701171,
"step": 24110
},
{
"epoch": 3.087168,
"grad_norm": 0.5429461598396301,
"learning_rate": 0.00023550537983430545,
"loss": 1.9731021881103517,
"step": 24120
},
{
"epoch": 3.088448,
"grad_norm": 0.5744686722755432,
"learning_rate": 0.0002354556899397408,
"loss": 2.000034141540527,
"step": 24130
},
{
"epoch": 3.089728,
"grad_norm": 0.6077907085418701,
"learning_rate": 0.00023540598615744276,
"loss": 1.9937231063842773,
"step": 24140
},
{
"epoch": 3.091008,
"grad_norm": 0.6105172634124756,
"learning_rate": 0.00023535626849548885,
"loss": 1.9851383209228515,
"step": 24150
},
{
"epoch": 3.092288,
"grad_norm": 0.5446839332580566,
"learning_rate": 0.00023530653696195885,
"loss": 2.0077171325683594,
"step": 24160
},
{
"epoch": 3.093568,
"grad_norm": 0.5680665373802185,
"learning_rate": 0.00023525679156493487,
"loss": 1.9853170394897461,
"step": 24170
},
{
"epoch": 3.094848,
"grad_norm": 0.5555473566055298,
"learning_rate": 0.00023520703231250114,
"loss": 2.0012443542480467,
"step": 24180
},
{
"epoch": 3.096128,
"grad_norm": 0.5348846316337585,
"learning_rate": 0.0002351572592127443,
"loss": 1.988169288635254,
"step": 24190
},
{
"epoch": 3.097408,
"grad_norm": 0.5657079815864563,
"learning_rate": 0.00023510747227375322,
"loss": 1.9833412170410156,
"step": 24200
},
{
"epoch": 3.098688,
"grad_norm": 0.6253102421760559,
"learning_rate": 0.00023505767150361882,
"loss": 1.9989374160766602,
"step": 24210
},
{
"epoch": 3.099968,
"grad_norm": 0.634135901927948,
"learning_rate": 0.00023500785691043453,
"loss": 1.9974393844604492,
"step": 24220
},
{
"epoch": 3.101248,
"grad_norm": 0.5518096089363098,
"learning_rate": 0.00023495802850229586,
"loss": 1.990166473388672,
"step": 24230
},
{
"epoch": 3.102528,
"grad_norm": 0.650854229927063,
"learning_rate": 0.00023490818628730067,
"loss": 1.9911649703979493,
"step": 24240
},
{
"epoch": 3.103808,
"grad_norm": 0.5608437657356262,
"learning_rate": 0.00023485833027354891,
"loss": 1.9896230697631836,
"step": 24250
},
{
"epoch": 3.105088,
"grad_norm": 0.5557622313499451,
"learning_rate": 0.000234808460469143,
"loss": 1.9885791778564452,
"step": 24260
},
{
"epoch": 3.106368,
"grad_norm": 0.5130448937416077,
"learning_rate": 0.00023475857688218737,
"loss": 1.9946943283081056,
"step": 24270
},
{
"epoch": 3.107648,
"grad_norm": 0.6254590749740601,
"learning_rate": 0.00023470867952078887,
"loss": 1.982680892944336,
"step": 24280
},
{
"epoch": 3.108928,
"grad_norm": 0.6610821485519409,
"learning_rate": 0.0002346587683930565,
"loss": 1.9815378189086914,
"step": 24290
},
{
"epoch": 3.110208,
"grad_norm": 0.5842161178588867,
"learning_rate": 0.00023460884350710143,
"loss": 1.9911502838134765,
"step": 24300
},
{
"epoch": 3.111488,
"grad_norm": 0.5703555345535278,
"learning_rate": 0.0002345589048710372,
"loss": 1.9919004440307617,
"step": 24310
},
{
"epoch": 3.112768,
"grad_norm": 0.5435205101966858,
"learning_rate": 0.00023450895249297956,
"loss": 1.9708248138427735,
"step": 24320
},
{
"epoch": 3.114048,
"grad_norm": 0.5477770566940308,
"learning_rate": 0.00023445898638104647,
"loss": 1.990170669555664,
"step": 24330
},
{
"epoch": 3.115328,
"grad_norm": 0.5144780278205872,
"learning_rate": 0.00023440900654335802,
"loss": 1.980717086791992,
"step": 24340
},
{
"epoch": 3.116608,
"grad_norm": 0.5944603085517883,
"learning_rate": 0.0002343590129880367,
"loss": 1.9938461303710937,
"step": 24350
},
{
"epoch": 3.117888,
"grad_norm": 0.5773472785949707,
"learning_rate": 0.00023430900572320718,
"loss": 2.0137517929077147,
"step": 24360
},
{
"epoch": 3.119168,
"grad_norm": 0.5520356893539429,
"learning_rate": 0.0002342589847569963,
"loss": 1.9941003799438477,
"step": 24370
},
{
"epoch": 3.120448,
"grad_norm": 0.5808164477348328,
"learning_rate": 0.0002342089500975331,
"loss": 1.9862529754638671,
"step": 24380
},
{
"epoch": 3.121728,
"grad_norm": 0.6027176380157471,
"learning_rate": 0.000234158901752949,
"loss": 1.987776756286621,
"step": 24390
},
{
"epoch": 3.123008,
"grad_norm": 0.6502209901809692,
"learning_rate": 0.0002341088397313775,
"loss": 1.991359329223633,
"step": 24400
},
{
"epoch": 3.124288,
"grad_norm": 0.5783761739730835,
"learning_rate": 0.0002340587640409544,
"loss": 1.9986988067626954,
"step": 24410
},
{
"epoch": 3.125568,
"grad_norm": 0.5866621136665344,
"learning_rate": 0.00023400867468981765,
"loss": 2.004291534423828,
"step": 24420
},
{
"epoch": 3.126848,
"grad_norm": 0.627935528755188,
"learning_rate": 0.0002339585716861075,
"loss": 1.994461441040039,
"step": 24430
},
{
"epoch": 3.128128,
"grad_norm": 0.6197845339775085,
"learning_rate": 0.00023390845503796636,
"loss": 1.977349090576172,
"step": 24440
},
{
"epoch": 3.129408,
"grad_norm": 0.6650096774101257,
"learning_rate": 0.00023385832475353893,
"loss": 1.979800033569336,
"step": 24450
},
{
"epoch": 3.130688,
"grad_norm": 0.5335533618927002,
"learning_rate": 0.00023380818084097203,
"loss": 1.9957963943481445,
"step": 24460
},
{
"epoch": 3.131968,
"grad_norm": 0.5574480295181274,
"learning_rate": 0.00023375802330841472,
"loss": 1.9823905944824218,
"step": 24470
},
{
"epoch": 3.133248,
"grad_norm": 0.561566174030304,
"learning_rate": 0.00023370785216401835,
"loss": 1.9903274536132813,
"step": 24480
},
{
"epoch": 3.134528,
"grad_norm": 0.5479939579963684,
"learning_rate": 0.0002336576674159364,
"loss": 1.9766128540039063,
"step": 24490
},
{
"epoch": 3.135808,
"grad_norm": 0.5059587955474854,
"learning_rate": 0.00023360746907232454,
"loss": 1.9858789443969727,
"step": 24500
},
{
"epoch": 3.137088,
"grad_norm": 0.5911202430725098,
"learning_rate": 0.00023355725714134076,
"loss": 1.9907167434692383,
"step": 24510
},
{
"epoch": 3.138368,
"grad_norm": 0.5590784549713135,
"learning_rate": 0.0002335070316311452,
"loss": 2.011886405944824,
"step": 24520
},
{
"epoch": 3.139648,
"grad_norm": 0.608490526676178,
"learning_rate": 0.00023345679254990016,
"loss": 1.9871561050415039,
"step": 24530
},
{
"epoch": 3.140928,
"grad_norm": 0.5686827301979065,
"learning_rate": 0.0002334065399057702,
"loss": 1.9754255294799805,
"step": 24540
},
{
"epoch": 3.142208,
"grad_norm": 0.552444338798523,
"learning_rate": 0.00023335627370692208,
"loss": 2.012767028808594,
"step": 24550
},
{
"epoch": 3.143488,
"grad_norm": 0.6630028486251831,
"learning_rate": 0.00023330599396152474,
"loss": 1.9758792877197267,
"step": 24560
},
{
"epoch": 3.144768,
"grad_norm": 0.6015341281890869,
"learning_rate": 0.00023325570067774933,
"loss": 1.973785972595215,
"step": 24570
},
{
"epoch": 3.146048,
"grad_norm": 0.5953616499900818,
"learning_rate": 0.00023320539386376924,
"loss": 1.9960126876831055,
"step": 24580
},
{
"epoch": 3.147328,
"grad_norm": 0.49537065625190735,
"learning_rate": 0.00023315507352776002,
"loss": 1.9968629837036134,
"step": 24590
},
{
"epoch": 3.148608,
"grad_norm": 0.6414424777030945,
"learning_rate": 0.00023310473967789937,
"loss": 1.9909296035766602,
"step": 24600
},
{
"epoch": 3.149888,
"grad_norm": 0.5670834183692932,
"learning_rate": 0.0002330543923223673,
"loss": 1.977849578857422,
"step": 24610
},
{
"epoch": 3.151168,
"grad_norm": 0.5764061212539673,
"learning_rate": 0.00023300403146934588,
"loss": 1.9865072250366211,
"step": 24620
},
{
"epoch": 3.152448,
"grad_norm": 0.5635042190551758,
"learning_rate": 0.0002329536571270195,
"loss": 1.989523696899414,
"step": 24630
},
{
"epoch": 3.153728,
"grad_norm": 0.5438238382339478,
"learning_rate": 0.00023290326930357476,
"loss": 1.9882726669311523,
"step": 24640
},
{
"epoch": 3.155008,
"grad_norm": 0.5321423411369324,
"learning_rate": 0.00023285286800720021,
"loss": 1.986033821105957,
"step": 24650
},
{
"epoch": 3.156288,
"grad_norm": 0.6436901688575745,
"learning_rate": 0.00023280245324608687,
"loss": 1.9812103271484376,
"step": 24660
},
{
"epoch": 3.157568,
"grad_norm": 0.6874346137046814,
"learning_rate": 0.0002327520250284278,
"loss": 1.991452980041504,
"step": 24670
},
{
"epoch": 3.158848,
"grad_norm": 0.6873441934585571,
"learning_rate": 0.00023270158336241836,
"loss": 1.9889459609985352,
"step": 24680
},
{
"epoch": 3.160128,
"grad_norm": 0.5413915514945984,
"learning_rate": 0.00023265112825625588,
"loss": 1.9882686614990235,
"step": 24690
},
{
"epoch": 3.1614079999999998,
"grad_norm": 0.6368159651756287,
"learning_rate": 0.00023260065971814013,
"loss": 2.0011133193969726,
"step": 24700
},
{
"epoch": 3.162688,
"grad_norm": 0.5715042352676392,
"learning_rate": 0.0002325501777562729,
"loss": 1.969057273864746,
"step": 24710
},
{
"epoch": 3.163968,
"grad_norm": 0.5132859945297241,
"learning_rate": 0.00023249968237885822,
"loss": 1.9842401504516602,
"step": 24720
},
{
"epoch": 3.165248,
"grad_norm": 0.48780661821365356,
"learning_rate": 0.0002324491735941023,
"loss": 1.9755685806274415,
"step": 24730
},
{
"epoch": 3.166528,
"grad_norm": 0.6054424047470093,
"learning_rate": 0.00023239865141021348,
"loss": 1.955612564086914,
"step": 24740
},
{
"epoch": 3.167808,
"grad_norm": 0.6506003141403198,
"learning_rate": 0.00023234811583540236,
"loss": 1.997670555114746,
"step": 24750
},
{
"epoch": 3.169088,
"grad_norm": 0.5567013621330261,
"learning_rate": 0.00023229756687788162,
"loss": 1.9816686630249023,
"step": 24760
},
{
"epoch": 3.170368,
"grad_norm": 0.6238530874252319,
"learning_rate": 0.00023224700454586622,
"loss": 1.973410415649414,
"step": 24770
},
{
"epoch": 3.171648,
"grad_norm": 0.5906739234924316,
"learning_rate": 0.0002321964288475732,
"loss": 1.9880643844604493,
"step": 24780
},
{
"epoch": 3.172928,
"grad_norm": 0.563016414642334,
"learning_rate": 0.0002321458397912218,
"loss": 1.967965316772461,
"step": 24790
},
{
"epoch": 3.174208,
"grad_norm": 0.6094998717308044,
"learning_rate": 0.00023209523738503348,
"loss": 2.00931339263916,
"step": 24800
},
{
"epoch": 3.175488,
"grad_norm": 0.5282471179962158,
"learning_rate": 0.00023204462163723186,
"loss": 1.9967885971069337,
"step": 24810
},
{
"epoch": 3.176768,
"grad_norm": 0.540429949760437,
"learning_rate": 0.00023199399255604262,
"loss": 1.9779693603515625,
"step": 24820
},
{
"epoch": 3.178048,
"grad_norm": 0.527767539024353,
"learning_rate": 0.00023194335014969374,
"loss": 1.9777732849121095,
"step": 24830
},
{
"epoch": 3.179328,
"grad_norm": 0.5744035243988037,
"learning_rate": 0.00023189269442641528,
"loss": 1.9790773391723633,
"step": 24840
},
{
"epoch": 3.180608,
"grad_norm": 0.6080548167228699,
"learning_rate": 0.00023184202539443952,
"loss": 2.002645492553711,
"step": 24850
},
{
"epoch": 3.181888,
"grad_norm": 0.5465280413627625,
"learning_rate": 0.00023179134306200092,
"loss": 1.9831550598144532,
"step": 24860
},
{
"epoch": 3.183168,
"grad_norm": 0.6074657440185547,
"learning_rate": 0.00023174064743733598,
"loss": 1.9838821411132812,
"step": 24870
},
{
"epoch": 3.184448,
"grad_norm": 0.6524350047111511,
"learning_rate": 0.00023168993852868344,
"loss": 1.9819204330444335,
"step": 24880
},
{
"epoch": 3.185728,
"grad_norm": 0.5440987348556519,
"learning_rate": 0.0002316392163442843,
"loss": 1.993447494506836,
"step": 24890
},
{
"epoch": 3.187008,
"grad_norm": 0.5262812376022339,
"learning_rate": 0.00023158848089238155,
"loss": 1.9817289352416991,
"step": 24900
},
{
"epoch": 3.188288,
"grad_norm": 0.5689319372177124,
"learning_rate": 0.00023153773218122036,
"loss": 2.008134651184082,
"step": 24910
},
{
"epoch": 3.189568,
"grad_norm": 0.541843593120575,
"learning_rate": 0.0002314869702190482,
"loss": 2.0019962310791017,
"step": 24920
},
{
"epoch": 3.190848,
"grad_norm": 0.638310968875885,
"learning_rate": 0.0002314361950141145,
"loss": 1.9735862731933593,
"step": 24930
},
{
"epoch": 3.192128,
"grad_norm": 0.5001397728919983,
"learning_rate": 0.00023138540657467098,
"loss": 1.9955354690551759,
"step": 24940
},
{
"epoch": 3.193408,
"grad_norm": 0.6100158095359802,
"learning_rate": 0.0002313346049089715,
"loss": 2.00225830078125,
"step": 24950
},
{
"epoch": 3.194688,
"grad_norm": 0.5394402742385864,
"learning_rate": 0.00023128379002527191,
"loss": 1.9944787979125977,
"step": 24960
},
{
"epoch": 3.195968,
"grad_norm": 0.677653968334198,
"learning_rate": 0.0002312329619318305,
"loss": 1.9953351974487306,
"step": 24970
},
{
"epoch": 3.197248,
"grad_norm": 0.5648075938224792,
"learning_rate": 0.00023118212063690742,
"loss": 1.9760156631469727,
"step": 24980
},
{
"epoch": 3.198528,
"grad_norm": 0.6293234825134277,
"learning_rate": 0.00023113126614876514,
"loss": 1.9821134567260743,
"step": 24990
},
{
"epoch": 3.199808,
"grad_norm": 0.5262898206710815,
"learning_rate": 0.00023108039847566817,
"loss": 1.9973690032958984,
"step": 25000
},
{
"epoch": 3.201088,
"grad_norm": 0.5845648050308228,
"learning_rate": 0.00023102951762588324,
"loss": 1.9782295227050781,
"step": 25010
},
{
"epoch": 3.202368,
"grad_norm": 0.5487980842590332,
"learning_rate": 0.00023097862360767923,
"loss": 1.9920047760009765,
"step": 25020
},
{
"epoch": 3.203648,
"grad_norm": 0.5452269911766052,
"learning_rate": 0.00023092771642932704,
"loss": 1.989668083190918,
"step": 25030
},
{
"epoch": 3.2049279999999998,
"grad_norm": 0.5880693793296814,
"learning_rate": 0.0002308767960990999,
"loss": 2.0039955139160157,
"step": 25040
},
{
"epoch": 3.206208,
"grad_norm": 0.643682062625885,
"learning_rate": 0.00023082586262527296,
"loss": 1.9718509674072267,
"step": 25050
},
{
"epoch": 3.207488,
"grad_norm": 0.5424628257751465,
"learning_rate": 0.00023077491601612365,
"loss": 2.000941848754883,
"step": 25060
},
{
"epoch": 3.208768,
"grad_norm": 0.510338306427002,
"learning_rate": 0.00023072395627993155,
"loss": 1.986880111694336,
"step": 25070
},
{
"epoch": 3.210048,
"grad_norm": 0.5362533926963806,
"learning_rate": 0.0002306729834249783,
"loss": 1.9790258407592773,
"step": 25080
},
{
"epoch": 3.211328,
"grad_norm": 0.6376779675483704,
"learning_rate": 0.0002306219974595476,
"loss": 2.000022125244141,
"step": 25090
},
{
"epoch": 3.212608,
"grad_norm": 0.6507259011268616,
"learning_rate": 0.0002305709983919255,
"loss": 1.973978042602539,
"step": 25100
},
{
"epoch": 3.213888,
"grad_norm": 0.5760257244110107,
"learning_rate": 0.00023051998623039997,
"loss": 1.9775646209716797,
"step": 25110
},
{
"epoch": 3.215168,
"grad_norm": 0.581234872341156,
"learning_rate": 0.00023046896098326134,
"loss": 1.9825855255126954,
"step": 25120
},
{
"epoch": 3.216448,
"grad_norm": 0.6260392665863037,
"learning_rate": 0.00023041792265880172,
"loss": 1.9761247634887695,
"step": 25130
},
{
"epoch": 3.217728,
"grad_norm": 0.6558592915534973,
"learning_rate": 0.00023036687126531567,
"loss": 1.9727117538452148,
"step": 25140
},
{
"epoch": 3.219008,
"grad_norm": 0.5699613690376282,
"learning_rate": 0.00023031580681109973,
"loss": 1.982707405090332,
"step": 25150
},
{
"epoch": 3.220288,
"grad_norm": 0.5609349608421326,
"learning_rate": 0.0002302647293044525,
"loss": 1.9840452194213867,
"step": 25160
},
{
"epoch": 3.221568,
"grad_norm": 0.6656886339187622,
"learning_rate": 0.0002302136387536749,
"loss": 1.9855270385742188,
"step": 25170
},
{
"epoch": 3.222848,
"grad_norm": 0.6227736473083496,
"learning_rate": 0.00023016253516706985,
"loss": 1.9742204666137695,
"step": 25180
},
{
"epoch": 3.224128,
"grad_norm": 0.6104076504707336,
"learning_rate": 0.0002301114185529423,
"loss": 1.9900693893432617,
"step": 25190
},
{
"epoch": 3.225408,
"grad_norm": 0.5289227962493896,
"learning_rate": 0.00023006028891959946,
"loss": 1.9850357055664063,
"step": 25200
},
{
"epoch": 3.2266880000000002,
"grad_norm": 0.6033598184585571,
"learning_rate": 0.00023000914627535054,
"loss": 1.9980274200439454,
"step": 25210
},
{
"epoch": 3.227968,
"grad_norm": 0.6151910424232483,
"learning_rate": 0.00022995799062850705,
"loss": 1.9819875717163087,
"step": 25220
},
{
"epoch": 3.229248,
"grad_norm": 0.6556711196899414,
"learning_rate": 0.0002299068219873824,
"loss": 1.983561134338379,
"step": 25230
},
{
"epoch": 3.230528,
"grad_norm": 0.5222577452659607,
"learning_rate": 0.00022985564036029225,
"loss": 1.982244110107422,
"step": 25240
},
{
"epoch": 3.231808,
"grad_norm": 0.5988575220108032,
"learning_rate": 0.00022980444575555428,
"loss": 1.9832239151000977,
"step": 25250
},
{
"epoch": 3.233088,
"grad_norm": 0.5118327736854553,
"learning_rate": 0.00022975323818148833,
"loss": 1.9960365295410156,
"step": 25260
},
{
"epoch": 3.234368,
"grad_norm": 0.6430198550224304,
"learning_rate": 0.00022970201764641634,
"loss": 1.9758806228637695,
"step": 25270
},
{
"epoch": 3.235648,
"grad_norm": 0.6431428790092468,
"learning_rate": 0.0002296507841586624,
"loss": 1.9754806518554688,
"step": 25280
},
{
"epoch": 3.236928,
"grad_norm": 0.5644878149032593,
"learning_rate": 0.0002295995377265526,
"loss": 1.9946617126464843,
"step": 25290
},
{
"epoch": 3.238208,
"grad_norm": 0.5899432301521301,
"learning_rate": 0.00022954827835841533,
"loss": 1.987223243713379,
"step": 25300
},
{
"epoch": 3.239488,
"grad_norm": 0.5841452479362488,
"learning_rate": 0.00022949700606258072,
"loss": 1.9808109283447266,
"step": 25310
},
{
"epoch": 3.240768,
"grad_norm": 0.5357463955879211,
"learning_rate": 0.00022944572084738138,
"loss": 1.9840631484985352,
"step": 25320
},
{
"epoch": 3.242048,
"grad_norm": 0.5990767478942871,
"learning_rate": 0.00022939442272115187,
"loss": 1.99046630859375,
"step": 25330
},
{
"epoch": 3.243328,
"grad_norm": 0.5736744999885559,
"learning_rate": 0.00022934311169222878,
"loss": 1.9944417953491211,
"step": 25340
},
{
"epoch": 3.244608,
"grad_norm": 0.5440366864204407,
"learning_rate": 0.0002292917877689509,
"loss": 1.9653417587280273,
"step": 25350
},
{
"epoch": 3.245888,
"grad_norm": 0.5249848961830139,
"learning_rate": 0.00022924045095965912,
"loss": 1.9987998962402345,
"step": 25360
},
{
"epoch": 3.247168,
"grad_norm": 0.5255021452903748,
"learning_rate": 0.00022918910127269634,
"loss": 1.9666845321655273,
"step": 25370
},
{
"epoch": 3.248448,
"grad_norm": 0.5041667222976685,
"learning_rate": 0.00022913773871640762,
"loss": 1.9795921325683594,
"step": 25380
},
{
"epoch": 3.249728,
"grad_norm": 0.7786513566970825,
"learning_rate": 0.00022908636329913997,
"loss": 2.0017948150634766,
"step": 25390
},
{
"epoch": 3.251008,
"grad_norm": 0.6393799185752869,
"learning_rate": 0.00022903497502924279,
"loss": 1.9859493255615235,
"step": 25400
},
{
"epoch": 3.252288,
"grad_norm": 0.6293373107910156,
"learning_rate": 0.00022898357391506723,
"loss": 1.990733528137207,
"step": 25410
},
{
"epoch": 3.253568,
"grad_norm": 0.6206292510032654,
"learning_rate": 0.00022893215996496678,
"loss": 1.9848413467407227,
"step": 25420
},
{
"epoch": 3.254848,
"grad_norm": 0.5599654316902161,
"learning_rate": 0.0002288807331872969,
"loss": 1.9900089263916017,
"step": 25430
},
{
"epoch": 3.256128,
"grad_norm": 0.583229660987854,
"learning_rate": 0.00022882929359041514,
"loss": 1.984672737121582,
"step": 25440
},
{
"epoch": 3.257408,
"grad_norm": 0.564947247505188,
"learning_rate": 0.00022877784118268112,
"loss": 1.9710359573364258,
"step": 25450
},
{
"epoch": 3.258688,
"grad_norm": 0.5558592081069946,
"learning_rate": 0.00022872637597245657,
"loss": 1.9933019638061524,
"step": 25460
},
{
"epoch": 3.2599679999999998,
"grad_norm": 0.6236609220504761,
"learning_rate": 0.00022867489796810542,
"loss": 1.9877862930297852,
"step": 25470
},
{
"epoch": 3.261248,
"grad_norm": 0.577813446521759,
"learning_rate": 0.0002286234071779934,
"loss": 1.9868661880493164,
"step": 25480
},
{
"epoch": 3.262528,
"grad_norm": 0.6737725138664246,
"learning_rate": 0.00022857190361048855,
"loss": 1.9833360671997071,
"step": 25490
},
{
"epoch": 3.263808,
"grad_norm": 0.59235018491745,
"learning_rate": 0.00022852038727396088,
"loss": 1.9865446090698242,
"step": 25500
},
{
"epoch": 3.265088,
"grad_norm": 0.5808746218681335,
"learning_rate": 0.00022846885817678263,
"loss": 1.9849153518676759,
"step": 25510
},
{
"epoch": 3.266368,
"grad_norm": 0.5626165866851807,
"learning_rate": 0.0002284173163273278,
"loss": 1.9847883224487304,
"step": 25520
},
{
"epoch": 3.267648,
"grad_norm": 0.5162211060523987,
"learning_rate": 0.00022836576173397281,
"loss": 1.9868818283081056,
"step": 25530
},
{
"epoch": 3.268928,
"grad_norm": 0.6702050566673279,
"learning_rate": 0.00022831419440509594,
"loss": 1.9832344055175781,
"step": 25540
},
{
"epoch": 3.2702080000000002,
"grad_norm": 0.569055438041687,
"learning_rate": 0.00022826261434907751,
"loss": 1.9740659713745117,
"step": 25550
},
{
"epoch": 3.271488,
"grad_norm": 0.7077053189277649,
"learning_rate": 0.00022821102157430015,
"loss": 1.9866657257080078,
"step": 25560
},
{
"epoch": 3.272768,
"grad_norm": 0.5884865522384644,
"learning_rate": 0.0002281594160891483,
"loss": 1.9961114883422852,
"step": 25570
},
{
"epoch": 3.274048,
"grad_norm": 0.5947074890136719,
"learning_rate": 0.0002281077979020086,
"loss": 1.989340591430664,
"step": 25580
},
{
"epoch": 3.275328,
"grad_norm": 0.5695871710777283,
"learning_rate": 0.0002280561670212697,
"loss": 1.9763702392578124,
"step": 25590
},
{
"epoch": 3.276608,
"grad_norm": 0.5365485548973083,
"learning_rate": 0.00022800452345532238,
"loss": 1.9948841094970704,
"step": 25600
},
{
"epoch": 3.277888,
"grad_norm": 0.549697995185852,
"learning_rate": 0.00022795286721255937,
"loss": 1.9908466339111328,
"step": 25610
},
{
"epoch": 3.279168,
"grad_norm": 0.6736475825309753,
"learning_rate": 0.0002279011983013756,
"loss": 1.9834627151489257,
"step": 25620
},
{
"epoch": 3.280448,
"grad_norm": 0.5713078379631042,
"learning_rate": 0.00022784951673016784,
"loss": 1.9666187286376953,
"step": 25630
},
{
"epoch": 3.281728,
"grad_norm": 0.5527968406677246,
"learning_rate": 0.00022779782250733523,
"loss": 1.9971948623657227,
"step": 25640
},
{
"epoch": 3.283008,
"grad_norm": 0.5583022236824036,
"learning_rate": 0.00022774611564127875,
"loss": 1.9863199234008788,
"step": 25650
},
{
"epoch": 3.284288,
"grad_norm": 0.5897033214569092,
"learning_rate": 0.00022769439614040145,
"loss": 1.96350154876709,
"step": 25660
},
{
"epoch": 3.285568,
"grad_norm": 0.6889257431030273,
"learning_rate": 0.00022764266401310845,
"loss": 1.98193359375,
"step": 25670
},
{
"epoch": 3.286848,
"grad_norm": 0.5120753645896912,
"learning_rate": 0.00022759091926780696,
"loss": 1.9850614547729493,
"step": 25680
},
{
"epoch": 3.288128,
"grad_norm": 0.5214012265205383,
"learning_rate": 0.00022753916191290625,
"loss": 1.995151901245117,
"step": 25690
},
{
"epoch": 3.289408,
"grad_norm": 0.5900340676307678,
"learning_rate": 0.00022748739195681755,
"loss": 1.989593505859375,
"step": 25700
},
{
"epoch": 3.290688,
"grad_norm": 0.6398248076438904,
"learning_rate": 0.00022743560940795426,
"loss": 1.9853450775146484,
"step": 25710
},
{
"epoch": 3.291968,
"grad_norm": 0.5377202033996582,
"learning_rate": 0.0002273838142747317,
"loss": 1.9927574157714845,
"step": 25720
},
{
"epoch": 3.293248,
"grad_norm": 0.6756455302238464,
"learning_rate": 0.0002273320065655673,
"loss": 1.9680391311645509,
"step": 25730
},
{
"epoch": 3.294528,
"grad_norm": 0.5579285025596619,
"learning_rate": 0.0002272801862888806,
"loss": 1.987359619140625,
"step": 25740
},
{
"epoch": 3.295808,
"grad_norm": 0.6385338306427002,
"learning_rate": 0.00022722835345309305,
"loss": 2.003000831604004,
"step": 25750
},
{
"epoch": 3.297088,
"grad_norm": 0.5434996485710144,
"learning_rate": 0.0002271765080666282,
"loss": 1.986551856994629,
"step": 25760
},
{
"epoch": 3.298368,
"grad_norm": 0.5533255338668823,
"learning_rate": 0.00022712465013791168,
"loss": 1.9938627243041993,
"step": 25770
},
{
"epoch": 3.299648,
"grad_norm": 0.5487186908721924,
"learning_rate": 0.0002270727796753711,
"loss": 1.97025146484375,
"step": 25780
},
{
"epoch": 3.300928,
"grad_norm": 0.5994226336479187,
"learning_rate": 0.0002270208966874361,
"loss": 2.0124526977539063,
"step": 25790
},
{
"epoch": 3.302208,
"grad_norm": 0.5293451547622681,
"learning_rate": 0.00022696900118253848,
"loss": 1.9865531921386719,
"step": 25800
},
{
"epoch": 3.3034879999999998,
"grad_norm": 0.5846413969993591,
"learning_rate": 0.00022691709316911192,
"loss": 1.9832498550415039,
"step": 25810
},
{
"epoch": 3.304768,
"grad_norm": 0.5115662217140198,
"learning_rate": 0.00022686517265559214,
"loss": 1.9946771621704102,
"step": 25820
},
{
"epoch": 3.306048,
"grad_norm": 0.5601929426193237,
"learning_rate": 0.00022681323965041697,
"loss": 2.012480545043945,
"step": 25830
},
{
"epoch": 3.307328,
"grad_norm": 0.5980732440948486,
"learning_rate": 0.00022676129416202633,
"loss": 1.9826223373413085,
"step": 25840
},
{
"epoch": 3.308608,
"grad_norm": 0.5125299096107483,
"learning_rate": 0.00022670933619886198,
"loss": 1.9835927963256836,
"step": 25850
},
{
"epoch": 3.309888,
"grad_norm": 0.6313246488571167,
"learning_rate": 0.0002266573657693678,
"loss": 1.9833383560180664,
"step": 25860
},
{
"epoch": 3.311168,
"grad_norm": 0.5341280698776245,
"learning_rate": 0.00022660538288198982,
"loss": 1.9842321395874023,
"step": 25870
},
{
"epoch": 3.312448,
"grad_norm": 0.5254926085472107,
"learning_rate": 0.00022655338754517585,
"loss": 1.9739671707153321,
"step": 25880
},
{
"epoch": 3.3137280000000002,
"grad_norm": 0.5650182962417603,
"learning_rate": 0.00022650137976737596,
"loss": 1.9911828994750977,
"step": 25890
},
{
"epoch": 3.315008,
"grad_norm": 0.5924813151359558,
"learning_rate": 0.00022644935955704206,
"loss": 1.964651107788086,
"step": 25900
},
{
"epoch": 3.316288,
"grad_norm": 0.5930665731430054,
"learning_rate": 0.00022639732692262814,
"loss": 1.9989532470703124,
"step": 25910
},
{
"epoch": 3.317568,
"grad_norm": 0.5740436315536499,
"learning_rate": 0.0002263452818725903,
"loss": 2.0066146850585938,
"step": 25920
},
{
"epoch": 3.318848,
"grad_norm": 0.5584340691566467,
"learning_rate": 0.0002262932244153866,
"loss": 1.987064743041992,
"step": 25930
},
{
"epoch": 3.320128,
"grad_norm": 0.7139118313789368,
"learning_rate": 0.00022624115455947694,
"loss": 1.9751407623291015,
"step": 25940
},
{
"epoch": 3.321408,
"grad_norm": 0.6018709540367126,
"learning_rate": 0.00022618907231332358,
"loss": 1.995305061340332,
"step": 25950
},
{
"epoch": 3.322688,
"grad_norm": 0.6208556294441223,
"learning_rate": 0.0002261369776853904,
"loss": 1.957587432861328,
"step": 25960
},
{
"epoch": 3.323968,
"grad_norm": 0.5538727045059204,
"learning_rate": 0.00022608487068414375,
"loss": 1.992540168762207,
"step": 25970
},
{
"epoch": 3.325248,
"grad_norm": 0.5340443253517151,
"learning_rate": 0.00022603275131805156,
"loss": 1.9694568634033203,
"step": 25980
},
{
"epoch": 3.326528,
"grad_norm": 0.6680694818496704,
"learning_rate": 0.000225980619595584,
"loss": 1.974074935913086,
"step": 25990
},
{
"epoch": 3.327808,
"grad_norm": 0.540829598903656,
"learning_rate": 0.0002259284755252132,
"loss": 1.986400032043457,
"step": 26000
},
{
"epoch": 3.329088,
"grad_norm": 0.5475504398345947,
"learning_rate": 0.00022587631911541326,
"loss": 1.9922298431396483,
"step": 26010
},
{
"epoch": 3.330368,
"grad_norm": 0.5302706360816956,
"learning_rate": 0.00022582415037466038,
"loss": 1.9848403930664062,
"step": 26020
},
{
"epoch": 3.331648,
"grad_norm": 0.5404679775238037,
"learning_rate": 0.00022577196931143265,
"loss": 2.0109537124633787,
"step": 26030
},
{
"epoch": 3.332928,
"grad_norm": 0.6001347899436951,
"learning_rate": 0.00022571977593421023,
"loss": 1.9888656616210938,
"step": 26040
},
{
"epoch": 3.334208,
"grad_norm": 0.5916277170181274,
"learning_rate": 0.00022566757025147533,
"loss": 1.97720947265625,
"step": 26050
},
{
"epoch": 3.335488,
"grad_norm": 0.6411023139953613,
"learning_rate": 0.000225615352271712,
"loss": 2.0060354232788087,
"step": 26060
},
{
"epoch": 3.336768,
"grad_norm": 0.5732954144477844,
"learning_rate": 0.00022556312200340642,
"loss": 1.980931854248047,
"step": 26070
},
{
"epoch": 3.338048,
"grad_norm": 0.5761641263961792,
"learning_rate": 0.00022551087945504677,
"loss": 1.9934026718139648,
"step": 26080
},
{
"epoch": 3.339328,
"grad_norm": 0.5444904565811157,
"learning_rate": 0.00022545862463512308,
"loss": 1.9966951370239259,
"step": 26090
},
{
"epoch": 3.340608,
"grad_norm": 0.5345814824104309,
"learning_rate": 0.00022540635755212768,
"loss": 1.9774633407592774,
"step": 26100
},
{
"epoch": 3.341888,
"grad_norm": 0.6960216164588928,
"learning_rate": 0.0002253540782145545,
"loss": 1.992002296447754,
"step": 26110
},
{
"epoch": 3.343168,
"grad_norm": 0.558958113193512,
"learning_rate": 0.0002253017866308997,
"loss": 2.0043169021606446,
"step": 26120
},
{
"epoch": 3.344448,
"grad_norm": 0.6332840323448181,
"learning_rate": 0.00022524948280966146,
"loss": 1.9771883010864257,
"step": 26130
},
{
"epoch": 3.345728,
"grad_norm": 0.5647940039634705,
"learning_rate": 0.0002251971667593398,
"loss": 2.0034500122070313,
"step": 26140
},
{
"epoch": 3.3470079999999998,
"grad_norm": 0.616619884967804,
"learning_rate": 0.00022514483848843687,
"loss": 1.9721206665039062,
"step": 26150
},
{
"epoch": 3.348288,
"grad_norm": 0.5086396932601929,
"learning_rate": 0.00022509249800545673,
"loss": 1.9859348297119142,
"step": 26160
},
{
"epoch": 3.349568,
"grad_norm": 1.1445276737213135,
"learning_rate": 0.00022504014531890535,
"loss": 1.977640151977539,
"step": 26170
},
{
"epoch": 3.350848,
"grad_norm": 0.5840352773666382,
"learning_rate": 0.0002249877804372908,
"loss": 1.9785541534423827,
"step": 26180
},
{
"epoch": 3.352128,
"grad_norm": 0.6394190788269043,
"learning_rate": 0.00022493540336912318,
"loss": 1.9806873321533203,
"step": 26190
},
{
"epoch": 3.353408,
"grad_norm": 0.5296914577484131,
"learning_rate": 0.00022488301412291442,
"loss": 1.986460304260254,
"step": 26200
},
{
"epoch": 3.354688,
"grad_norm": 0.5996544361114502,
"learning_rate": 0.00022483061270717853,
"loss": 1.9831855773925782,
"step": 26210
},
{
"epoch": 3.355968,
"grad_norm": 0.5922337770462036,
"learning_rate": 0.0002247781991304314,
"loss": 1.975562286376953,
"step": 26220
},
{
"epoch": 3.3572480000000002,
"grad_norm": 0.5477715134620667,
"learning_rate": 0.00022472577340119107,
"loss": 1.9828922271728515,
"step": 26230
},
{
"epoch": 3.358528,
"grad_norm": 0.510864794254303,
"learning_rate": 0.0002246733355279773,
"loss": 1.9963268280029296,
"step": 26240
},
{
"epoch": 3.359808,
"grad_norm": 0.5823022723197937,
"learning_rate": 0.0002246208855193121,
"loss": 1.967047882080078,
"step": 26250
},
{
"epoch": 3.361088,
"grad_norm": 0.6401575803756714,
"learning_rate": 0.0002245684233837193,
"loss": 1.9741010665893555,
"step": 26260
},
{
"epoch": 3.362368,
"grad_norm": 0.5336191654205322,
"learning_rate": 0.00022451594912972469,
"loss": 2.0022872924804687,
"step": 26270
},
{
"epoch": 3.363648,
"grad_norm": 0.5632333755493164,
"learning_rate": 0.0002244634627658561,
"loss": 1.9703046798706054,
"step": 26280
},
{
"epoch": 3.364928,
"grad_norm": 0.6095065474510193,
"learning_rate": 0.00022441096430064324,
"loss": 1.9715984344482422,
"step": 26290
},
{
"epoch": 3.366208,
"grad_norm": 0.5298207402229309,
"learning_rate": 0.00022435845374261786,
"loss": 1.9834218978881837,
"step": 26300
},
{
"epoch": 3.367488,
"grad_norm": 0.602534830570221,
"learning_rate": 0.00022430593110031363,
"loss": 1.9809667587280273,
"step": 26310
},
{
"epoch": 3.368768,
"grad_norm": 0.6661078333854675,
"learning_rate": 0.00022425339638226628,
"loss": 1.988561248779297,
"step": 26320
},
{
"epoch": 3.370048,
"grad_norm": 0.524481475353241,
"learning_rate": 0.00022420084959701338,
"loss": 1.9899652481079102,
"step": 26330
},
{
"epoch": 3.371328,
"grad_norm": 0.7108169198036194,
"learning_rate": 0.00022414829075309453,
"loss": 1.985029411315918,
"step": 26340
},
{
"epoch": 3.372608,
"grad_norm": 0.5233047604560852,
"learning_rate": 0.00022409571985905122,
"loss": 2.0042726516723635,
"step": 26350
},
{
"epoch": 3.373888,
"grad_norm": 0.5562222599983215,
"learning_rate": 0.00022404313692342697,
"loss": 2.000959014892578,
"step": 26360
},
{
"epoch": 3.375168,
"grad_norm": 0.5300836563110352,
"learning_rate": 0.00022399054195476728,
"loss": 1.9846923828125,
"step": 26370
},
{
"epoch": 3.376448,
"grad_norm": 0.5596098303794861,
"learning_rate": 0.00022393793496161958,
"loss": 1.998135757446289,
"step": 26380
},
{
"epoch": 3.377728,
"grad_norm": 0.5750000476837158,
"learning_rate": 0.00022388531595253313,
"loss": 1.9935836791992188,
"step": 26390
},
{
"epoch": 3.379008,
"grad_norm": 0.6085788607597351,
"learning_rate": 0.0002238326849360593,
"loss": 1.9868192672729492,
"step": 26400
},
{
"epoch": 3.380288,
"grad_norm": 0.5175146460533142,
"learning_rate": 0.00022378004192075135,
"loss": 1.983330535888672,
"step": 26410
},
{
"epoch": 3.381568,
"grad_norm": 0.5330653190612793,
"learning_rate": 0.00022372738691516455,
"loss": 1.9800052642822266,
"step": 26420
},
{
"epoch": 3.382848,
"grad_norm": 0.5616282224655151,
"learning_rate": 0.00022367471992785598,
"loss": 1.9885684967041015,
"step": 26430
},
{
"epoch": 3.384128,
"grad_norm": 0.6576964855194092,
"learning_rate": 0.0002236220409673849,
"loss": 1.9771598815917968,
"step": 26440
},
{
"epoch": 3.385408,
"grad_norm": 0.59928297996521,
"learning_rate": 0.0002235693500423122,
"loss": 1.9929443359375,
"step": 26450
},
{
"epoch": 3.386688,
"grad_norm": 0.5475125312805176,
"learning_rate": 0.00022351664716120098,
"loss": 1.9967332839965821,
"step": 26460
},
{
"epoch": 3.387968,
"grad_norm": 0.5720624327659607,
"learning_rate": 0.00022346393233261614,
"loss": 1.9788862228393556,
"step": 26470
},
{
"epoch": 3.389248,
"grad_norm": 0.6221330165863037,
"learning_rate": 0.00022341120556512462,
"loss": 1.9952735900878906,
"step": 26480
},
{
"epoch": 3.3905279999999998,
"grad_norm": 0.4886365830898285,
"learning_rate": 0.00022335846686729523,
"loss": 1.9983684539794921,
"step": 26490
},
{
"epoch": 3.391808,
"grad_norm": 0.585231602191925,
"learning_rate": 0.0002233057162476988,
"loss": 1.972970199584961,
"step": 26500
},
{
"epoch": 3.393088,
"grad_norm": 0.504214882850647,
"learning_rate": 0.0002232529537149079,
"loss": 2.0070215225219727,
"step": 26510
},
{
"epoch": 3.394368,
"grad_norm": 0.5471949577331543,
"learning_rate": 0.00022320017927749728,
"loss": 1.9989194869995117,
"step": 26520
},
{
"epoch": 3.395648,
"grad_norm": 0.5572674870491028,
"learning_rate": 0.00022314739294404344,
"loss": 1.9829879760742188,
"step": 26530
},
{
"epoch": 3.396928,
"grad_norm": 0.5234968066215515,
"learning_rate": 0.00022309459472312496,
"loss": 2.0099626541137696,
"step": 26540
},
{
"epoch": 3.398208,
"grad_norm": 0.5730016231536865,
"learning_rate": 0.00022304178462332228,
"loss": 1.9876594543457031,
"step": 26550
},
{
"epoch": 3.399488,
"grad_norm": 0.557110071182251,
"learning_rate": 0.00022298896265321777,
"loss": 2.0010894775390624,
"step": 26560
},
{
"epoch": 3.4007680000000002,
"grad_norm": 0.5575364232063293,
"learning_rate": 0.00022293612882139566,
"loss": 1.9967948913574218,
"step": 26570
},
{
"epoch": 3.402048,
"grad_norm": 0.8081061244010925,
"learning_rate": 0.00022288328313644222,
"loss": 1.9872371673583984,
"step": 26580
},
{
"epoch": 3.403328,
"grad_norm": 0.6154770255088806,
"learning_rate": 0.00022283042560694566,
"loss": 1.9996952056884765,
"step": 26590
},
{
"epoch": 3.404608,
"grad_norm": 0.5542266368865967,
"learning_rate": 0.000222777556241496,
"loss": 1.98583927154541,
"step": 26600
},
{
"epoch": 3.405888,
"grad_norm": 0.6101301908493042,
"learning_rate": 0.00022272467504868532,
"loss": 1.9941965103149415,
"step": 26610
},
{
"epoch": 3.407168,
"grad_norm": 0.5594171285629272,
"learning_rate": 0.0002226717820371074,
"loss": 2.0043703079223634,
"step": 26620
},
{
"epoch": 3.408448,
"grad_norm": 0.5299621224403381,
"learning_rate": 0.00022261887721535822,
"loss": 1.9584278106689452,
"step": 26630
},
{
"epoch": 3.409728,
"grad_norm": 0.564645528793335,
"learning_rate": 0.0002225659605920355,
"loss": 1.9910812377929688,
"step": 26640
},
{
"epoch": 3.411008,
"grad_norm": 0.5531313419342041,
"learning_rate": 0.00022251303217573894,
"loss": 1.9966140747070313,
"step": 26650
},
{
"epoch": 3.412288,
"grad_norm": 0.6407268047332764,
"learning_rate": 0.0002224600919750701,
"loss": 1.9987991333007813,
"step": 26660
},
{
"epoch": 3.413568,
"grad_norm": 0.5819632411003113,
"learning_rate": 0.00022240713999863255,
"loss": 1.9891889572143555,
"step": 26670
},
{
"epoch": 3.414848,
"grad_norm": 0.5754876732826233,
"learning_rate": 0.00022235417625503167,
"loss": 1.992224884033203,
"step": 26680
},
{
"epoch": 3.416128,
"grad_norm": 0.6382068991661072,
"learning_rate": 0.0002223012007528748,
"loss": 1.9674076080322265,
"step": 26690
},
{
"epoch": 3.417408,
"grad_norm": 0.5638684034347534,
"learning_rate": 0.00022224821350077127,
"loss": 1.9865243911743165,
"step": 26700
},
{
"epoch": 3.418688,
"grad_norm": 0.5460231900215149,
"learning_rate": 0.00022219521450733215,
"loss": 1.9905261993408203,
"step": 26710
},
{
"epoch": 3.419968,
"grad_norm": 0.5359580516815186,
"learning_rate": 0.00022214220378117058,
"loss": 1.9872230529785155,
"step": 26720
},
{
"epoch": 3.421248,
"grad_norm": 0.5470555424690247,
"learning_rate": 0.00022208918133090151,
"loss": 1.9935911178588868,
"step": 26730
},
{
"epoch": 3.422528,
"grad_norm": 0.604181170463562,
"learning_rate": 0.0002220361471651418,
"loss": 1.9790651321411132,
"step": 26740
},
{
"epoch": 3.423808,
"grad_norm": 0.5588186979293823,
"learning_rate": 0.0002219831012925103,
"loss": 1.99047794342041,
"step": 26750
},
{
"epoch": 3.425088,
"grad_norm": 0.6078959107398987,
"learning_rate": 0.00022193004372162765,
"loss": 1.9925849914550782,
"step": 26760
},
{
"epoch": 3.426368,
"grad_norm": 0.6026549935340881,
"learning_rate": 0.00022187697446111643,
"loss": 2.0095529556274414,
"step": 26770
},
{
"epoch": 3.427648,
"grad_norm": 0.496967077255249,
"learning_rate": 0.00022182389351960112,
"loss": 1.9883018493652345,
"step": 26780
},
{
"epoch": 3.428928,
"grad_norm": 0.53969407081604,
"learning_rate": 0.0002217708009057082,
"loss": 1.9825569152832032,
"step": 26790
},
{
"epoch": 3.430208,
"grad_norm": 0.4943142831325531,
"learning_rate": 0.00022171769662806585,
"loss": 1.9934322357177734,
"step": 26800
},
{
"epoch": 3.431488,
"grad_norm": 0.5422329306602478,
"learning_rate": 0.00022166458069530429,
"loss": 1.9912120819091796,
"step": 26810
},
{
"epoch": 3.432768,
"grad_norm": 0.5828312635421753,
"learning_rate": 0.0002216114531160556,
"loss": 1.993133544921875,
"step": 26820
},
{
"epoch": 3.4340479999999998,
"grad_norm": 0.4947112202644348,
"learning_rate": 0.00022155831389895377,
"loss": 1.9754547119140624,
"step": 26830
},
{
"epoch": 3.435328,
"grad_norm": 0.5286633968353271,
"learning_rate": 0.0002215051630526346,
"loss": 1.9901077270507812,
"step": 26840
},
{
"epoch": 3.436608,
"grad_norm": 0.6643515825271606,
"learning_rate": 0.0002214520005857359,
"loss": 1.9919002532958985,
"step": 26850
},
{
"epoch": 3.437888,
"grad_norm": 0.5636131763458252,
"learning_rate": 0.00022139882650689722,
"loss": 1.9926467895507813,
"step": 26860
},
{
"epoch": 3.439168,
"grad_norm": 0.610214352607727,
"learning_rate": 0.00022134564082476022,
"loss": 1.9856657028198241,
"step": 26870
},
{
"epoch": 3.440448,
"grad_norm": 0.5394167900085449,
"learning_rate": 0.00022129244354796823,
"loss": 2.001126289367676,
"step": 26880
},
{
"epoch": 3.441728,
"grad_norm": 0.5697112679481506,
"learning_rate": 0.00022123923468516648,
"loss": 2.001607894897461,
"step": 26890
},
{
"epoch": 3.443008,
"grad_norm": 0.6184862852096558,
"learning_rate": 0.0002211860142450023,
"loss": 1.9962215423583984,
"step": 26900
},
{
"epoch": 3.4442880000000002,
"grad_norm": 0.5632733702659607,
"learning_rate": 0.00022113278223612459,
"loss": 1.9978540420532227,
"step": 26910
},
{
"epoch": 3.445568,
"grad_norm": 0.5086802244186401,
"learning_rate": 0.0002210795386671844,
"loss": 1.9833139419555663,
"step": 26920
},
{
"epoch": 3.446848,
"grad_norm": 0.5562728047370911,
"learning_rate": 0.0002210262835468345,
"loss": 1.9737894058227539,
"step": 26930
},
{
"epoch": 3.448128,
"grad_norm": 0.5738405585289001,
"learning_rate": 0.0002209730168837296,
"loss": 1.979767417907715,
"step": 26940
},
{
"epoch": 3.449408,
"grad_norm": 0.568520724773407,
"learning_rate": 0.0002209197386865263,
"loss": 1.9933338165283203,
"step": 26950
},
{
"epoch": 3.450688,
"grad_norm": 0.5804442763328552,
"learning_rate": 0.00022086644896388298,
"loss": 1.9821647644042968,
"step": 26960
},
{
"epoch": 3.451968,
"grad_norm": 0.4598086476325989,
"learning_rate": 0.00022081314772445999,
"loss": 1.9700786590576171,
"step": 26970
},
{
"epoch": 3.453248,
"grad_norm": 0.5403249859809875,
"learning_rate": 0.0002207598349769195,
"loss": 1.9746721267700196,
"step": 26980
},
{
"epoch": 3.454528,
"grad_norm": 0.5497088432312012,
"learning_rate": 0.00022070651072992562,
"loss": 1.982335662841797,
"step": 26990
},
{
"epoch": 3.455808,
"grad_norm": 0.6411692500114441,
"learning_rate": 0.00022065317499214422,
"loss": 1.9873348236083985,
"step": 27000
},
{
"epoch": 3.457088,
"grad_norm": 0.543631911277771,
"learning_rate": 0.00022059982777224317,
"loss": 1.979437255859375,
"step": 27010
},
{
"epoch": 3.458368,
"grad_norm": 0.5339155793190002,
"learning_rate": 0.00022054646907889203,
"loss": 1.986314582824707,
"step": 27020
},
{
"epoch": 3.459648,
"grad_norm": 0.5448115468025208,
"learning_rate": 0.00022049309892076244,
"loss": 1.9787910461425782,
"step": 27030
},
{
"epoch": 3.460928,
"grad_norm": 0.5912877917289734,
"learning_rate": 0.0002204397173065277,
"loss": 1.990377426147461,
"step": 27040
},
{
"epoch": 3.462208,
"grad_norm": 0.5684054493904114,
"learning_rate": 0.0002203863242448631,
"loss": 1.9756654739379882,
"step": 27050
},
{
"epoch": 3.463488,
"grad_norm": 0.5991564393043518,
"learning_rate": 0.00022033291974444578,
"loss": 1.982996940612793,
"step": 27060
},
{
"epoch": 3.464768,
"grad_norm": 0.5537754893302917,
"learning_rate": 0.00022027950381395465,
"loss": 1.9867424011230468,
"step": 27070
},
{
"epoch": 3.466048,
"grad_norm": 0.6527571082115173,
"learning_rate": 0.00022022607646207055,
"loss": 1.9904155731201172,
"step": 27080
},
{
"epoch": 3.467328,
"grad_norm": 0.5535604953765869,
"learning_rate": 0.0002201726376974762,
"loss": 2.002224349975586,
"step": 27090
},
{
"epoch": 3.468608,
"grad_norm": 0.5156789422035217,
"learning_rate": 0.00022011918752885612,
"loss": 1.9815191268920898,
"step": 27100
},
{
"epoch": 3.469888,
"grad_norm": 0.5747023224830627,
"learning_rate": 0.00022006572596489673,
"loss": 1.9789878845214843,
"step": 27110
},
{
"epoch": 3.471168,
"grad_norm": 0.8120809197425842,
"learning_rate": 0.00022001225301428623,
"loss": 1.9896406173706054,
"step": 27120
},
{
"epoch": 3.472448,
"grad_norm": 0.5500006079673767,
"learning_rate": 0.00021995876868571474,
"loss": 1.992913818359375,
"step": 27130
},
{
"epoch": 3.473728,
"grad_norm": 0.579857349395752,
"learning_rate": 0.00021990527298787422,
"loss": 2.0003177642822267,
"step": 27140
},
{
"epoch": 3.475008,
"grad_norm": 0.526107132434845,
"learning_rate": 0.00021985176592945839,
"loss": 1.9917516708374023,
"step": 27150
},
{
"epoch": 3.476288,
"grad_norm": 0.5414380431175232,
"learning_rate": 0.00021979824751916293,
"loss": 1.9925329208374023,
"step": 27160
},
{
"epoch": 3.4775679999999998,
"grad_norm": 0.5670140385627747,
"learning_rate": 0.00021974471776568542,
"loss": 1.9810220718383789,
"step": 27170
},
{
"epoch": 3.478848,
"grad_norm": 0.529721200466156,
"learning_rate": 0.00021969117667772502,
"loss": 1.9813186645507812,
"step": 27180
},
{
"epoch": 3.480128,
"grad_norm": 0.48862379789352417,
"learning_rate": 0.000219637624263983,
"loss": 1.988352394104004,
"step": 27190
},
{
"epoch": 3.481408,
"grad_norm": 0.5519262552261353,
"learning_rate": 0.00021958406053316235,
"loss": 1.9809022903442384,
"step": 27200
},
{
"epoch": 3.482688,
"grad_norm": 0.5499802231788635,
"learning_rate": 0.00021953048549396792,
"loss": 1.9720312118530274,
"step": 27210
},
{
"epoch": 3.483968,
"grad_norm": 0.5983313918113708,
"learning_rate": 0.00021947689915510633,
"loss": 1.9878826141357422,
"step": 27220
},
{
"epoch": 3.485248,
"grad_norm": 0.5668785572052002,
"learning_rate": 0.0002194233015252863,
"loss": 1.9752262115478516,
"step": 27230
},
{
"epoch": 3.486528,
"grad_norm": 0.5226091146469116,
"learning_rate": 0.00021936969261321794,
"loss": 1.9839624404907226,
"step": 27240
},
{
"epoch": 3.4878080000000002,
"grad_norm": 0.5870707631111145,
"learning_rate": 0.00021931607242761355,
"loss": 1.9687854766845703,
"step": 27250
},
{
"epoch": 3.489088,
"grad_norm": 0.5663028955459595,
"learning_rate": 0.00021926244097718718,
"loss": 1.99157657623291,
"step": 27260
},
{
"epoch": 3.490368,
"grad_norm": 0.5825117230415344,
"learning_rate": 0.00021920879827065466,
"loss": 1.9959434509277343,
"step": 27270
},
{
"epoch": 3.491648,
"grad_norm": 0.6357275247573853,
"learning_rate": 0.00021915514431673368,
"loss": 1.99389705657959,
"step": 27280
},
{
"epoch": 3.492928,
"grad_norm": 0.5106639266014099,
"learning_rate": 0.00021910147912414374,
"loss": 1.9861896514892579,
"step": 27290
},
{
"epoch": 3.494208,
"grad_norm": 0.5409963726997375,
"learning_rate": 0.0002190478027016062,
"loss": 1.9838197708129883,
"step": 27300
},
{
"epoch": 3.495488,
"grad_norm": 0.5385698080062866,
"learning_rate": 0.0002189941150578442,
"loss": 1.9856269836425782,
"step": 27310
},
{
"epoch": 3.496768,
"grad_norm": 0.6536942720413208,
"learning_rate": 0.00021894041620158268,
"loss": 1.986029815673828,
"step": 27320
},
{
"epoch": 3.498048,
"grad_norm": 0.676060140132904,
"learning_rate": 0.0002188867061415485,
"loss": 1.9904102325439452,
"step": 27330
},
{
"epoch": 3.499328,
"grad_norm": 0.6567533016204834,
"learning_rate": 0.0002188329848864703,
"loss": 2.0003116607666014,
"step": 27340
},
{
"epoch": 3.500608,
"grad_norm": 0.6727558374404907,
"learning_rate": 0.0002187792524450785,
"loss": 2.000511360168457,
"step": 27350
},
{
"epoch": 3.501888,
"grad_norm": 0.5893589854240417,
"learning_rate": 0.00021872550882610543,
"loss": 1.9867849349975586,
"step": 27360
},
{
"epoch": 3.503168,
"grad_norm": 0.5534513592720032,
"learning_rate": 0.0002186717540382851,
"loss": 1.9950983047485351,
"step": 27370
},
{
"epoch": 3.504448,
"grad_norm": 0.5639446973800659,
"learning_rate": 0.00021861798809035338,
"loss": 2.0019479751586915,
"step": 27380
},
{
"epoch": 3.505728,
"grad_norm": 0.5824128985404968,
"learning_rate": 0.00021856421099104804,
"loss": 2.00390739440918,
"step": 27390
},
{
"epoch": 3.507008,
"grad_norm": 0.539023756980896,
"learning_rate": 0.0002185104227491086,
"loss": 1.9779850006103517,
"step": 27400
},
{
"epoch": 3.508288,
"grad_norm": 0.6104794144630432,
"learning_rate": 0.00021845662337327643,
"loss": 1.9929458618164062,
"step": 27410
},
{
"epoch": 3.509568,
"grad_norm": 0.5943665504455566,
"learning_rate": 0.00021840281287229462,
"loss": 1.989596176147461,
"step": 27420
},
{
"epoch": 3.510848,
"grad_norm": 0.5819639563560486,
"learning_rate": 0.0002183489912549081,
"loss": 1.994131851196289,
"step": 27430
},
{
"epoch": 3.512128,
"grad_norm": 0.6449604034423828,
"learning_rate": 0.00021829515852986368,
"loss": 1.974349594116211,
"step": 27440
},
{
"epoch": 3.513408,
"grad_norm": 0.5457702875137329,
"learning_rate": 0.00021824131470590986,
"loss": 1.9785287857055665,
"step": 27450
},
{
"epoch": 3.514688,
"grad_norm": 0.6023463010787964,
"learning_rate": 0.00021818745979179714,
"loss": 1.9811319351196288,
"step": 27460
},
{
"epoch": 3.515968,
"grad_norm": 0.6144632697105408,
"learning_rate": 0.00021813359379627758,
"loss": 1.9683956146240233,
"step": 27470
},
{
"epoch": 3.517248,
"grad_norm": 0.5966708660125732,
"learning_rate": 0.00021807971672810513,
"loss": 1.9963912963867188,
"step": 27480
},
{
"epoch": 3.518528,
"grad_norm": 0.5037618279457092,
"learning_rate": 0.00021802582859603564,
"loss": 1.9996419906616212,
"step": 27490
},
{
"epoch": 3.5198080000000003,
"grad_norm": 0.548179566860199,
"learning_rate": 0.00021797192940882664,
"loss": 1.9855461120605469,
"step": 27500
},
{
"epoch": 3.5210879999999998,
"grad_norm": 0.6035701036453247,
"learning_rate": 0.00021791801917523746,
"loss": 1.9923065185546875,
"step": 27510
},
{
"epoch": 3.522368,
"grad_norm": 0.5762219429016113,
"learning_rate": 0.00021786409790402935,
"loss": 1.9889404296875,
"step": 27520
},
{
"epoch": 3.523648,
"grad_norm": 0.530866265296936,
"learning_rate": 0.00021781016560396514,
"loss": 1.9744157791137695,
"step": 27530
},
{
"epoch": 3.524928,
"grad_norm": 0.5759620666503906,
"learning_rate": 0.00021775622228380968,
"loss": 1.989402961730957,
"step": 27540
},
{
"epoch": 3.526208,
"grad_norm": 0.5873883366584778,
"learning_rate": 0.0002177022679523295,
"loss": 1.9913141250610351,
"step": 27550
},
{
"epoch": 3.527488,
"grad_norm": 0.5034027695655823,
"learning_rate": 0.00021764830261829284,
"loss": 1.981001853942871,
"step": 27560
},
{
"epoch": 3.528768,
"grad_norm": 0.6205025315284729,
"learning_rate": 0.0002175943262904699,
"loss": 1.9897785186767578,
"step": 27570
},
{
"epoch": 3.530048,
"grad_norm": 0.5059072375297546,
"learning_rate": 0.00021754033897763257,
"loss": 1.9716508865356446,
"step": 27580
},
{
"epoch": 3.5313280000000002,
"grad_norm": 0.5587280988693237,
"learning_rate": 0.0002174863406885545,
"loss": 2.003332328796387,
"step": 27590
},
{
"epoch": 3.5326079999999997,
"grad_norm": 0.5736079216003418,
"learning_rate": 0.0002174323314320112,
"loss": 1.982004737854004,
"step": 27600
},
{
"epoch": 3.533888,
"grad_norm": 0.600450873374939,
"learning_rate": 0.00021737831121677986,
"loss": 1.9813684463500976,
"step": 27610
},
{
"epoch": 3.535168,
"grad_norm": 0.5454357266426086,
"learning_rate": 0.0002173242800516396,
"loss": 1.9877498626708985,
"step": 27620
},
{
"epoch": 3.536448,
"grad_norm": 0.6329535841941833,
"learning_rate": 0.00021727023794537118,
"loss": 1.9785518646240234,
"step": 27630
},
{
"epoch": 3.537728,
"grad_norm": 0.5240281224250793,
"learning_rate": 0.00021721618490675717,
"loss": 1.9938764572143555,
"step": 27640
},
{
"epoch": 3.539008,
"grad_norm": 0.5118955373764038,
"learning_rate": 0.000217162120944582,
"loss": 1.9886499404907227,
"step": 27650
},
{
"epoch": 3.540288,
"grad_norm": 0.6505560874938965,
"learning_rate": 0.0002171080460676317,
"loss": 1.9969818115234375,
"step": 27660
},
{
"epoch": 3.541568,
"grad_norm": 0.5759509205818176,
"learning_rate": 0.00021705396028469438,
"loss": 2.0024511337280275,
"step": 27670
},
{
"epoch": 3.542848,
"grad_norm": 0.62619549036026,
"learning_rate": 0.00021699986360455956,
"loss": 2.0029170989990233,
"step": 27680
},
{
"epoch": 3.5441279999999997,
"grad_norm": 0.535536527633667,
"learning_rate": 0.00021694575603601875,
"loss": 1.9783918380737304,
"step": 27690
},
{
"epoch": 3.545408,
"grad_norm": 0.5759283900260925,
"learning_rate": 0.0002168916375878652,
"loss": 2.012949562072754,
"step": 27700
},
{
"epoch": 3.546688,
"grad_norm": 0.6536675691604614,
"learning_rate": 0.00021683750826889388,
"loss": 1.9897195816040039,
"step": 27710
},
{
"epoch": 3.547968,
"grad_norm": 0.4974117875099182,
"learning_rate": 0.00021678336808790156,
"loss": 1.9831945419311523,
"step": 27720
},
{
"epoch": 3.549248,
"grad_norm": 0.5179481506347656,
"learning_rate": 0.00021672921705368677,
"loss": 2.000334548950195,
"step": 27730
},
{
"epoch": 3.550528,
"grad_norm": 0.5521727204322815,
"learning_rate": 0.00021667505517504985,
"loss": 1.9734397888183595,
"step": 27740
},
{
"epoch": 3.551808,
"grad_norm": 0.5469948649406433,
"learning_rate": 0.00021662088246079278,
"loss": 1.9953804016113281,
"step": 27750
},
{
"epoch": 3.553088,
"grad_norm": 0.6153531670570374,
"learning_rate": 0.0002165666989197194,
"loss": 1.9672393798828125,
"step": 27760
},
{
"epoch": 3.554368,
"grad_norm": 0.5717499852180481,
"learning_rate": 0.00021651250456063533,
"loss": 1.9851177215576172,
"step": 27770
},
{
"epoch": 3.555648,
"grad_norm": 0.5950706005096436,
"learning_rate": 0.00021645829939234785,
"loss": 1.9822568893432617,
"step": 27780
},
{
"epoch": 3.556928,
"grad_norm": 0.6211817860603333,
"learning_rate": 0.00021640408342366607,
"loss": 1.9743057250976563,
"step": 27790
},
{
"epoch": 3.558208,
"grad_norm": 0.6092166304588318,
"learning_rate": 0.00021634985666340087,
"loss": 1.9791032791137695,
"step": 27800
},
{
"epoch": 3.559488,
"grad_norm": 0.5236141681671143,
"learning_rate": 0.00021629561912036476,
"loss": 1.9728475570678712,
"step": 27810
},
{
"epoch": 3.560768,
"grad_norm": 0.541706383228302,
"learning_rate": 0.00021624137080337216,
"loss": 1.990790557861328,
"step": 27820
},
{
"epoch": 3.562048,
"grad_norm": 0.5353380441665649,
"learning_rate": 0.0002161871117212392,
"loss": 1.9596197128295898,
"step": 27830
},
{
"epoch": 3.5633280000000003,
"grad_norm": 0.608723521232605,
"learning_rate": 0.00021613284188278366,
"loss": 1.9787612915039063,
"step": 27840
},
{
"epoch": 3.5646079999999998,
"grad_norm": 0.4905088245868683,
"learning_rate": 0.0002160785612968252,
"loss": 1.9780260086059571,
"step": 27850
},
{
"epoch": 3.565888,
"grad_norm": 0.6041679978370667,
"learning_rate": 0.00021602426997218517,
"loss": 1.9880363464355468,
"step": 27860
},
{
"epoch": 3.567168,
"grad_norm": 0.5986327528953552,
"learning_rate": 0.00021596996791768656,
"loss": 2.002728271484375,
"step": 27870
},
{
"epoch": 3.568448,
"grad_norm": 0.584351658821106,
"learning_rate": 0.00021591565514215435,
"loss": 1.990821647644043,
"step": 27880
},
{
"epoch": 3.569728,
"grad_norm": 0.6142448782920837,
"learning_rate": 0.000215861331654415,
"loss": 1.9811918258666992,
"step": 27890
},
{
"epoch": 3.571008,
"grad_norm": 0.554929792881012,
"learning_rate": 0.00021580699746329688,
"loss": 1.9977935791015624,
"step": 27900
},
{
"epoch": 3.572288,
"grad_norm": 0.6107694506645203,
"learning_rate": 0.00021575265257763008,
"loss": 1.976531982421875,
"step": 27910
},
{
"epoch": 3.573568,
"grad_norm": 0.5219458341598511,
"learning_rate": 0.00021569829700624635,
"loss": 1.995957374572754,
"step": 27920
},
{
"epoch": 3.5748480000000002,
"grad_norm": 0.5386251211166382,
"learning_rate": 0.00021564393075797924,
"loss": 1.9903945922851562,
"step": 27930
},
{
"epoch": 3.5761279999999998,
"grad_norm": 0.5742165446281433,
"learning_rate": 0.00021558955384166402,
"loss": 1.9949895858764648,
"step": 27940
},
{
"epoch": 3.577408,
"grad_norm": 0.5746867060661316,
"learning_rate": 0.00021553516626613765,
"loss": 1.9978286743164062,
"step": 27950
},
{
"epoch": 3.578688,
"grad_norm": 0.5418838858604431,
"learning_rate": 0.00021548076804023893,
"loss": 1.9855541229248046,
"step": 27960
},
{
"epoch": 3.579968,
"grad_norm": 0.6021138429641724,
"learning_rate": 0.0002154263591728083,
"loss": 1.9962167739868164,
"step": 27970
},
{
"epoch": 3.581248,
"grad_norm": 0.5667937397956848,
"learning_rate": 0.00021537193967268798,
"loss": 1.9953760147094726,
"step": 27980
},
{
"epoch": 3.582528,
"grad_norm": 0.5987287163734436,
"learning_rate": 0.00021531750954872185,
"loss": 1.991709327697754,
"step": 27990
},
{
"epoch": 3.583808,
"grad_norm": 0.5572004914283752,
"learning_rate": 0.00021526306880975554,
"loss": 1.966143798828125,
"step": 28000
},
{
"epoch": 3.585088,
"grad_norm": 0.5169771313667297,
"learning_rate": 0.00021520861746463646,
"loss": 1.9951797485351563,
"step": 28010
},
{
"epoch": 3.5863680000000002,
"grad_norm": 0.49965041875839233,
"learning_rate": 0.00021515415552221374,
"loss": 1.9818460464477539,
"step": 28020
},
{
"epoch": 3.5876479999999997,
"grad_norm": 0.586247444152832,
"learning_rate": 0.0002150996829913382,
"loss": 1.9905126571655274,
"step": 28030
},
{
"epoch": 3.588928,
"grad_norm": 0.5780482292175293,
"learning_rate": 0.0002150451998808623,
"loss": 1.986124610900879,
"step": 28040
},
{
"epoch": 3.590208,
"grad_norm": 0.535301923751831,
"learning_rate": 0.00021499070619964032,
"loss": 1.969681739807129,
"step": 28050
},
{
"epoch": 3.591488,
"grad_norm": 0.5765060782432556,
"learning_rate": 0.00021493620195652829,
"loss": 2.012920379638672,
"step": 28060
},
{
"epoch": 3.592768,
"grad_norm": 0.5878505110740662,
"learning_rate": 0.00021488168716038387,
"loss": 1.980586624145508,
"step": 28070
},
{
"epoch": 3.594048,
"grad_norm": 0.5603281259536743,
"learning_rate": 0.0002148271618200665,
"loss": 1.9970598220825195,
"step": 28080
},
{
"epoch": 3.595328,
"grad_norm": 0.5028684735298157,
"learning_rate": 0.00021477262594443732,
"loss": 1.9816917419433593,
"step": 28090
},
{
"epoch": 3.596608,
"grad_norm": 0.5116479396820068,
"learning_rate": 0.00021471807954235905,
"loss": 1.9734943389892579,
"step": 28100
},
{
"epoch": 3.597888,
"grad_norm": 0.5142369866371155,
"learning_rate": 0.00021466352262269637,
"loss": 1.9714818954467774,
"step": 28110
},
{
"epoch": 3.599168,
"grad_norm": 0.5683803558349609,
"learning_rate": 0.0002146089551943154,
"loss": 1.9573966979980468,
"step": 28120
},
{
"epoch": 3.600448,
"grad_norm": 0.5387345552444458,
"learning_rate": 0.00021455437726608428,
"loss": 1.9688138961791992,
"step": 28130
},
{
"epoch": 3.601728,
"grad_norm": 0.5637044906616211,
"learning_rate": 0.00021449978884687256,
"loss": 2.004421043395996,
"step": 28140
},
{
"epoch": 3.603008,
"grad_norm": 0.5459534525871277,
"learning_rate": 0.00021444518994555162,
"loss": 1.9907466888427734,
"step": 28150
},
{
"epoch": 3.604288,
"grad_norm": 0.574813723564148,
"learning_rate": 0.00021439058057099456,
"loss": 1.9914409637451171,
"step": 28160
},
{
"epoch": 3.605568,
"grad_norm": 0.513828456401825,
"learning_rate": 0.00021433596073207616,
"loss": 1.998603630065918,
"step": 28170
},
{
"epoch": 3.6068480000000003,
"grad_norm": 0.6206038594245911,
"learning_rate": 0.00021428133043767286,
"loss": 1.9774324417114257,
"step": 28180
},
{
"epoch": 3.608128,
"grad_norm": 0.6333898901939392,
"learning_rate": 0.00021422668969666293,
"loss": 1.9794776916503907,
"step": 28190
},
{
"epoch": 3.609408,
"grad_norm": 0.5514783263206482,
"learning_rate": 0.00021417203851792615,
"loss": 1.991372299194336,
"step": 28200
},
{
"epoch": 3.610688,
"grad_norm": 0.6588954925537109,
"learning_rate": 0.00021411737691034416,
"loss": 1.9956167221069336,
"step": 28210
},
{
"epoch": 3.611968,
"grad_norm": 0.6265916228294373,
"learning_rate": 0.00021406270488280018,
"loss": 1.9586856842041016,
"step": 28220
},
{
"epoch": 3.613248,
"grad_norm": 0.5440872311592102,
"learning_rate": 0.0002140080224441792,
"loss": 1.9838111877441407,
"step": 28230
},
{
"epoch": 3.614528,
"grad_norm": 0.56548672914505,
"learning_rate": 0.00021395332960336786,
"loss": 1.9867000579833984,
"step": 28240
},
{
"epoch": 3.615808,
"grad_norm": 0.5862581729888916,
"learning_rate": 0.00021389862636925452,
"loss": 1.998979377746582,
"step": 28250
},
{
"epoch": 3.617088,
"grad_norm": 0.5543072819709778,
"learning_rate": 0.00021384391275072924,
"loss": 2.0067070007324217,
"step": 28260
},
{
"epoch": 3.6183680000000003,
"grad_norm": 0.5610848665237427,
"learning_rate": 0.0002137891887566837,
"loss": 1.9645145416259766,
"step": 28270
},
{
"epoch": 3.6196479999999998,
"grad_norm": 0.5380170345306396,
"learning_rate": 0.00021373445439601128,
"loss": 1.9736440658569336,
"step": 28280
},
{
"epoch": 3.620928,
"grad_norm": 0.5458701252937317,
"learning_rate": 0.0002136797096776071,
"loss": 1.9838777542114259,
"step": 28290
},
{
"epoch": 3.622208,
"grad_norm": 0.5660322308540344,
"learning_rate": 0.00021362495461036795,
"loss": 2.00638370513916,
"step": 28300
},
{
"epoch": 3.623488,
"grad_norm": 0.5156005024909973,
"learning_rate": 0.00021357018920319231,
"loss": 1.9834522247314452,
"step": 28310
},
{
"epoch": 3.624768,
"grad_norm": 0.5439391136169434,
"learning_rate": 0.0002135154134649803,
"loss": 1.976597785949707,
"step": 28320
},
{
"epoch": 3.626048,
"grad_norm": 0.587530791759491,
"learning_rate": 0.00021346062740463368,
"loss": 2.002365493774414,
"step": 28330
},
{
"epoch": 3.627328,
"grad_norm": 0.5792186260223389,
"learning_rate": 0.000213405831031056,
"loss": 1.9705774307250976,
"step": 28340
},
{
"epoch": 3.628608,
"grad_norm": 0.5124531984329224,
"learning_rate": 0.0002133510243531524,
"loss": 1.9941570281982421,
"step": 28350
},
{
"epoch": 3.6298880000000002,
"grad_norm": 0.5746634602546692,
"learning_rate": 0.00021329620737982975,
"loss": 1.9930782318115234,
"step": 28360
},
{
"epoch": 3.6311679999999997,
"grad_norm": 0.5270441770553589,
"learning_rate": 0.0002132413801199966,
"loss": 1.9603591918945313,
"step": 28370
},
{
"epoch": 3.632448,
"grad_norm": 0.5977653861045837,
"learning_rate": 0.000213186542582563,
"loss": 1.9689033508300782,
"step": 28380
},
{
"epoch": 3.633728,
"grad_norm": 0.5912472605705261,
"learning_rate": 0.000213131694776441,
"loss": 1.9960309982299804,
"step": 28390
},
{
"epoch": 3.635008,
"grad_norm": 0.5767495036125183,
"learning_rate": 0.00021307683671054398,
"loss": 1.9865341186523438,
"step": 28400
},
{
"epoch": 3.636288,
"grad_norm": 0.6715563535690308,
"learning_rate": 0.0002130219683937872,
"loss": 1.9919937133789063,
"step": 28410
},
{
"epoch": 3.637568,
"grad_norm": 0.6190712451934814,
"learning_rate": 0.00021296708983508753,
"loss": 1.9822353363037108,
"step": 28420
},
{
"epoch": 3.638848,
"grad_norm": 0.5579679012298584,
"learning_rate": 0.00021291220104336349,
"loss": 1.989960289001465,
"step": 28430
},
{
"epoch": 3.640128,
"grad_norm": 0.5669401288032532,
"learning_rate": 0.00021285730202753525,
"loss": 2.0009809494018556,
"step": 28440
},
{
"epoch": 3.641408,
"grad_norm": 0.546239972114563,
"learning_rate": 0.00021280239279652464,
"loss": 2.0107975006103516,
"step": 28450
},
{
"epoch": 3.642688,
"grad_norm": 0.5876777768135071,
"learning_rate": 0.0002127474733592552,
"loss": 1.989878273010254,
"step": 28460
},
{
"epoch": 3.643968,
"grad_norm": 0.5683197379112244,
"learning_rate": 0.0002126925437246521,
"loss": 1.9948509216308594,
"step": 28470
},
{
"epoch": 3.645248,
"grad_norm": 0.6339470148086548,
"learning_rate": 0.0002126376039016422,
"loss": 1.9756332397460938,
"step": 28480
},
{
"epoch": 3.646528,
"grad_norm": 0.5522933006286621,
"learning_rate": 0.00021258265389915396,
"loss": 1.9939540863037108,
"step": 28490
},
{
"epoch": 3.647808,
"grad_norm": 0.5460194945335388,
"learning_rate": 0.00021252769372611746,
"loss": 1.9832334518432617,
"step": 28500
},
{
"epoch": 3.649088,
"grad_norm": 0.7249252796173096,
"learning_rate": 0.0002124727233914645,
"loss": 1.9871631622314454,
"step": 28510
},
{
"epoch": 3.650368,
"grad_norm": 0.5550248622894287,
"learning_rate": 0.00021241774290412863,
"loss": 1.9926271438598633,
"step": 28520
},
{
"epoch": 3.651648,
"grad_norm": 0.6031103134155273,
"learning_rate": 0.00021236275227304484,
"loss": 1.9721454620361327,
"step": 28530
},
{
"epoch": 3.652928,
"grad_norm": 0.6091546416282654,
"learning_rate": 0.00021230775150714987,
"loss": 1.9983280181884766,
"step": 28540
},
{
"epoch": 3.654208,
"grad_norm": 0.5428852438926697,
"learning_rate": 0.00021225274061538216,
"loss": 1.9815176010131836,
"step": 28550
},
{
"epoch": 3.655488,
"grad_norm": 0.5597147941589355,
"learning_rate": 0.00021219771960668169,
"loss": 1.9842142105102538,
"step": 28560
},
{
"epoch": 3.656768,
"grad_norm": 0.592098593711853,
"learning_rate": 0.00021214268848999016,
"loss": 1.975527000427246,
"step": 28570
},
{
"epoch": 3.658048,
"grad_norm": 0.5894666314125061,
"learning_rate": 0.00021208764727425087,
"loss": 1.9808032989501954,
"step": 28580
},
{
"epoch": 3.659328,
"grad_norm": 0.550961971282959,
"learning_rate": 0.00021203259596840884,
"loss": 1.9934980392456054,
"step": 28590
},
{
"epoch": 3.660608,
"grad_norm": 0.5491728782653809,
"learning_rate": 0.0002119775345814106,
"loss": 1.9789344787597656,
"step": 28600
},
{
"epoch": 3.6618880000000003,
"grad_norm": 0.6378230452537537,
"learning_rate": 0.00021192246312220443,
"loss": 1.9917186737060546,
"step": 28610
},
{
"epoch": 3.6631679999999998,
"grad_norm": 0.5464106202125549,
"learning_rate": 0.0002118673815997402,
"loss": 1.9864612579345704,
"step": 28620
},
{
"epoch": 3.664448,
"grad_norm": 0.5600960850715637,
"learning_rate": 0.00021181229002296934,
"loss": 1.9829896926879882,
"step": 28630
},
{
"epoch": 3.665728,
"grad_norm": 0.6128487586975098,
"learning_rate": 0.00021175718840084514,
"loss": 1.984199333190918,
"step": 28640
},
{
"epoch": 3.667008,
"grad_norm": 0.5754013657569885,
"learning_rate": 0.0002117020767423223,
"loss": 1.9972517013549804,
"step": 28650
},
{
"epoch": 3.668288,
"grad_norm": 0.6718581914901733,
"learning_rate": 0.0002116469550563572,
"loss": 1.9923513412475586,
"step": 28660
},
{
"epoch": 3.669568,
"grad_norm": 0.5851135849952698,
"learning_rate": 0.0002115918233519079,
"loss": 1.9722537994384766,
"step": 28670
},
{
"epoch": 3.670848,
"grad_norm": 0.5529770851135254,
"learning_rate": 0.0002115366816379341,
"loss": 1.9942699432373048,
"step": 28680
},
{
"epoch": 3.672128,
"grad_norm": 0.5680275559425354,
"learning_rate": 0.00021148152992339704,
"loss": 2.005533790588379,
"step": 28690
},
{
"epoch": 3.6734080000000002,
"grad_norm": 0.5226141214370728,
"learning_rate": 0.0002114263682172597,
"loss": 1.9817398071289063,
"step": 28700
},
{
"epoch": 3.6746879999999997,
"grad_norm": 0.5302145481109619,
"learning_rate": 0.00021137119652848657,
"loss": 2.0020517349243163,
"step": 28710
},
{
"epoch": 3.675968,
"grad_norm": 0.5450255870819092,
"learning_rate": 0.00021131601486604383,
"loss": 1.9974952697753907,
"step": 28720
},
{
"epoch": 3.677248,
"grad_norm": 0.5823283791542053,
"learning_rate": 0.00021126082323889926,
"loss": 1.9754655838012696,
"step": 28730
},
{
"epoch": 3.678528,
"grad_norm": 0.5041142106056213,
"learning_rate": 0.00021120562165602224,
"loss": 1.9796337127685546,
"step": 28740
},
{
"epoch": 3.679808,
"grad_norm": 0.618359386920929,
"learning_rate": 0.00021115041012638382,
"loss": 1.9896469116210938,
"step": 28750
},
{
"epoch": 3.681088,
"grad_norm": 0.5301356911659241,
"learning_rate": 0.00021109518865895666,
"loss": 1.995879554748535,
"step": 28760
},
{
"epoch": 3.682368,
"grad_norm": 0.5943943858146667,
"learning_rate": 0.000211039957262715,
"loss": 1.963615608215332,
"step": 28770
},
{
"epoch": 3.683648,
"grad_norm": 0.5753364562988281,
"learning_rate": 0.00021098471594663474,
"loss": 1.9781982421875,
"step": 28780
},
{
"epoch": 3.684928,
"grad_norm": 0.5356373190879822,
"learning_rate": 0.00021092946471969323,
"loss": 1.9934642791748047,
"step": 28790
},
{
"epoch": 3.686208,
"grad_norm": 0.5520766377449036,
"learning_rate": 0.00021087420359086967,
"loss": 1.988004493713379,
"step": 28800
},
{
"epoch": 3.687488,
"grad_norm": 0.6242434978485107,
"learning_rate": 0.00021081893256914478,
"loss": 1.9729034423828125,
"step": 28810
},
{
"epoch": 3.688768,
"grad_norm": 0.6445499062538147,
"learning_rate": 0.00021076365166350076,
"loss": 1.9686292648315429,
"step": 28820
},
{
"epoch": 3.690048,
"grad_norm": 0.6093785166740417,
"learning_rate": 0.00021070836088292163,
"loss": 1.9849760055541992,
"step": 28830
},
{
"epoch": 3.691328,
"grad_norm": 0.5356111526489258,
"learning_rate": 0.00021065306023639282,
"loss": 1.9984220504760741,
"step": 28840
},
{
"epoch": 3.692608,
"grad_norm": 0.6063131093978882,
"learning_rate": 0.00021059774973290155,
"loss": 1.9902584075927734,
"step": 28850
},
{
"epoch": 3.693888,
"grad_norm": 0.5904937386512756,
"learning_rate": 0.00021054242938143647,
"loss": 1.9926582336425782,
"step": 28860
},
{
"epoch": 3.695168,
"grad_norm": 0.5374875068664551,
"learning_rate": 0.00021048709919098784,
"loss": 1.9860862731933593,
"step": 28870
},
{
"epoch": 3.696448,
"grad_norm": 0.5586413145065308,
"learning_rate": 0.00021043175917054782,
"loss": 1.9863187789916992,
"step": 28880
},
{
"epoch": 3.697728,
"grad_norm": 0.6076497435569763,
"learning_rate": 0.00021037640932910966,
"loss": 1.9747234344482423,
"step": 28890
},
{
"epoch": 3.699008,
"grad_norm": 0.5487990975379944,
"learning_rate": 0.00021032104967566863,
"loss": 1.9922229766845703,
"step": 28900
},
{
"epoch": 3.700288,
"grad_norm": 0.5689164996147156,
"learning_rate": 0.0002102656802192214,
"loss": 1.9784936904907227,
"step": 28910
},
{
"epoch": 3.701568,
"grad_norm": 0.4948176443576813,
"learning_rate": 0.0002102103009687662,
"loss": 1.9780052185058594,
"step": 28920
},
{
"epoch": 3.702848,
"grad_norm": 0.5095639228820801,
"learning_rate": 0.00021015491193330306,
"loss": 1.9759859085083007,
"step": 28930
},
{
"epoch": 3.704128,
"grad_norm": 0.5393545031547546,
"learning_rate": 0.00021009951312183343,
"loss": 2.000740623474121,
"step": 28940
},
{
"epoch": 3.7054080000000003,
"grad_norm": 0.5978825092315674,
"learning_rate": 0.00021004410454336032,
"loss": 1.977119255065918,
"step": 28950
},
{
"epoch": 3.7066879999999998,
"grad_norm": 0.5866334438323975,
"learning_rate": 0.00020998868620688847,
"loss": 2.0037263870239257,
"step": 28960
},
{
"epoch": 3.707968,
"grad_norm": 0.5709126591682434,
"learning_rate": 0.00020993325812142404,
"loss": 1.9704929351806642,
"step": 28970
},
{
"epoch": 3.709248,
"grad_norm": 0.5161747932434082,
"learning_rate": 0.00020987782029597491,
"loss": 1.9768083572387696,
"step": 28980
},
{
"epoch": 3.710528,
"grad_norm": 0.515489399433136,
"learning_rate": 0.0002098223727395505,
"loss": 1.9899738311767579,
"step": 28990
},
{
"epoch": 3.711808,
"grad_norm": 0.5386382937431335,
"learning_rate": 0.00020976691546116186,
"loss": 1.979398536682129,
"step": 29000
},
{
"epoch": 3.713088,
"grad_norm": 0.5071958303451538,
"learning_rate": 0.00020971144846982146,
"loss": 1.997949981689453,
"step": 29010
},
{
"epoch": 3.714368,
"grad_norm": 0.5398463010787964,
"learning_rate": 0.00020965597177454348,
"loss": 1.983870506286621,
"step": 29020
},
{
"epoch": 3.715648,
"grad_norm": 0.5263432860374451,
"learning_rate": 0.00020960048538434367,
"loss": 1.9882617950439454,
"step": 29030
},
{
"epoch": 3.7169280000000002,
"grad_norm": 0.6346035599708557,
"learning_rate": 0.00020954498930823937,
"loss": 1.9846860885620117,
"step": 29040
},
{
"epoch": 3.7182079999999997,
"grad_norm": 0.5057269334793091,
"learning_rate": 0.00020948948355524943,
"loss": 1.9782440185546875,
"step": 29050
},
{
"epoch": 3.719488,
"grad_norm": 0.541222095489502,
"learning_rate": 0.0002094339681343943,
"loss": 1.9872257232666015,
"step": 29060
},
{
"epoch": 3.720768,
"grad_norm": 0.5738321542739868,
"learning_rate": 0.000209378443054696,
"loss": 1.9655555725097655,
"step": 29070
},
{
"epoch": 3.722048,
"grad_norm": 0.5881938338279724,
"learning_rate": 0.00020932290832517813,
"loss": 1.9852842330932616,
"step": 29080
},
{
"epoch": 3.723328,
"grad_norm": 0.6293831467628479,
"learning_rate": 0.00020926736395486588,
"loss": 1.9732063293457032,
"step": 29090
},
{
"epoch": 3.724608,
"grad_norm": 0.5662873983383179,
"learning_rate": 0.0002092118099527859,
"loss": 1.9854938507080078,
"step": 29100
},
{
"epoch": 3.725888,
"grad_norm": 0.5348583459854126,
"learning_rate": 0.0002091562463279666,
"loss": 1.985159683227539,
"step": 29110
},
{
"epoch": 3.727168,
"grad_norm": 0.7337904572486877,
"learning_rate": 0.0002091006730894378,
"loss": 1.989360237121582,
"step": 29120
},
{
"epoch": 3.728448,
"grad_norm": 0.548698365688324,
"learning_rate": 0.0002090450902462308,
"loss": 1.987696075439453,
"step": 29130
},
{
"epoch": 3.729728,
"grad_norm": 0.5655688643455505,
"learning_rate": 0.00020898949780737876,
"loss": 1.986178207397461,
"step": 29140
},
{
"epoch": 3.731008,
"grad_norm": 0.5448610782623291,
"learning_rate": 0.00020893389578191608,
"loss": 1.980559730529785,
"step": 29150
},
{
"epoch": 3.732288,
"grad_norm": 0.5854073762893677,
"learning_rate": 0.00020887828417887904,
"loss": 1.978692626953125,
"step": 29160
},
{
"epoch": 3.733568,
"grad_norm": 0.4877609312534332,
"learning_rate": 0.0002088226630073051,
"loss": 1.9733474731445313,
"step": 29170
},
{
"epoch": 3.734848,
"grad_norm": 0.8218435049057007,
"learning_rate": 0.00020876703227623353,
"loss": 1.9814172744750977,
"step": 29180
},
{
"epoch": 3.736128,
"grad_norm": 0.5663524270057678,
"learning_rate": 0.00020871139199470518,
"loss": 1.9863300323486328,
"step": 29190
},
{
"epoch": 3.737408,
"grad_norm": 0.5868373513221741,
"learning_rate": 0.00020865574217176225,
"loss": 1.9814794540405274,
"step": 29200
},
{
"epoch": 3.738688,
"grad_norm": 0.4930722415447235,
"learning_rate": 0.00020860008281644866,
"loss": 1.986924171447754,
"step": 29210
},
{
"epoch": 3.739968,
"grad_norm": 0.6298187971115112,
"learning_rate": 0.00020854441393780988,
"loss": 1.9953485488891602,
"step": 29220
},
{
"epoch": 3.741248,
"grad_norm": 0.5191821455955505,
"learning_rate": 0.0002084887355448928,
"loss": 2.002189826965332,
"step": 29230
},
{
"epoch": 3.742528,
"grad_norm": 0.5548213720321655,
"learning_rate": 0.0002084330476467459,
"loss": 1.978570556640625,
"step": 29240
},
{
"epoch": 3.743808,
"grad_norm": 0.5658348202705383,
"learning_rate": 0.00020837735025241934,
"loss": 1.9735542297363282,
"step": 29250
},
{
"epoch": 3.745088,
"grad_norm": 0.6535835862159729,
"learning_rate": 0.0002083216433709646,
"loss": 1.9949039459228515,
"step": 29260
},
{
"epoch": 3.746368,
"grad_norm": 0.5450387001037598,
"learning_rate": 0.0002082659270114349,
"loss": 1.9937076568603516,
"step": 29270
},
{
"epoch": 3.747648,
"grad_norm": 0.7123132348060608,
"learning_rate": 0.00020821020118288495,
"loss": 2.004031753540039,
"step": 29280
},
{
"epoch": 3.7489280000000003,
"grad_norm": 0.5287041068077087,
"learning_rate": 0.00020815446589437088,
"loss": 1.9786949157714844,
"step": 29290
},
{
"epoch": 3.7502079999999998,
"grad_norm": 0.5034719705581665,
"learning_rate": 0.00020809872115495052,
"loss": 1.9864692687988281,
"step": 29300
},
{
"epoch": 3.751488,
"grad_norm": 0.5581765174865723,
"learning_rate": 0.0002080429669736831,
"loss": 1.9839128494262694,
"step": 29310
},
{
"epoch": 3.752768,
"grad_norm": 0.6420619487762451,
"learning_rate": 0.00020798720335962948,
"loss": 1.997878074645996,
"step": 29320
},
{
"epoch": 3.754048,
"grad_norm": 0.5652140974998474,
"learning_rate": 0.00020793143032185203,
"loss": 1.9833480834960937,
"step": 29330
},
{
"epoch": 3.755328,
"grad_norm": 0.5337820649147034,
"learning_rate": 0.0002078756478694146,
"loss": 1.9760307312011718,
"step": 29340
},
{
"epoch": 3.756608,
"grad_norm": 0.5075362324714661,
"learning_rate": 0.0002078198560113827,
"loss": 1.987095260620117,
"step": 29350
},
{
"epoch": 3.757888,
"grad_norm": 0.575356125831604,
"learning_rate": 0.00020776405475682314,
"loss": 1.9819770812988282,
"step": 29360
},
{
"epoch": 3.759168,
"grad_norm": 0.5982807874679565,
"learning_rate": 0.0002077082441148045,
"loss": 1.9694828033447265,
"step": 29370
},
{
"epoch": 3.7604480000000002,
"grad_norm": 0.5873699188232422,
"learning_rate": 0.00020765242409439682,
"loss": 1.988532829284668,
"step": 29380
},
{
"epoch": 3.7617279999999997,
"grad_norm": 0.5766106247901917,
"learning_rate": 0.00020759659470467152,
"loss": 1.9802318572998048,
"step": 29390
},
{
"epoch": 3.763008,
"grad_norm": 0.7062702775001526,
"learning_rate": 0.00020754075595470168,
"loss": 1.9836202621459962,
"step": 29400
},
{
"epoch": 3.764288,
"grad_norm": 0.5728334188461304,
"learning_rate": 0.0002074849078535619,
"loss": 1.9860113143920899,
"step": 29410
},
{
"epoch": 3.765568,
"grad_norm": 0.515667736530304,
"learning_rate": 0.00020742905041032825,
"loss": 1.9730705261230468,
"step": 29420
},
{
"epoch": 3.766848,
"grad_norm": 0.5372771620750427,
"learning_rate": 0.00020737318363407836,
"loss": 1.980555534362793,
"step": 29430
},
{
"epoch": 3.768128,
"grad_norm": 0.6481306552886963,
"learning_rate": 0.0002073173075338913,
"loss": 1.9992198944091797,
"step": 29440
},
{
"epoch": 3.769408,
"grad_norm": 0.5424243211746216,
"learning_rate": 0.00020726142211884783,
"loss": 1.9547388076782226,
"step": 29450
},
{
"epoch": 3.770688,
"grad_norm": 0.5248872637748718,
"learning_rate": 0.00020720552739802992,
"loss": 1.9673873901367187,
"step": 29460
},
{
"epoch": 3.771968,
"grad_norm": 0.5678983330726624,
"learning_rate": 0.00020714962338052142,
"loss": 1.9932964324951172,
"step": 29470
},
{
"epoch": 3.773248,
"grad_norm": 0.5630221962928772,
"learning_rate": 0.00020709371007540737,
"loss": 1.977674674987793,
"step": 29480
},
{
"epoch": 3.774528,
"grad_norm": 0.6201947331428528,
"learning_rate": 0.00020703778749177454,
"loss": 2.001751518249512,
"step": 29490
},
{
"epoch": 3.775808,
"grad_norm": 0.5714350342750549,
"learning_rate": 0.00020698185563871105,
"loss": 1.991265869140625,
"step": 29500
},
{
"epoch": 3.777088,
"grad_norm": 0.5975375175476074,
"learning_rate": 0.00020692591452530672,
"loss": 1.9838735580444335,
"step": 29510
},
{
"epoch": 3.778368,
"grad_norm": 0.5758293867111206,
"learning_rate": 0.00020686996416065264,
"loss": 1.9839805603027343,
"step": 29520
},
{
"epoch": 3.779648,
"grad_norm": 0.5605624914169312,
"learning_rate": 0.0002068140045538416,
"loss": 1.9763347625732421,
"step": 29530
},
{
"epoch": 3.780928,
"grad_norm": 0.6834815144538879,
"learning_rate": 0.00020675803571396775,
"loss": 1.9914129257202149,
"step": 29540
},
{
"epoch": 3.782208,
"grad_norm": 0.5629785060882568,
"learning_rate": 0.00020670205765012682,
"loss": 2.0023303985595704,
"step": 29550
},
{
"epoch": 3.783488,
"grad_norm": 0.5293847322463989,
"learning_rate": 0.00020664607037141608,
"loss": 2.008919525146484,
"step": 29560
},
{
"epoch": 3.784768,
"grad_norm": 0.6430532932281494,
"learning_rate": 0.00020659007388693413,
"loss": 1.9894001007080078,
"step": 29570
},
{
"epoch": 3.786048,
"grad_norm": 0.5431298017501831,
"learning_rate": 0.00020653406820578127,
"loss": 1.9858413696289063,
"step": 29580
},
{
"epoch": 3.787328,
"grad_norm": 0.5471967458724976,
"learning_rate": 0.00020647805333705912,
"loss": 1.979313278198242,
"step": 29590
},
{
"epoch": 3.788608,
"grad_norm": 0.5782037973403931,
"learning_rate": 0.0002064220292898709,
"loss": 1.958416748046875,
"step": 29600
},
{
"epoch": 3.789888,
"grad_norm": 0.5658454298973083,
"learning_rate": 0.00020636599607332137,
"loss": 1.9677610397338867,
"step": 29610
},
{
"epoch": 3.791168,
"grad_norm": 0.5749077796936035,
"learning_rate": 0.0002063099536965166,
"loss": 2.0072866439819337,
"step": 29620
},
{
"epoch": 3.7924480000000003,
"grad_norm": 0.5311256647109985,
"learning_rate": 0.0002062539021685643,
"loss": 1.9979808807373047,
"step": 29630
},
{
"epoch": 3.7937279999999998,
"grad_norm": 0.5327048301696777,
"learning_rate": 0.0002061978414985736,
"loss": 2.002699851989746,
"step": 29640
},
{
"epoch": 3.795008,
"grad_norm": 0.5706093907356262,
"learning_rate": 0.00020614177169565515,
"loss": 1.9909536361694335,
"step": 29650
},
{
"epoch": 3.796288,
"grad_norm": 0.5305303931236267,
"learning_rate": 0.00020608569276892106,
"loss": 2.0073110580444338,
"step": 29660
},
{
"epoch": 3.797568,
"grad_norm": 0.5891302227973938,
"learning_rate": 0.00020602960472748495,
"loss": 1.9849267959594727,
"step": 29670
},
{
"epoch": 3.798848,
"grad_norm": 0.5420774817466736,
"learning_rate": 0.00020597350758046193,
"loss": 1.9905881881713867,
"step": 29680
},
{
"epoch": 3.800128,
"grad_norm": 0.5280233025550842,
"learning_rate": 0.00020591740133696845,
"loss": 1.9735836029052733,
"step": 29690
},
{
"epoch": 3.801408,
"grad_norm": 0.5348743200302124,
"learning_rate": 0.00020586128600612267,
"loss": 1.9924886703491211,
"step": 29700
},
{
"epoch": 3.802688,
"grad_norm": 0.5282723903656006,
"learning_rate": 0.00020580516159704408,
"loss": 1.9904544830322266,
"step": 29710
},
{
"epoch": 3.8039680000000002,
"grad_norm": 0.5412614941596985,
"learning_rate": 0.00020574902811885365,
"loss": 1.987430953979492,
"step": 29720
},
{
"epoch": 3.8052479999999997,
"grad_norm": 0.5511971116065979,
"learning_rate": 0.0002056928855806739,
"loss": 1.9998693466186523,
"step": 29730
},
{
"epoch": 3.806528,
"grad_norm": 0.5306313037872314,
"learning_rate": 0.00020563673399162866,
"loss": 1.9861530303955077,
"step": 29740
},
{
"epoch": 3.807808,
"grad_norm": 0.5924667716026306,
"learning_rate": 0.00020558057336084342,
"loss": 2.006414794921875,
"step": 29750
},
{
"epoch": 3.809088,
"grad_norm": 0.5124107003211975,
"learning_rate": 0.0002055244036974451,
"loss": 1.978062057495117,
"step": 29760
},
{
"epoch": 3.810368,
"grad_norm": 0.5125358700752258,
"learning_rate": 0.0002054682250105619,
"loss": 2.0081048965454102,
"step": 29770
},
{
"epoch": 3.811648,
"grad_norm": 0.5302267670631409,
"learning_rate": 0.00020541203730932382,
"loss": 1.9879358291625977,
"step": 29780
},
{
"epoch": 3.812928,
"grad_norm": 0.5916584730148315,
"learning_rate": 0.00020535584060286209,
"loss": 1.9677852630615233,
"step": 29790
},
{
"epoch": 3.814208,
"grad_norm": 0.5989006757736206,
"learning_rate": 0.00020529963490030934,
"loss": 1.9797948837280273,
"step": 29800
},
{
"epoch": 3.815488,
"grad_norm": 0.5187551975250244,
"learning_rate": 0.00020524342021079986,
"loss": 2.002546501159668,
"step": 29810
},
{
"epoch": 3.816768,
"grad_norm": 0.582127571105957,
"learning_rate": 0.0002051871965434693,
"loss": 1.978790283203125,
"step": 29820
},
{
"epoch": 3.818048,
"grad_norm": 0.543649435043335,
"learning_rate": 0.0002051309639074548,
"loss": 1.9863243103027344,
"step": 29830
},
{
"epoch": 3.819328,
"grad_norm": 0.5650764107704163,
"learning_rate": 0.00020507472231189497,
"loss": 1.9810718536376952,
"step": 29840
},
{
"epoch": 3.820608,
"grad_norm": 0.5411292314529419,
"learning_rate": 0.00020501847176592982,
"loss": 2.005443000793457,
"step": 29850
},
{
"epoch": 3.821888,
"grad_norm": 0.6170166730880737,
"learning_rate": 0.00020496221227870082,
"loss": 1.9795961380004883,
"step": 29860
},
{
"epoch": 3.823168,
"grad_norm": 0.5266174674034119,
"learning_rate": 0.0002049059438593509,
"loss": 1.9736883163452148,
"step": 29870
},
{
"epoch": 3.824448,
"grad_norm": 0.5521631240844727,
"learning_rate": 0.0002048496665170245,
"loss": 1.9826141357421876,
"step": 29880
},
{
"epoch": 3.825728,
"grad_norm": 0.588783323764801,
"learning_rate": 0.0002047933802608675,
"loss": 1.9838504791259766,
"step": 29890
},
{
"epoch": 3.827008,
"grad_norm": 0.6067324876785278,
"learning_rate": 0.00020473708510002712,
"loss": 1.9863182067871095,
"step": 29900
},
{
"epoch": 3.828288,
"grad_norm": 0.638071596622467,
"learning_rate": 0.00020468078104365217,
"loss": 1.9789527893066405,
"step": 29910
},
{
"epoch": 3.829568,
"grad_norm": 0.5177901983261108,
"learning_rate": 0.00020462446810089278,
"loss": 1.988624382019043,
"step": 29920
},
{
"epoch": 3.830848,
"grad_norm": 0.5844088196754456,
"learning_rate": 0.00020456814628090058,
"loss": 2.0012149810791016,
"step": 29930
},
{
"epoch": 3.832128,
"grad_norm": 0.6778616905212402,
"learning_rate": 0.00020451181559282873,
"loss": 1.99754638671875,
"step": 29940
},
{
"epoch": 3.833408,
"grad_norm": 0.48814669251441956,
"learning_rate": 0.00020445547604583167,
"loss": 1.9948690414428711,
"step": 29950
},
{
"epoch": 3.834688,
"grad_norm": 0.567865788936615,
"learning_rate": 0.0002043991276490654,
"loss": 1.9695240020751954,
"step": 29960
},
{
"epoch": 3.8359680000000003,
"grad_norm": 0.5611554980278015,
"learning_rate": 0.0002043427704116873,
"loss": 2.003871536254883,
"step": 29970
},
{
"epoch": 3.8372479999999998,
"grad_norm": 0.5449111461639404,
"learning_rate": 0.00020428640434285613,
"loss": 1.9721059799194336,
"step": 29980
},
{
"epoch": 3.838528,
"grad_norm": 0.5736333727836609,
"learning_rate": 0.00020423002945173225,
"loss": 1.9730342864990233,
"step": 29990
},
{
"epoch": 3.839808,
"grad_norm": 0.5991712808609009,
"learning_rate": 0.00020417364574747737,
"loss": 1.9920562744140624,
"step": 30000
},
{
"epoch": 3.841088,
"grad_norm": 0.559112012386322,
"learning_rate": 0.00020411725323925457,
"loss": 1.9865190505981445,
"step": 30010
},
{
"epoch": 3.842368,
"grad_norm": 0.6059279441833496,
"learning_rate": 0.00020406085193622845,
"loss": 2.0139705657958986,
"step": 30020
},
{
"epoch": 3.843648,
"grad_norm": 0.5712509155273438,
"learning_rate": 0.00020400444184756497,
"loss": 1.9801897048950194,
"step": 30030
},
{
"epoch": 3.844928,
"grad_norm": 0.5186220407485962,
"learning_rate": 0.00020394802298243157,
"loss": 1.9822515487670898,
"step": 30040
},
{
"epoch": 3.846208,
"grad_norm": 0.550894558429718,
"learning_rate": 0.00020389159534999705,
"loss": 1.9714841842651367,
"step": 30050
},
{
"epoch": 3.8474880000000002,
"grad_norm": 0.5740315318107605,
"learning_rate": 0.00020383515895943177,
"loss": 1.9778398513793944,
"step": 30060
},
{
"epoch": 3.8487679999999997,
"grad_norm": 0.5553203225135803,
"learning_rate": 0.0002037787138199074,
"loss": 1.9868932723999024,
"step": 30070
},
{
"epoch": 3.850048,
"grad_norm": 0.6429391503334045,
"learning_rate": 0.00020372225994059703,
"loss": 1.9733606338500977,
"step": 30080
},
{
"epoch": 3.851328,
"grad_norm": 0.5938458442687988,
"learning_rate": 0.00020366579733067518,
"loss": 2.004710006713867,
"step": 30090
},
{
"epoch": 3.852608,
"grad_norm": 0.5436522364616394,
"learning_rate": 0.00020360932599931784,
"loss": 1.993332862854004,
"step": 30100
},
{
"epoch": 3.853888,
"grad_norm": 0.5521186590194702,
"learning_rate": 0.00020355284595570234,
"loss": 1.9861724853515625,
"step": 30110
},
{
"epoch": 3.855168,
"grad_norm": 0.5518317222595215,
"learning_rate": 0.00020349635720900758,
"loss": 1.979178237915039,
"step": 30120
},
{
"epoch": 3.856448,
"grad_norm": 0.548599898815155,
"learning_rate": 0.0002034398597684136,
"loss": 1.9704620361328125,
"step": 30130
},
{
"epoch": 3.857728,
"grad_norm": 0.5208097696304321,
"learning_rate": 0.00020338335364310218,
"loss": 1.984796905517578,
"step": 30140
},
{
"epoch": 3.859008,
"grad_norm": 0.5526499152183533,
"learning_rate": 0.00020332683884225625,
"loss": 1.9921659469604491,
"step": 30150
},
{
"epoch": 3.860288,
"grad_norm": 0.5405840277671814,
"learning_rate": 0.00020327031537506024,
"loss": 1.990637969970703,
"step": 30160
},
{
"epoch": 3.861568,
"grad_norm": 0.550037145614624,
"learning_rate": 0.00020321378325070002,
"loss": 1.9712507247924804,
"step": 30170
},
{
"epoch": 3.862848,
"grad_norm": 0.581608772277832,
"learning_rate": 0.00020315724247836292,
"loss": 1.9887773513793945,
"step": 30180
},
{
"epoch": 3.864128,
"grad_norm": 0.5157130360603333,
"learning_rate": 0.0002031006930672375,
"loss": 1.9832012176513671,
"step": 30190
},
{
"epoch": 3.865408,
"grad_norm": 0.5875346660614014,
"learning_rate": 0.00020304413502651383,
"loss": 1.9960704803466798,
"step": 30200
},
{
"epoch": 3.866688,
"grad_norm": 0.5131496787071228,
"learning_rate": 0.00020298756836538342,
"loss": 2.004130554199219,
"step": 30210
},
{
"epoch": 3.867968,
"grad_norm": 0.5128728151321411,
"learning_rate": 0.00020293099309303907,
"loss": 1.9925081253051757,
"step": 30220
},
{
"epoch": 3.869248,
"grad_norm": 0.5256906747817993,
"learning_rate": 0.00020287440921867514,
"loss": 1.9943161010742188,
"step": 30230
},
{
"epoch": 3.870528,
"grad_norm": 0.5374130010604858,
"learning_rate": 0.00020281781675148724,
"loss": 1.9870458602905274,
"step": 30240
},
{
"epoch": 3.871808,
"grad_norm": 0.5639137625694275,
"learning_rate": 0.00020276121570067242,
"loss": 1.9721153259277344,
"step": 30250
},
{
"epoch": 3.873088,
"grad_norm": 0.5527125000953674,
"learning_rate": 0.0002027046060754291,
"loss": 1.9646047592163085,
"step": 30260
},
{
"epoch": 3.874368,
"grad_norm": 0.5678842067718506,
"learning_rate": 0.00020264798788495727,
"loss": 1.9808963775634765,
"step": 30270
},
{
"epoch": 3.875648,
"grad_norm": 0.5315206050872803,
"learning_rate": 0.000202591361138458,
"loss": 2.002364158630371,
"step": 30280
},
{
"epoch": 3.876928,
"grad_norm": 0.6290958523750305,
"learning_rate": 0.000202534725845134,
"loss": 1.9690723419189453,
"step": 30290
},
{
"epoch": 3.878208,
"grad_norm": 0.5837169885635376,
"learning_rate": 0.00020247808201418935,
"loss": 1.9779605865478516,
"step": 30300
},
{
"epoch": 3.8794880000000003,
"grad_norm": 0.598372757434845,
"learning_rate": 0.00020242142965482934,
"loss": 1.980092430114746,
"step": 30310
},
{
"epoch": 3.8807679999999998,
"grad_norm": 0.7139766812324524,
"learning_rate": 0.00020236476877626082,
"loss": 1.9786115646362306,
"step": 30320
},
{
"epoch": 3.882048,
"grad_norm": 0.5008103847503662,
"learning_rate": 0.00020230809938769198,
"loss": 1.9874671936035155,
"step": 30330
},
{
"epoch": 3.883328,
"grad_norm": 0.5452128648757935,
"learning_rate": 0.00020225142149833233,
"loss": 1.9882085800170899,
"step": 30340
},
{
"epoch": 3.884608,
"grad_norm": 0.567658007144928,
"learning_rate": 0.00020219473511739287,
"loss": 1.9878250122070313,
"step": 30350
},
{
"epoch": 3.885888,
"grad_norm": 0.5099652409553528,
"learning_rate": 0.0002021380402540859,
"loss": 1.9784721374511718,
"step": 30360
},
{
"epoch": 3.887168,
"grad_norm": 0.5228867530822754,
"learning_rate": 0.00020208133691762515,
"loss": 1.985776710510254,
"step": 30370
},
{
"epoch": 3.888448,
"grad_norm": 0.5330978631973267,
"learning_rate": 0.00020202462511722565,
"loss": 1.97340087890625,
"step": 30380
},
{
"epoch": 3.889728,
"grad_norm": 0.5471864342689514,
"learning_rate": 0.00020196790486210384,
"loss": 1.990519905090332,
"step": 30390
},
{
"epoch": 3.8910080000000002,
"grad_norm": 0.635163426399231,
"learning_rate": 0.00020191117616147762,
"loss": 2.009369659423828,
"step": 30400
},
{
"epoch": 3.8922879999999997,
"grad_norm": 0.5854194164276123,
"learning_rate": 0.00020185443902456614,
"loss": 1.998093032836914,
"step": 30410
},
{
"epoch": 3.893568,
"grad_norm": 0.5346847772598267,
"learning_rate": 0.00020179769346058998,
"loss": 1.9833505630493165,
"step": 30420
},
{
"epoch": 3.894848,
"grad_norm": 0.5997456312179565,
"learning_rate": 0.00020174093947877107,
"loss": 2.0182485580444336,
"step": 30430
},
{
"epoch": 3.896128,
"grad_norm": 0.5178316831588745,
"learning_rate": 0.00020168417708833267,
"loss": 1.9771265029907226,
"step": 30440
},
{
"epoch": 3.897408,
"grad_norm": 0.626436173915863,
"learning_rate": 0.00020162740629849958,
"loss": 1.9975809097290038,
"step": 30450
},
{
"epoch": 3.898688,
"grad_norm": 0.7488899230957031,
"learning_rate": 0.00020157062711849777,
"loss": 1.9910070419311523,
"step": 30460
},
{
"epoch": 3.899968,
"grad_norm": 0.616863489151001,
"learning_rate": 0.00020151383955755464,
"loss": 1.9985078811645507,
"step": 30470
},
{
"epoch": 3.901248,
"grad_norm": 0.49252620339393616,
"learning_rate": 0.00020145704362489896,
"loss": 1.9826824188232421,
"step": 30480
},
{
"epoch": 3.902528,
"grad_norm": 0.5173830986022949,
"learning_rate": 0.00020140023932976083,
"loss": 1.9799739837646484,
"step": 30490
},
{
"epoch": 3.9038079999999997,
"grad_norm": 0.5470153093338013,
"learning_rate": 0.00020134342668137178,
"loss": 1.967597770690918,
"step": 30500
},
{
"epoch": 3.905088,
"grad_norm": 0.543853759765625,
"learning_rate": 0.00020128660568896462,
"loss": 1.979436492919922,
"step": 30510
},
{
"epoch": 3.906368,
"grad_norm": 0.5402469635009766,
"learning_rate": 0.0002012297763617736,
"loss": 1.987936019897461,
"step": 30520
},
{
"epoch": 3.907648,
"grad_norm": 0.5462364554405212,
"learning_rate": 0.0002011729387090342,
"loss": 1.975601577758789,
"step": 30530
},
{
"epoch": 3.908928,
"grad_norm": 0.5932900905609131,
"learning_rate": 0.00020111609273998335,
"loss": 1.960386085510254,
"step": 30540
},
{
"epoch": 3.910208,
"grad_norm": 0.5455654263496399,
"learning_rate": 0.00020105923846385935,
"loss": 1.9911722183227538,
"step": 30550
},
{
"epoch": 3.911488,
"grad_norm": 0.5863791704177856,
"learning_rate": 0.0002010023758899018,
"loss": 1.9720617294311524,
"step": 30560
},
{
"epoch": 3.912768,
"grad_norm": 0.5738329291343689,
"learning_rate": 0.00020094550502735154,
"loss": 1.9839431762695312,
"step": 30570
},
{
"epoch": 3.914048,
"grad_norm": 0.6378779411315918,
"learning_rate": 0.00020088862588545106,
"loss": 1.9898338317871094,
"step": 30580
},
{
"epoch": 3.915328,
"grad_norm": 0.5495602488517761,
"learning_rate": 0.00020083173847344388,
"loss": 1.983449935913086,
"step": 30590
},
{
"epoch": 3.916608,
"grad_norm": 0.6093432903289795,
"learning_rate": 0.000200774842800575,
"loss": 1.9882675170898438,
"step": 30600
},
{
"epoch": 3.917888,
"grad_norm": 0.679192304611206,
"learning_rate": 0.00020071793887609077,
"loss": 2.0041906356811525,
"step": 30610
},
{
"epoch": 3.919168,
"grad_norm": 0.5759492516517639,
"learning_rate": 0.0002006610267092389,
"loss": 1.9799560546875,
"step": 30620
},
{
"epoch": 3.920448,
"grad_norm": 0.558263897895813,
"learning_rate": 0.0002006041063092684,
"loss": 1.9861858367919922,
"step": 30630
},
{
"epoch": 3.921728,
"grad_norm": 0.5205216407775879,
"learning_rate": 0.0002005471776854296,
"loss": 1.9806264877319335,
"step": 30640
},
{
"epoch": 3.9230080000000003,
"grad_norm": 0.5546402335166931,
"learning_rate": 0.00020049024084697417,
"loss": 1.9911415100097656,
"step": 30650
},
{
"epoch": 3.9242879999999998,
"grad_norm": 0.6539683938026428,
"learning_rate": 0.00020043329580315521,
"loss": 2.009406852722168,
"step": 30660
},
{
"epoch": 3.925568,
"grad_norm": 0.494182825088501,
"learning_rate": 0.000200376342563227,
"loss": 1.9875524520874024,
"step": 30670
},
{
"epoch": 3.926848,
"grad_norm": 0.6101253032684326,
"learning_rate": 0.00020031938113644523,
"loss": 1.9945423126220703,
"step": 30680
},
{
"epoch": 3.928128,
"grad_norm": 0.5383170247077942,
"learning_rate": 0.00020026241153206698,
"loss": 1.9889867782592774,
"step": 30690
},
{
"epoch": 3.929408,
"grad_norm": 0.6210159659385681,
"learning_rate": 0.00020020543375935058,
"loss": 1.9935251235961915,
"step": 30700
},
{
"epoch": 3.930688,
"grad_norm": 0.5684430003166199,
"learning_rate": 0.0002001484478275557,
"loss": 2.003163146972656,
"step": 30710
},
{
"epoch": 3.931968,
"grad_norm": 0.5938053131103516,
"learning_rate": 0.00020009145374594328,
"loss": 1.9993892669677735,
"step": 30720
},
{
"epoch": 3.933248,
"grad_norm": 0.6462310552597046,
"learning_rate": 0.0002000344515237758,
"loss": 1.987071418762207,
"step": 30730
},
{
"epoch": 3.9345280000000002,
"grad_norm": 0.47106269001960754,
"learning_rate": 0.00019997744117031678,
"loss": 1.9753177642822266,
"step": 30740
},
{
"epoch": 3.9358079999999998,
"grad_norm": 0.6171962022781372,
"learning_rate": 0.00019992042269483123,
"loss": 1.9926753997802735,
"step": 30750
},
{
"epoch": 3.937088,
"grad_norm": 0.63820481300354,
"learning_rate": 0.00019986339610658543,
"loss": 1.992531967163086,
"step": 30760
},
{
"epoch": 3.938368,
"grad_norm": 0.5814592838287354,
"learning_rate": 0.00019980636141484704,
"loss": 1.9981204986572265,
"step": 30770
},
{
"epoch": 3.939648,
"grad_norm": 0.6870337128639221,
"learning_rate": 0.0001997493186288849,
"loss": 1.9836124420166015,
"step": 30780
},
{
"epoch": 3.940928,
"grad_norm": 0.5421463251113892,
"learning_rate": 0.00019969226775796935,
"loss": 1.964405632019043,
"step": 30790
},
{
"epoch": 3.942208,
"grad_norm": 0.6081421971321106,
"learning_rate": 0.00019963520881137188,
"loss": 1.9815546035766602,
"step": 30800
},
{
"epoch": 3.943488,
"grad_norm": 0.5832293629646301,
"learning_rate": 0.00019957814179836543,
"loss": 1.9935243606567383,
"step": 30810
},
{
"epoch": 3.944768,
"grad_norm": 0.7498242855072021,
"learning_rate": 0.00019952106672822408,
"loss": 1.9812671661376953,
"step": 30820
},
{
"epoch": 3.9460480000000002,
"grad_norm": 0.6353603601455688,
"learning_rate": 0.00019946398361022337,
"loss": 1.9832763671875,
"step": 30830
},
{
"epoch": 3.9473279999999997,
"grad_norm": 0.5163182020187378,
"learning_rate": 0.0001994068924536401,
"loss": 1.9939390182495118,
"step": 30840
},
{
"epoch": 3.948608,
"grad_norm": 0.4961836636066437,
"learning_rate": 0.00019934979326775235,
"loss": 1.9752267837524413,
"step": 30850
},
{
"epoch": 3.949888,
"grad_norm": 0.5573530197143555,
"learning_rate": 0.00019929268606183956,
"loss": 1.9768529891967774,
"step": 30860
},
{
"epoch": 3.951168,
"grad_norm": 0.5306168794631958,
"learning_rate": 0.0001992355708451825,
"loss": 1.9791009902954102,
"step": 30870
},
{
"epoch": 3.952448,
"grad_norm": 0.6373990178108215,
"learning_rate": 0.00019917844762706307,
"loss": 2.0112537384033202,
"step": 30880
},
{
"epoch": 3.953728,
"grad_norm": 0.5437690019607544,
"learning_rate": 0.00019912131641676464,
"loss": 1.995201301574707,
"step": 30890
},
{
"epoch": 3.955008,
"grad_norm": 0.5694563984870911,
"learning_rate": 0.00019906417722357177,
"loss": 1.9776130676269532,
"step": 30900
},
{
"epoch": 3.956288,
"grad_norm": 0.6150057911872864,
"learning_rate": 0.00019900703005677045,
"loss": 1.965877151489258,
"step": 30910
},
{
"epoch": 3.957568,
"grad_norm": 0.5811395049095154,
"learning_rate": 0.00019894987492564788,
"loss": 1.9923713684082032,
"step": 30920
},
{
"epoch": 3.958848,
"grad_norm": 0.5867418646812439,
"learning_rate": 0.00019889271183949248,
"loss": 2.011946105957031,
"step": 30930
},
{
"epoch": 3.960128,
"grad_norm": 0.5507497787475586,
"learning_rate": 0.00019883554080759414,
"loss": 1.9859262466430665,
"step": 30940
},
{
"epoch": 3.961408,
"grad_norm": 0.5661574602127075,
"learning_rate": 0.00019877836183924388,
"loss": 1.9864067077636718,
"step": 30950
},
{
"epoch": 3.962688,
"grad_norm": 0.5792911052703857,
"learning_rate": 0.0001987211749437341,
"loss": 1.9799402236938477,
"step": 30960
},
{
"epoch": 3.963968,
"grad_norm": 0.5232501029968262,
"learning_rate": 0.00019866398013035846,
"loss": 1.9754589080810547,
"step": 30970
},
{
"epoch": 3.965248,
"grad_norm": 0.5879788398742676,
"learning_rate": 0.00019860677740841189,
"loss": 1.969335174560547,
"step": 30980
},
{
"epoch": 3.9665280000000003,
"grad_norm": 0.5124989748001099,
"learning_rate": 0.00019854956678719068,
"loss": 1.981199073791504,
"step": 30990
},
{
"epoch": 3.967808,
"grad_norm": 0.6190961599349976,
"learning_rate": 0.00019849234827599226,
"loss": 1.9984474182128906,
"step": 31000
},
{
"epoch": 3.969088,
"grad_norm": 0.5928915739059448,
"learning_rate": 0.00019843512188411546,
"loss": 1.9831966400146483,
"step": 31010
},
{
"epoch": 3.970368,
"grad_norm": 0.5055640339851379,
"learning_rate": 0.00019837788762086045,
"loss": 1.987381935119629,
"step": 31020
},
{
"epoch": 3.971648,
"grad_norm": 0.589326024055481,
"learning_rate": 0.0001983206454955285,
"loss": 1.9970569610595703,
"step": 31030
},
{
"epoch": 3.972928,
"grad_norm": 0.5640303492546082,
"learning_rate": 0.00019826339551742226,
"loss": 2.0109716415405274,
"step": 31040
},
{
"epoch": 3.974208,
"grad_norm": 0.5930779576301575,
"learning_rate": 0.00019820613769584564,
"loss": 1.985532569885254,
"step": 31050
},
{
"epoch": 3.975488,
"grad_norm": 0.6032840013504028,
"learning_rate": 0.00019814887204010385,
"loss": 1.9989219665527345,
"step": 31060
},
{
"epoch": 3.976768,
"grad_norm": 0.6488826274871826,
"learning_rate": 0.00019809159855950338,
"loss": 1.9778594970703125,
"step": 31070
},
{
"epoch": 3.9780480000000003,
"grad_norm": 0.5536602735519409,
"learning_rate": 0.00019803431726335188,
"loss": 1.9910259246826172,
"step": 31080
},
{
"epoch": 3.9793279999999998,
"grad_norm": 0.5660229325294495,
"learning_rate": 0.00019797702816095846,
"loss": 1.9930913925170899,
"step": 31090
},
{
"epoch": 3.980608,
"grad_norm": 0.6235045194625854,
"learning_rate": 0.00019791973126163335,
"loss": 1.9850702285766602,
"step": 31100
},
{
"epoch": 3.981888,
"grad_norm": 0.6198228001594543,
"learning_rate": 0.000197862426574688,
"loss": 1.99481143951416,
"step": 31110
},
{
"epoch": 3.983168,
"grad_norm": 0.6004642248153687,
"learning_rate": 0.00019780511410943531,
"loss": 1.9768100738525392,
"step": 31120
},
{
"epoch": 3.984448,
"grad_norm": 0.5241604447364807,
"learning_rate": 0.0001977477938751894,
"loss": 1.980127716064453,
"step": 31130
},
{
"epoch": 3.985728,
"grad_norm": 0.5309976935386658,
"learning_rate": 0.00019769046588126544,
"loss": 1.9917280197143554,
"step": 31140
},
{
"epoch": 3.987008,
"grad_norm": 0.6423666477203369,
"learning_rate": 0.00019763313013698022,
"loss": 1.9918857574462892,
"step": 31150
},
{
"epoch": 3.988288,
"grad_norm": 0.5453601479530334,
"learning_rate": 0.0001975757866516514,
"loss": 1.9947084426879882,
"step": 31160
},
{
"epoch": 3.9895680000000002,
"grad_norm": 0.6284031271934509,
"learning_rate": 0.00019751843543459825,
"loss": 1.9872659683227538,
"step": 31170
},
{
"epoch": 3.9908479999999997,
"grad_norm": 0.5978078246116638,
"learning_rate": 0.00019746107649514102,
"loss": 1.9899866104125976,
"step": 31180
},
{
"epoch": 3.992128,
"grad_norm": 0.5772927403450012,
"learning_rate": 0.00019740370984260139,
"loss": 1.9744977951049805,
"step": 31190
},
{
"epoch": 3.993408,
"grad_norm": 0.5741131901741028,
"learning_rate": 0.00019734633548630225,
"loss": 1.9976312637329101,
"step": 31200
},
{
"epoch": 3.994688,
"grad_norm": 0.5574102401733398,
"learning_rate": 0.0001972889534355677,
"loss": 1.984219741821289,
"step": 31210
},
{
"epoch": 3.995968,
"grad_norm": 0.6186317205429077,
"learning_rate": 0.00019723156369972307,
"loss": 1.9665685653686524,
"step": 31220
},
{
"epoch": 3.997248,
"grad_norm": 0.5251540541648865,
"learning_rate": 0.0001971741662880951,
"loss": 1.9627262115478517,
"step": 31230
},
{
"epoch": 3.998528,
"grad_norm": 0.5522096753120422,
"learning_rate": 0.00019711676121001158,
"loss": 1.9839990615844727,
"step": 31240
},
{
"epoch": 3.999808,
"grad_norm": 0.5859482288360596,
"learning_rate": 0.00019705934847480167,
"loss": 1.9857580184936523,
"step": 31250
},
{
"epoch": 4.0,
"eval_loss": 1.9913235902786255,
"eval_runtime": 26.6754,
"eval_samples_per_second": 374.877,
"eval_steps_per_second": 46.86,
"step": 31252
},
{
"epoch": 4.001024,
"grad_norm": 0.6299041509628296,
"learning_rate": 0.0001970019280917957,
"loss": 1.9830154418945312,
"step": 31260
},
{
"epoch": 4.002304,
"grad_norm": 0.5626737475395203,
"learning_rate": 0.00019694450007032532,
"loss": 1.996224594116211,
"step": 31270
},
{
"epoch": 4.003584,
"grad_norm": 0.5618051886558533,
"learning_rate": 0.00019688706441972335,
"loss": 2.005790901184082,
"step": 31280
},
{
"epoch": 4.004864,
"grad_norm": 0.5545622110366821,
"learning_rate": 0.00019682962114932385,
"loss": 1.9890554428100586,
"step": 31290
},
{
"epoch": 4.006144,
"grad_norm": 0.5232539176940918,
"learning_rate": 0.00019677217026846223,
"loss": 1.993143653869629,
"step": 31300
},
{
"epoch": 4.007424,
"grad_norm": 0.6090359091758728,
"learning_rate": 0.000196714711786475,
"loss": 1.9840103149414063,
"step": 31310
},
{
"epoch": 4.008704,
"grad_norm": 0.5603802800178528,
"learning_rate": 0.00019665724571269993,
"loss": 2.007435989379883,
"step": 31320
},
{
"epoch": 4.009984,
"grad_norm": 0.5910757184028625,
"learning_rate": 0.00019659977205647612,
"loss": 1.9896909713745117,
"step": 31330
},
{
"epoch": 4.011264,
"grad_norm": 0.6823562383651733,
"learning_rate": 0.00019654229082714377,
"loss": 1.9678262710571288,
"step": 31340
},
{
"epoch": 4.012544,
"grad_norm": 0.661990225315094,
"learning_rate": 0.00019648480203404445,
"loss": 1.974587631225586,
"step": 31350
},
{
"epoch": 4.013824,
"grad_norm": 0.5895488262176514,
"learning_rate": 0.00019642730568652085,
"loss": 1.9831224441528321,
"step": 31360
},
{
"epoch": 4.015104,
"grad_norm": 0.5408278107643127,
"learning_rate": 0.00019636980179391685,
"loss": 1.991963768005371,
"step": 31370
},
{
"epoch": 4.016384,
"grad_norm": 0.640742301940918,
"learning_rate": 0.0001963122903655778,
"loss": 1.9676647186279297,
"step": 31380
},
{
"epoch": 4.017664,
"grad_norm": 0.6142351627349854,
"learning_rate": 0.00019625477141084992,
"loss": 1.9698530197143556,
"step": 31390
},
{
"epoch": 4.018944,
"grad_norm": 0.5462142825126648,
"learning_rate": 0.00019619724493908096,
"loss": 1.995458221435547,
"step": 31400
},
{
"epoch": 4.020224,
"grad_norm": 0.5107971429824829,
"learning_rate": 0.0001961397109596197,
"loss": 2.0100448608398436,
"step": 31410
},
{
"epoch": 4.021504,
"grad_norm": 0.6024925708770752,
"learning_rate": 0.00019608216948181623,
"loss": 1.9783119201660155,
"step": 31420
},
{
"epoch": 4.022784,
"grad_norm": 0.5744830369949341,
"learning_rate": 0.00019602462051502187,
"loss": 1.9791351318359376,
"step": 31430
},
{
"epoch": 4.024064,
"grad_norm": 0.5777241587638855,
"learning_rate": 0.00019596706406858907,
"loss": 1.9925649642944336,
"step": 31440
},
{
"epoch": 4.025344,
"grad_norm": 0.5455799102783203,
"learning_rate": 0.0001959095001518716,
"loss": 1.9730594635009766,
"step": 31450
},
{
"epoch": 4.026624,
"grad_norm": 0.6336265802383423,
"learning_rate": 0.00019585192877422436,
"loss": 1.9811168670654298,
"step": 31460
},
{
"epoch": 4.027904,
"grad_norm": 0.6307604312896729,
"learning_rate": 0.0001957943499450035,
"loss": 1.9950862884521485,
"step": 31470
},
{
"epoch": 4.029184,
"grad_norm": 0.6669643521308899,
"learning_rate": 0.0001957367636735664,
"loss": 1.9837976455688477,
"step": 31480
},
{
"epoch": 4.030464,
"grad_norm": 0.5565756559371948,
"learning_rate": 0.0001956791699692716,
"loss": 1.9761388778686524,
"step": 31490
},
{
"epoch": 4.031744,
"grad_norm": 0.5867668390274048,
"learning_rate": 0.00019562156884147893,
"loss": 1.9814401626586915,
"step": 31500
},
{
"epoch": 4.033024,
"grad_norm": 0.5480380654335022,
"learning_rate": 0.00019556396029954927,
"loss": 1.9699199676513672,
"step": 31510
},
{
"epoch": 4.034304,
"grad_norm": 0.5411675572395325,
"learning_rate": 0.00019550634435284484,
"loss": 1.988614845275879,
"step": 31520
},
{
"epoch": 4.035584,
"grad_norm": 0.5257730484008789,
"learning_rate": 0.00019544872101072912,
"loss": 1.9837587356567383,
"step": 31530
},
{
"epoch": 4.036864,
"grad_norm": 0.6645423173904419,
"learning_rate": 0.0001953910902825666,
"loss": 1.9780921936035156,
"step": 31540
},
{
"epoch": 4.038144,
"grad_norm": 0.5767107009887695,
"learning_rate": 0.00019533345217772314,
"loss": 1.9912710189819336,
"step": 31550
},
{
"epoch": 4.039424,
"grad_norm": 0.6621580123901367,
"learning_rate": 0.00019527580670556567,
"loss": 1.960418128967285,
"step": 31560
},
{
"epoch": 4.040704,
"grad_norm": 0.5896260738372803,
"learning_rate": 0.00019521815387546238,
"loss": 1.9787818908691406,
"step": 31570
},
{
"epoch": 4.041984,
"grad_norm": 0.5379810929298401,
"learning_rate": 0.00019516049369678272,
"loss": 1.9789587020874024,
"step": 31580
},
{
"epoch": 4.043264,
"grad_norm": 0.6495339274406433,
"learning_rate": 0.00019510282617889722,
"loss": 1.9799354553222657,
"step": 31590
},
{
"epoch": 4.044544,
"grad_norm": 0.5287909507751465,
"learning_rate": 0.00019504515133117765,
"loss": 1.9619480133056642,
"step": 31600
},
{
"epoch": 4.045824,
"grad_norm": 0.4918554723262787,
"learning_rate": 0.00019498746916299698,
"loss": 1.9877389907836913,
"step": 31610
},
{
"epoch": 4.047104,
"grad_norm": 0.558770477771759,
"learning_rate": 0.00019492977968372933,
"loss": 1.9832231521606445,
"step": 31620
},
{
"epoch": 4.048384,
"grad_norm": 0.5642580389976501,
"learning_rate": 0.00019487208290275008,
"loss": 1.972151756286621,
"step": 31630
},
{
"epoch": 4.049664,
"grad_norm": 0.8230816125869751,
"learning_rate": 0.00019481437882943575,
"loss": 1.9912504196166991,
"step": 31640
},
{
"epoch": 4.050944,
"grad_norm": 0.5231454968452454,
"learning_rate": 0.00019475666747316406,
"loss": 1.9839841842651367,
"step": 31650
},
{
"epoch": 4.052224,
"grad_norm": 0.5577403903007507,
"learning_rate": 0.0001946989488433139,
"loss": 1.9962522506713867,
"step": 31660
},
{
"epoch": 4.053504,
"grad_norm": 0.5184106230735779,
"learning_rate": 0.00019464122294926534,
"loss": 1.9722087860107422,
"step": 31670
},
{
"epoch": 4.054784,
"grad_norm": 0.555223286151886,
"learning_rate": 0.00019458348980039958,
"loss": 1.9840164184570312,
"step": 31680
},
{
"epoch": 4.056064,
"grad_norm": 0.5095366835594177,
"learning_rate": 0.00019452574940609916,
"loss": 2.002980041503906,
"step": 31690
},
{
"epoch": 4.057344,
"grad_norm": 0.5739324688911438,
"learning_rate": 0.00019446800177574762,
"loss": 1.9860118865966796,
"step": 31700
},
{
"epoch": 4.058624,
"grad_norm": 0.5177751183509827,
"learning_rate": 0.00019441024691872982,
"loss": 2.0005125045776366,
"step": 31710
},
{
"epoch": 4.059904,
"grad_norm": 0.5459211468696594,
"learning_rate": 0.00019435248484443168,
"loss": 1.9783573150634766,
"step": 31720
},
{
"epoch": 4.061184,
"grad_norm": 0.5414858460426331,
"learning_rate": 0.00019429471556224026,
"loss": 1.9842920303344727,
"step": 31730
},
{
"epoch": 4.062464,
"grad_norm": 0.5159125328063965,
"learning_rate": 0.000194236939081544,
"loss": 1.984844970703125,
"step": 31740
},
{
"epoch": 4.063744,
"grad_norm": 0.48380246758461,
"learning_rate": 0.0001941791554117323,
"loss": 1.9993803024291992,
"step": 31750
},
{
"epoch": 4.065024,
"grad_norm": 0.5028738379478455,
"learning_rate": 0.00019412136456219587,
"loss": 1.983896827697754,
"step": 31760
},
{
"epoch": 4.066304,
"grad_norm": 0.6009868383407593,
"learning_rate": 0.00019406356654232647,
"loss": 1.9693897247314454,
"step": 31770
},
{
"epoch": 4.067584,
"grad_norm": 0.5476009845733643,
"learning_rate": 0.0001940057613615171,
"loss": 1.9709745407104493,
"step": 31780
},
{
"epoch": 4.068864,
"grad_norm": 0.5578610897064209,
"learning_rate": 0.00019394794902916187,
"loss": 1.983639907836914,
"step": 31790
},
{
"epoch": 4.070144,
"grad_norm": 0.6218410134315491,
"learning_rate": 0.00019389012955465615,
"loss": 1.9879779815673828,
"step": 31800
},
{
"epoch": 4.071424,
"grad_norm": 0.5557071566581726,
"learning_rate": 0.00019383230294739635,
"loss": 1.987039566040039,
"step": 31810
},
{
"epoch": 4.072704,
"grad_norm": 0.6463297605514526,
"learning_rate": 0.0001937744692167801,
"loss": 1.9768577575683595,
"step": 31820
},
{
"epoch": 4.073984,
"grad_norm": 0.5399960279464722,
"learning_rate": 0.0001937166283722062,
"loss": 1.9915435791015625,
"step": 31830
},
{
"epoch": 4.075264,
"grad_norm": 0.5559431910514832,
"learning_rate": 0.0001936587804230746,
"loss": 1.9794988632202148,
"step": 31840
},
{
"epoch": 4.076544,
"grad_norm": 0.5098065733909607,
"learning_rate": 0.00019360092537878634,
"loss": 1.9873838424682617,
"step": 31850
},
{
"epoch": 4.077824,
"grad_norm": 0.5517367720603943,
"learning_rate": 0.0001935430632487437,
"loss": 1.994942855834961,
"step": 31860
},
{
"epoch": 4.079104,
"grad_norm": 0.6437150239944458,
"learning_rate": 0.0001934851940423501,
"loss": 2.0045568466186525,
"step": 31870
},
{
"epoch": 4.080384,
"grad_norm": 0.5515249967575073,
"learning_rate": 0.00019342731776901008,
"loss": 1.9759395599365235,
"step": 31880
},
{
"epoch": 4.081664,
"grad_norm": 0.548764705657959,
"learning_rate": 0.00019336943443812934,
"loss": 2.003133201599121,
"step": 31890
},
{
"epoch": 4.082944,
"grad_norm": 0.5144739151000977,
"learning_rate": 0.00019331154405911467,
"loss": 1.9858793258666991,
"step": 31900
},
{
"epoch": 4.084224,
"grad_norm": 0.6169164776802063,
"learning_rate": 0.00019325364664137407,
"loss": 1.9671730041503905,
"step": 31910
},
{
"epoch": 4.085504,
"grad_norm": 0.5277315378189087,
"learning_rate": 0.00019319574219431676,
"loss": 1.9723012924194336,
"step": 31920
},
{
"epoch": 4.086784,
"grad_norm": 0.5037543177604675,
"learning_rate": 0.0001931378307273529,
"loss": 1.9889310836791991,
"step": 31930
},
{
"epoch": 4.088064,
"grad_norm": 0.5634151101112366,
"learning_rate": 0.00019307991224989403,
"loss": 1.9866125106811523,
"step": 31940
},
{
"epoch": 4.089344,
"grad_norm": 0.5284469127655029,
"learning_rate": 0.00019302198677135257,
"loss": 1.9823102951049805,
"step": 31950
},
{
"epoch": 4.090624,
"grad_norm": 0.6562960743904114,
"learning_rate": 0.00019296405430114225,
"loss": 1.9848323822021485,
"step": 31960
},
{
"epoch": 4.091904,
"grad_norm": 0.5423099398612976,
"learning_rate": 0.000192906114848678,
"loss": 1.9788240432739257,
"step": 31970
},
{
"epoch": 4.093184,
"grad_norm": 0.5476163625717163,
"learning_rate": 0.00019284816842337566,
"loss": 1.9982023239135742,
"step": 31980
},
{
"epoch": 4.094464,
"grad_norm": 0.49992725253105164,
"learning_rate": 0.00019279021503465239,
"loss": 1.9578582763671875,
"step": 31990
},
{
"epoch": 4.095744,
"grad_norm": 0.6355798244476318,
"learning_rate": 0.00019273225469192641,
"loss": 1.9672042846679687,
"step": 32000
},
{
"epoch": 4.097024,
"grad_norm": 0.6337215304374695,
"learning_rate": 0.00019267428740461705,
"loss": 1.99415283203125,
"step": 32010
},
{
"epoch": 4.098304,
"grad_norm": 0.643813967704773,
"learning_rate": 0.00019261631318214485,
"loss": 1.9919702529907226,
"step": 32020
},
{
"epoch": 4.099584,
"grad_norm": 0.5398332476615906,
"learning_rate": 0.0001925583320339314,
"loss": 1.9897029876708985,
"step": 32030
},
{
"epoch": 4.100864,
"grad_norm": 0.6040722131729126,
"learning_rate": 0.00019250034396939937,
"loss": 1.989752960205078,
"step": 32040
},
{
"epoch": 4.102144,
"grad_norm": 0.5288349390029907,
"learning_rate": 0.00019244234899797274,
"loss": 1.9982242584228516,
"step": 32050
},
{
"epoch": 4.103424,
"grad_norm": 0.5963200926780701,
"learning_rate": 0.0001923843471290764,
"loss": 1.980311393737793,
"step": 32060
},
{
"epoch": 4.104704,
"grad_norm": 0.553414523601532,
"learning_rate": 0.00019232633837213656,
"loss": 1.9720359802246095,
"step": 32070
},
{
"epoch": 4.105984,
"grad_norm": 0.588015079498291,
"learning_rate": 0.00019226832273658038,
"loss": 1.996793556213379,
"step": 32080
},
{
"epoch": 4.107264,
"grad_norm": 0.5195156931877136,
"learning_rate": 0.0001922103002318362,
"loss": 1.9842557907104492,
"step": 32090
},
{
"epoch": 4.108544,
"grad_norm": 0.5243674516677856,
"learning_rate": 0.00019215227086733346,
"loss": 1.9857933044433593,
"step": 32100
},
{
"epoch": 4.109824,
"grad_norm": 0.5834159255027771,
"learning_rate": 0.0001920942346525028,
"loss": 1.9834598541259765,
"step": 32110
},
{
"epoch": 4.111104,
"grad_norm": 0.6352337598800659,
"learning_rate": 0.0001920361915967759,
"loss": 1.9820768356323242,
"step": 32120
},
{
"epoch": 4.112384,
"grad_norm": 0.5453675985336304,
"learning_rate": 0.0001919781417095855,
"loss": 1.9524999618530274,
"step": 32130
},
{
"epoch": 4.113664,
"grad_norm": 0.5579755306243896,
"learning_rate": 0.00019192008500036554,
"loss": 1.9911815643310546,
"step": 32140
},
{
"epoch": 4.114944,
"grad_norm": 0.6380091309547424,
"learning_rate": 0.00019186202147855106,
"loss": 1.9761613845825194,
"step": 32150
},
{
"epoch": 4.116224,
"grad_norm": 0.592567503452301,
"learning_rate": 0.0001918039511535782,
"loss": 1.9945091247558593,
"step": 32160
},
{
"epoch": 4.117504,
"grad_norm": 0.5931141972541809,
"learning_rate": 0.00019174587403488416,
"loss": 1.985262680053711,
"step": 32170
},
{
"epoch": 4.118784,
"grad_norm": 0.6284838914871216,
"learning_rate": 0.00019168779013190726,
"loss": 1.9713857650756836,
"step": 32180
},
{
"epoch": 4.120064,
"grad_norm": 0.5342831015586853,
"learning_rate": 0.00019162969945408698,
"loss": 1.9840373992919922,
"step": 32190
},
{
"epoch": 4.121344,
"grad_norm": 0.5221065282821655,
"learning_rate": 0.00019157160201086382,
"loss": 1.9761585235595702,
"step": 32200
},
{
"epoch": 4.122624,
"grad_norm": 0.5410021543502808,
"learning_rate": 0.00019151349781167947,
"loss": 1.990442657470703,
"step": 32210
},
{
"epoch": 4.123904,
"grad_norm": 0.5128123760223389,
"learning_rate": 0.0001914553868659766,
"loss": 1.989105796813965,
"step": 32220
},
{
"epoch": 4.125184,
"grad_norm": 0.5067338943481445,
"learning_rate": 0.00019139726918319914,
"loss": 2.0040735244750976,
"step": 32230
},
{
"epoch": 4.126464,
"grad_norm": 0.5250660181045532,
"learning_rate": 0.0001913391447727919,
"loss": 1.9618148803710938,
"step": 32240
},
{
"epoch": 4.127744,
"grad_norm": 0.4967181384563446,
"learning_rate": 0.00019128101364420102,
"loss": 1.9810466766357422,
"step": 32250
},
{
"epoch": 4.129024,
"grad_norm": 0.618277370929718,
"learning_rate": 0.0001912228758068735,
"loss": 1.992283821105957,
"step": 32260
},
{
"epoch": 4.130304,
"grad_norm": 0.6157234311103821,
"learning_rate": 0.00019116473127025763,
"loss": 1.9841623306274414,
"step": 32270
},
{
"epoch": 4.131584,
"grad_norm": 0.523912250995636,
"learning_rate": 0.00019110658004380267,
"loss": 1.9619083404541016,
"step": 32280
},
{
"epoch": 4.132864,
"grad_norm": 0.5995690226554871,
"learning_rate": 0.00019104842213695905,
"loss": 2.009751892089844,
"step": 32290
},
{
"epoch": 4.134144,
"grad_norm": 0.5331161022186279,
"learning_rate": 0.00019099025755917813,
"loss": 1.9802642822265626,
"step": 32300
},
{
"epoch": 4.135424,
"grad_norm": 0.5899742841720581,
"learning_rate": 0.00019093208631991253,
"loss": 1.9880594253540038,
"step": 32310
},
{
"epoch": 4.136704,
"grad_norm": 0.5604852437973022,
"learning_rate": 0.00019087390842861582,
"loss": 1.9692310333251952,
"step": 32320
},
{
"epoch": 4.137984,
"grad_norm": 0.6354756951332092,
"learning_rate": 0.00019081572389474286,
"loss": 1.9707115173339844,
"step": 32330
},
{
"epoch": 4.139264,
"grad_norm": 0.5776991248130798,
"learning_rate": 0.00019075753272774931,
"loss": 1.978531837463379,
"step": 32340
},
{
"epoch": 4.140544,
"grad_norm": 0.5238893628120422,
"learning_rate": 0.0001906993349370921,
"loss": 1.9640451431274415,
"step": 32350
},
{
"epoch": 4.141824,
"grad_norm": 0.5438602566719055,
"learning_rate": 0.00019064113053222912,
"loss": 1.9786006927490234,
"step": 32360
},
{
"epoch": 4.143104,
"grad_norm": 0.59942626953125,
"learning_rate": 0.00019058291952261938,
"loss": 1.9742914199829102,
"step": 32370
},
{
"epoch": 4.144384,
"grad_norm": 0.5296668410301208,
"learning_rate": 0.00019052470191772308,
"loss": 1.9922739028930665,
"step": 32380
},
{
"epoch": 4.145664,
"grad_norm": 0.5321184992790222,
"learning_rate": 0.00019046647772700132,
"loss": 2.0009510040283205,
"step": 32390
},
{
"epoch": 4.146944,
"grad_norm": 0.5334282517433167,
"learning_rate": 0.00019040824695991637,
"loss": 1.9755399703979493,
"step": 32400
},
{
"epoch": 4.148224,
"grad_norm": 0.761122465133667,
"learning_rate": 0.00019035000962593153,
"loss": 1.983025360107422,
"step": 32410
},
{
"epoch": 4.149504,
"grad_norm": 0.5668951869010925,
"learning_rate": 0.00019029176573451108,
"loss": 2.005123519897461,
"step": 32420
},
{
"epoch": 4.150784,
"grad_norm": 0.5432547330856323,
"learning_rate": 0.00019023351529512059,
"loss": 1.9867059707641601,
"step": 32430
},
{
"epoch": 4.152064,
"grad_norm": 0.5667033791542053,
"learning_rate": 0.0001901752583172265,
"loss": 1.9951541900634766,
"step": 32440
},
{
"epoch": 4.153344,
"grad_norm": 0.5591143369674683,
"learning_rate": 0.00019011699481029636,
"loss": 1.9792530059814453,
"step": 32450
},
{
"epoch": 4.154624,
"grad_norm": 0.5710412263870239,
"learning_rate": 0.00019005872478379887,
"loss": 1.9877080917358398,
"step": 32460
},
{
"epoch": 4.155904,
"grad_norm": 0.5680413842201233,
"learning_rate": 0.0001900004482472036,
"loss": 1.9772388458251953,
"step": 32470
},
{
"epoch": 4.157184,
"grad_norm": 0.625892162322998,
"learning_rate": 0.00018994216520998143,
"loss": 1.9977046966552734,
"step": 32480
},
{
"epoch": 4.158464,
"grad_norm": 0.5320836305618286,
"learning_rate": 0.00018988387568160406,
"loss": 1.9836885452270507,
"step": 32490
},
{
"epoch": 4.159744,
"grad_norm": 0.5624877214431763,
"learning_rate": 0.0001898255796715444,
"loss": 1.986583137512207,
"step": 32500
},
{
"epoch": 4.161024,
"grad_norm": 0.5579854249954224,
"learning_rate": 0.00018976727718927633,
"loss": 1.981358528137207,
"step": 32510
},
{
"epoch": 4.162304,
"grad_norm": 0.5090166926383972,
"learning_rate": 0.00018970896824427484,
"loss": 1.9618722915649414,
"step": 32520
},
{
"epoch": 4.163584,
"grad_norm": 0.5622723698616028,
"learning_rate": 0.00018965065284601588,
"loss": 1.9710205078125,
"step": 32530
},
{
"epoch": 4.164864,
"grad_norm": 0.5880756974220276,
"learning_rate": 0.0001895923310039766,
"loss": 1.9757196426391601,
"step": 32540
},
{
"epoch": 4.166144,
"grad_norm": 0.5161948204040527,
"learning_rate": 0.00018953400272763498,
"loss": 1.9737361907958983,
"step": 32550
},
{
"epoch": 4.167424,
"grad_norm": 0.57667475938797,
"learning_rate": 0.0001894756680264703,
"loss": 1.9926284790039062,
"step": 32560
},
{
"epoch": 4.168704,
"grad_norm": 0.5206830501556396,
"learning_rate": 0.0001894173269099628,
"loss": 1.9814916610717774,
"step": 32570
},
{
"epoch": 4.169984,
"grad_norm": 0.5887254476547241,
"learning_rate": 0.0001893589793875935,
"loss": 1.970177459716797,
"step": 32580
},
{
"epoch": 4.171264,
"grad_norm": 0.5800731182098389,
"learning_rate": 0.00018930062546884484,
"loss": 1.9681640625,
"step": 32590
},
{
"epoch": 4.172544,
"grad_norm": 0.5334901213645935,
"learning_rate": 0.0001892422651632001,
"loss": 2.0054132461547853,
"step": 32600
},
{
"epoch": 4.173824,
"grad_norm": 0.5569890737533569,
"learning_rate": 0.00018918389848014366,
"loss": 1.9895713806152344,
"step": 32610
},
{
"epoch": 4.175104,
"grad_norm": 0.5742230415344238,
"learning_rate": 0.00018912552542916096,
"loss": 1.9917278289794922,
"step": 32620
},
{
"epoch": 4.176384,
"grad_norm": 0.6233923435211182,
"learning_rate": 0.0001890671460197383,
"loss": 1.9872404098510743,
"step": 32630
},
{
"epoch": 4.177664,
"grad_norm": 0.5486009120941162,
"learning_rate": 0.00018900876026136323,
"loss": 1.9698661804199218,
"step": 32640
},
{
"epoch": 4.1789439999999995,
"grad_norm": 0.625461220741272,
"learning_rate": 0.00018895036816352423,
"loss": 1.9938188552856446,
"step": 32650
},
{
"epoch": 4.180224,
"grad_norm": 0.5302927494049072,
"learning_rate": 0.00018889196973571084,
"loss": 1.9880809783935547,
"step": 32660
},
{
"epoch": 4.181504,
"grad_norm": 0.6061988472938538,
"learning_rate": 0.00018883356498741358,
"loss": 1.978337287902832,
"step": 32670
},
{
"epoch": 4.182784,
"grad_norm": 0.5159080028533936,
"learning_rate": 0.00018877515392812407,
"loss": 1.9703197479248047,
"step": 32680
},
{
"epoch": 4.184064,
"grad_norm": 0.5505797266960144,
"learning_rate": 0.0001887167365673349,
"loss": 1.9838781356811523,
"step": 32690
},
{
"epoch": 4.185344,
"grad_norm": 0.5347500443458557,
"learning_rate": 0.00018865831291453965,
"loss": 1.981992530822754,
"step": 32700
},
{
"epoch": 4.186624,
"grad_norm": 0.6274113059043884,
"learning_rate": 0.000188599882979233,
"loss": 1.988235092163086,
"step": 32710
},
{
"epoch": 4.187904,
"grad_norm": 0.5352274179458618,
"learning_rate": 0.0001885414467709107,
"loss": 1.9947893142700195,
"step": 32720
},
{
"epoch": 4.189184,
"grad_norm": 0.5334500670433044,
"learning_rate": 0.00018848300429906932,
"loss": 1.9833099365234375,
"step": 32730
},
{
"epoch": 4.190464,
"grad_norm": 0.6110740303993225,
"learning_rate": 0.00018842455557320674,
"loss": 1.9627702713012696,
"step": 32740
},
{
"epoch": 4.191744,
"grad_norm": 0.5029670596122742,
"learning_rate": 0.00018836610060282154,
"loss": 1.98756103515625,
"step": 32750
},
{
"epoch": 4.193024,
"grad_norm": 0.5223727226257324,
"learning_rate": 0.00018830763939741346,
"loss": 1.9864767074584961,
"step": 32760
},
{
"epoch": 4.194304,
"grad_norm": 0.5852099657058716,
"learning_rate": 0.00018824917196648335,
"loss": 1.9927217483520507,
"step": 32770
},
{
"epoch": 4.195584,
"grad_norm": 0.6278187036514282,
"learning_rate": 0.00018819069831953287,
"loss": 1.9811880111694335,
"step": 32780
},
{
"epoch": 4.196864,
"grad_norm": 0.5951676368713379,
"learning_rate": 0.0001881322184660649,
"loss": 1.9854673385620116,
"step": 32790
},
{
"epoch": 4.198144,
"grad_norm": 0.5660651922225952,
"learning_rate": 0.00018807373241558326,
"loss": 1.9939729690551757,
"step": 32800
},
{
"epoch": 4.199424,
"grad_norm": 0.5107500553131104,
"learning_rate": 0.00018801524017759253,
"loss": 1.9970722198486328,
"step": 32810
},
{
"epoch": 4.200704,
"grad_norm": 0.5875138640403748,
"learning_rate": 0.0001879567417615987,
"loss": 1.9913364410400392,
"step": 32820
},
{
"epoch": 4.201984,
"grad_norm": 0.5765213966369629,
"learning_rate": 0.00018789823717710856,
"loss": 1.9853361129760743,
"step": 32830
},
{
"epoch": 4.203264,
"grad_norm": 0.5696060061454773,
"learning_rate": 0.00018783972643362983,
"loss": 1.9879749298095704,
"step": 32840
},
{
"epoch": 4.204544,
"grad_norm": 0.7158764600753784,
"learning_rate": 0.00018778120954067136,
"loss": 1.9796258926391601,
"step": 32850
},
{
"epoch": 4.205824,
"grad_norm": 0.5262233018875122,
"learning_rate": 0.00018772268650774296,
"loss": 1.9872392654418944,
"step": 32860
},
{
"epoch": 4.207104,
"grad_norm": 0.5858497023582458,
"learning_rate": 0.00018766415734435546,
"loss": 2.0109174728393553,
"step": 32870
},
{
"epoch": 4.208384,
"grad_norm": 0.5245158076286316,
"learning_rate": 0.0001876056220600206,
"loss": 1.9781723022460938,
"step": 32880
},
{
"epoch": 4.209664,
"grad_norm": 0.5820968151092529,
"learning_rate": 0.00018754708066425117,
"loss": 1.9814178466796875,
"step": 32890
},
{
"epoch": 4.210944,
"grad_norm": 0.5532392263412476,
"learning_rate": 0.00018748853316656106,
"loss": 1.997748565673828,
"step": 32900
},
{
"epoch": 4.212224,
"grad_norm": 0.5401993989944458,
"learning_rate": 0.00018742997957646497,
"loss": 1.9698808670043946,
"step": 32910
},
{
"epoch": 4.213504,
"grad_norm": 0.5732303261756897,
"learning_rate": 0.00018737141990347867,
"loss": 1.9774776458740235,
"step": 32920
},
{
"epoch": 4.214784,
"grad_norm": 0.5871857404708862,
"learning_rate": 0.00018731285415711892,
"loss": 1.9812442779541015,
"step": 32930
},
{
"epoch": 4.216064,
"grad_norm": 0.5820578932762146,
"learning_rate": 0.00018725428234690343,
"loss": 1.9970525741577148,
"step": 32940
},
{
"epoch": 4.217344,
"grad_norm": 0.5760462880134583,
"learning_rate": 0.00018719570448235103,
"loss": 2.001485061645508,
"step": 32950
},
{
"epoch": 4.218624,
"grad_norm": 0.5382128953933716,
"learning_rate": 0.0001871371205729814,
"loss": 1.9888221740722656,
"step": 32960
},
{
"epoch": 4.219904,
"grad_norm": 0.5304638743400574,
"learning_rate": 0.00018707853062831516,
"loss": 1.9785356521606445,
"step": 32970
},
{
"epoch": 4.221184,
"grad_norm": 0.6414950489997864,
"learning_rate": 0.0001870199346578741,
"loss": 1.9860849380493164,
"step": 32980
},
{
"epoch": 4.222464,
"grad_norm": 0.5588897466659546,
"learning_rate": 0.00018696133267118078,
"loss": 1.9969505310058593,
"step": 32990
},
{
"epoch": 4.223744,
"grad_norm": 0.5319377779960632,
"learning_rate": 0.0001869027246777589,
"loss": 1.9771106719970704,
"step": 33000
},
{
"epoch": 4.225024,
"grad_norm": 0.5611428022384644,
"learning_rate": 0.00018684411068713302,
"loss": 1.972075843811035,
"step": 33010
},
{
"epoch": 4.226304,
"grad_norm": 0.5005406737327576,
"learning_rate": 0.00018678549070882876,
"loss": 1.986722183227539,
"step": 33020
},
{
"epoch": 4.227584,
"grad_norm": 0.5889732837677002,
"learning_rate": 0.0001867268647523727,
"loss": 1.9998470306396485,
"step": 33030
},
{
"epoch": 4.228864,
"grad_norm": 0.5291791558265686,
"learning_rate": 0.00018666823282729227,
"loss": 1.9806386947631835,
"step": 33040
},
{
"epoch": 4.230144,
"grad_norm": 0.5452983975410461,
"learning_rate": 0.0001866095949431161,
"loss": 1.9750017166137694,
"step": 33050
},
{
"epoch": 4.231424,
"grad_norm": 0.5166009664535522,
"learning_rate": 0.0001865509511093736,
"loss": 1.9851669311523437,
"step": 33060
},
{
"epoch": 4.232704,
"grad_norm": 0.5598189234733582,
"learning_rate": 0.00018649230133559516,
"loss": 1.9822526931762696,
"step": 33070
},
{
"epoch": 4.233984,
"grad_norm": 0.5906304717063904,
"learning_rate": 0.0001864336456313123,
"loss": 1.9957168579101563,
"step": 33080
},
{
"epoch": 4.235264,
"grad_norm": 0.5275633335113525,
"learning_rate": 0.00018637498400605725,
"loss": 1.9869281768798828,
"step": 33090
},
{
"epoch": 4.236544,
"grad_norm": 0.5636445879936218,
"learning_rate": 0.0001863163164693634,
"loss": 1.9805118560791015,
"step": 33100
},
{
"epoch": 4.237824,
"grad_norm": 0.5214247107505798,
"learning_rate": 0.00018625764303076503,
"loss": 1.994375991821289,
"step": 33110
},
{
"epoch": 4.239104,
"grad_norm": 0.5097304582595825,
"learning_rate": 0.00018619896369979737,
"loss": 1.9723014831542969,
"step": 33120
},
{
"epoch": 4.240384,
"grad_norm": 0.6612902879714966,
"learning_rate": 0.0001861402784859967,
"loss": 1.9854486465454102,
"step": 33130
},
{
"epoch": 4.241664,
"grad_norm": 0.5815019607543945,
"learning_rate": 0.0001860815873989001,
"loss": 1.9914060592651368,
"step": 33140
},
{
"epoch": 4.242944,
"grad_norm": 0.7632920742034912,
"learning_rate": 0.00018602289044804572,
"loss": 1.9987968444824218,
"step": 33150
},
{
"epoch": 4.244224,
"grad_norm": 0.616214394569397,
"learning_rate": 0.0001859641876429726,
"loss": 1.991929817199707,
"step": 33160
},
{
"epoch": 4.245504,
"grad_norm": 0.533692479133606,
"learning_rate": 0.00018590547899322078,
"loss": 1.9930892944335938,
"step": 33170
},
{
"epoch": 4.246784,
"grad_norm": 0.46948596835136414,
"learning_rate": 0.00018584676450833122,
"loss": 1.9882503509521485,
"step": 33180
},
{
"epoch": 4.248064,
"grad_norm": 0.5388482809066772,
"learning_rate": 0.00018578804419784587,
"loss": 1.9920934677124023,
"step": 33190
},
{
"epoch": 4.249344,
"grad_norm": 0.5786804556846619,
"learning_rate": 0.00018572931807130753,
"loss": 1.9664384841918945,
"step": 33200
},
{
"epoch": 4.250624,
"grad_norm": 0.5860249400138855,
"learning_rate": 0.00018567058613826008,
"loss": 1.9855789184570312,
"step": 33210
},
{
"epoch": 4.251904,
"grad_norm": 0.5384968519210815,
"learning_rate": 0.0001856118484082482,
"loss": 1.985384178161621,
"step": 33220
},
{
"epoch": 4.253184,
"grad_norm": 0.5234634280204773,
"learning_rate": 0.00018555310489081766,
"loss": 2.0004112243652346,
"step": 33230
},
{
"epoch": 4.2544640000000005,
"grad_norm": 0.5759642720222473,
"learning_rate": 0.00018549435559551503,
"loss": 1.9921632766723634,
"step": 33240
},
{
"epoch": 4.255744,
"grad_norm": 0.6346774697303772,
"learning_rate": 0.00018543560053188793,
"loss": 2.0058698654174805,
"step": 33250
},
{
"epoch": 4.257024,
"grad_norm": 0.5759117603302002,
"learning_rate": 0.00018537683970948484,
"loss": 2.0027191162109377,
"step": 33260
},
{
"epoch": 4.258304,
"grad_norm": 0.5591118931770325,
"learning_rate": 0.0001853180731378552,
"loss": 1.9904550552368163,
"step": 33270
},
{
"epoch": 4.259584,
"grad_norm": 0.5281338691711426,
"learning_rate": 0.00018525930082654942,
"loss": 1.9920619964599608,
"step": 33280
},
{
"epoch": 4.260864,
"grad_norm": 0.5985543131828308,
"learning_rate": 0.00018520052278511885,
"loss": 1.9788658142089843,
"step": 33290
},
{
"epoch": 4.262144,
"grad_norm": 0.535639762878418,
"learning_rate": 0.00018514173902311567,
"loss": 1.9624618530273437,
"step": 33300
},
{
"epoch": 4.263424,
"grad_norm": 0.5315822958946228,
"learning_rate": 0.0001850829495500931,
"loss": 1.9837608337402344,
"step": 33310
},
{
"epoch": 4.264704,
"grad_norm": 0.5767773389816284,
"learning_rate": 0.00018502415437560518,
"loss": 1.9701852798461914,
"step": 33320
},
{
"epoch": 4.2659839999999996,
"grad_norm": 0.5310953855514526,
"learning_rate": 0.000184965353509207,
"loss": 1.992589569091797,
"step": 33330
},
{
"epoch": 4.267264,
"grad_norm": 0.5356354117393494,
"learning_rate": 0.00018490654696045453,
"loss": 1.9894405364990235,
"step": 33340
},
{
"epoch": 4.268544,
"grad_norm": 0.6333763599395752,
"learning_rate": 0.00018484773473890457,
"loss": 1.9937240600585937,
"step": 33350
},
{
"epoch": 4.269824,
"grad_norm": 0.6143940091133118,
"learning_rate": 0.000184788916854115,
"loss": 1.9938791275024415,
"step": 33360
},
{
"epoch": 4.271104,
"grad_norm": 0.5527456402778625,
"learning_rate": 0.00018473009331564457,
"loss": 1.9837648391723632,
"step": 33370
},
{
"epoch": 4.272384,
"grad_norm": 0.5052431225776672,
"learning_rate": 0.00018467126413305277,
"loss": 1.9908990859985352,
"step": 33380
},
{
"epoch": 4.273664,
"grad_norm": 0.5940543413162231,
"learning_rate": 0.00018461242931590028,
"loss": 1.9889307022094727,
"step": 33390
},
{
"epoch": 4.274944,
"grad_norm": 0.49247485399246216,
"learning_rate": 0.0001845535888737485,
"loss": 1.9847494125366212,
"step": 33400
},
{
"epoch": 4.276224,
"grad_norm": 0.6349459290504456,
"learning_rate": 0.0001844947428161599,
"loss": 1.9817338943481446,
"step": 33410
},
{
"epoch": 4.277504,
"grad_norm": 0.6028229594230652,
"learning_rate": 0.00018443589115269772,
"loss": 1.977625846862793,
"step": 33420
},
{
"epoch": 4.278784,
"grad_norm": 0.5466931462287903,
"learning_rate": 0.0001843770338929262,
"loss": 1.9893648147583007,
"step": 33430
},
{
"epoch": 4.280064,
"grad_norm": 0.5603229403495789,
"learning_rate": 0.00018431817104641045,
"loss": 1.9892923355102539,
"step": 33440
},
{
"epoch": 4.281344,
"grad_norm": 0.5319015383720398,
"learning_rate": 0.00018425930262271645,
"loss": 1.9996934890747071,
"step": 33450
},
{
"epoch": 4.282624,
"grad_norm": 0.667831540107727,
"learning_rate": 0.00018420042863141124,
"loss": 1.9943742752075195,
"step": 33460
},
{
"epoch": 4.283904,
"grad_norm": 0.5256161689758301,
"learning_rate": 0.00018414154908206258,
"loss": 1.9953142166137696,
"step": 33470
},
{
"epoch": 4.285184,
"grad_norm": 0.6265203356742859,
"learning_rate": 0.00018408266398423924,
"loss": 1.9920263290405273,
"step": 33480
},
{
"epoch": 4.286464,
"grad_norm": 0.6362709403038025,
"learning_rate": 0.00018402377334751082,
"loss": 1.9916799545288086,
"step": 33490
},
{
"epoch": 4.287744,
"grad_norm": 0.5964462161064148,
"learning_rate": 0.00018396487718144792,
"loss": 1.9923322677612305,
"step": 33500
},
{
"epoch": 4.289024,
"grad_norm": 0.5507072806358337,
"learning_rate": 0.00018390597549562196,
"loss": 1.9985431671142577,
"step": 33510
},
{
"epoch": 4.290304,
"grad_norm": 0.5511053204536438,
"learning_rate": 0.00018384706829960526,
"loss": 1.9920490264892579,
"step": 33520
},
{
"epoch": 4.291584,
"grad_norm": 0.5287621021270752,
"learning_rate": 0.00018378815560297112,
"loss": 2.0104290008544923,
"step": 33530
},
{
"epoch": 4.292864,
"grad_norm": 0.5044808983802795,
"learning_rate": 0.00018372923741529363,
"loss": 1.9860254287719727,
"step": 33540
},
{
"epoch": 4.294144,
"grad_norm": 0.5707244277000427,
"learning_rate": 0.0001836703137461478,
"loss": 1.9870615005493164,
"step": 33550
},
{
"epoch": 4.295424,
"grad_norm": 0.5427442193031311,
"learning_rate": 0.0001836113846051095,
"loss": 1.9949073791503906,
"step": 33560
},
{
"epoch": 4.296704,
"grad_norm": 0.5911272764205933,
"learning_rate": 0.00018355245000175563,
"loss": 2.002837371826172,
"step": 33570
},
{
"epoch": 4.297984,
"grad_norm": 0.5736803412437439,
"learning_rate": 0.0001834935099456638,
"loss": 1.9740489959716796,
"step": 33580
},
{
"epoch": 4.299264,
"grad_norm": 0.5234352946281433,
"learning_rate": 0.00018343456444641266,
"loss": 1.997903823852539,
"step": 33590
},
{
"epoch": 4.300544,
"grad_norm": 0.5698778033256531,
"learning_rate": 0.0001833756135135816,
"loss": 1.9910005569458007,
"step": 33600
},
{
"epoch": 4.301824,
"grad_norm": 0.5431295037269592,
"learning_rate": 0.000183316657156751,
"loss": 1.9837766647338868,
"step": 33610
},
{
"epoch": 4.303104,
"grad_norm": 0.5022788643836975,
"learning_rate": 0.00018325769538550207,
"loss": 1.9953033447265625,
"step": 33620
},
{
"epoch": 4.304384,
"grad_norm": 0.5754233002662659,
"learning_rate": 0.00018319872820941692,
"loss": 1.9891111373901367,
"step": 33630
},
{
"epoch": 4.305664,
"grad_norm": 0.5681812763214111,
"learning_rate": 0.00018313975563807853,
"loss": 1.9767908096313476,
"step": 33640
},
{
"epoch": 4.306944,
"grad_norm": 0.5897827744483948,
"learning_rate": 0.00018308077768107081,
"loss": 1.9853351593017579,
"step": 33650
},
{
"epoch": 4.308224,
"grad_norm": 0.5421229600906372,
"learning_rate": 0.0001830217943479784,
"loss": 1.9856372833251954,
"step": 33660
},
{
"epoch": 4.3095040000000004,
"grad_norm": 0.5452390909194946,
"learning_rate": 0.00018296280564838698,
"loss": 1.997075080871582,
"step": 33670
},
{
"epoch": 4.310784,
"grad_norm": 0.5625180602073669,
"learning_rate": 0.00018290381159188297,
"loss": 1.9841188430786132,
"step": 33680
},
{
"epoch": 4.312064,
"grad_norm": 0.48434609174728394,
"learning_rate": 0.0001828448121880538,
"loss": 1.9624549865722656,
"step": 33690
},
{
"epoch": 4.313344,
"grad_norm": 0.5423107147216797,
"learning_rate": 0.00018278580744648763,
"loss": 1.9862525939941407,
"step": 33700
},
{
"epoch": 4.314624,
"grad_norm": 0.5753083825111389,
"learning_rate": 0.0001827267973767736,
"loss": 1.9806907653808594,
"step": 33710
},
{
"epoch": 4.315904,
"grad_norm": 0.555249035358429,
"learning_rate": 0.0001826677819885016,
"loss": 1.9859750747680665,
"step": 33720
},
{
"epoch": 4.317184,
"grad_norm": 0.5242694020271301,
"learning_rate": 0.00018260876129126252,
"loss": 1.9931392669677734,
"step": 33730
},
{
"epoch": 4.318464,
"grad_norm": 0.580824613571167,
"learning_rate": 0.00018254973529464793,
"loss": 1.9954347610473633,
"step": 33740
},
{
"epoch": 4.319744,
"grad_norm": 0.5955199003219604,
"learning_rate": 0.0001824907040082505,
"loss": 1.9964975357055663,
"step": 33750
},
{
"epoch": 4.3210239999999995,
"grad_norm": 0.587718665599823,
"learning_rate": 0.00018243166744166356,
"loss": 1.973196029663086,
"step": 33760
},
{
"epoch": 4.322304,
"grad_norm": 0.6043587327003479,
"learning_rate": 0.0001823726256044814,
"loss": 2.003477668762207,
"step": 33770
},
{
"epoch": 4.323584,
"grad_norm": 0.540220320224762,
"learning_rate": 0.00018231357850629907,
"loss": 1.9934576034545899,
"step": 33780
},
{
"epoch": 4.324864,
"grad_norm": 0.530486524105072,
"learning_rate": 0.00018225452615671262,
"loss": 1.9677120208740235,
"step": 33790
},
{
"epoch": 4.326144,
"grad_norm": 0.5758630633354187,
"learning_rate": 0.00018219546856531886,
"loss": 1.9953496932983399,
"step": 33800
},
{
"epoch": 4.327424,
"grad_norm": 0.5293718576431274,
"learning_rate": 0.00018213640574171543,
"loss": 1.9786293029785156,
"step": 33810
},
{
"epoch": 4.328704,
"grad_norm": 0.6676873564720154,
"learning_rate": 0.00018207733769550093,
"loss": 1.9967527389526367,
"step": 33820
},
{
"epoch": 4.329984,
"grad_norm": 0.6731985807418823,
"learning_rate": 0.00018201826443627465,
"loss": 1.9922828674316406,
"step": 33830
},
{
"epoch": 4.331264,
"grad_norm": 0.587509274482727,
"learning_rate": 0.0001819591859736368,
"loss": 1.9865249633789062,
"step": 33840
},
{
"epoch": 4.332544,
"grad_norm": 0.597335934638977,
"learning_rate": 0.00018190010231718858,
"loss": 1.963678741455078,
"step": 33850
},
{
"epoch": 4.333824,
"grad_norm": 0.5411670804023743,
"learning_rate": 0.00018184101347653176,
"loss": 1.974991226196289,
"step": 33860
},
{
"epoch": 4.335104,
"grad_norm": 0.5857498049736023,
"learning_rate": 0.00018178191946126925,
"loss": 1.9989984512329102,
"step": 33870
},
{
"epoch": 4.336384,
"grad_norm": 0.5691531896591187,
"learning_rate": 0.0001817228202810045,
"loss": 2.000857925415039,
"step": 33880
},
{
"epoch": 4.337664,
"grad_norm": 0.4956691563129425,
"learning_rate": 0.00018166371594534195,
"loss": 1.9887378692626954,
"step": 33890
},
{
"epoch": 4.338944,
"grad_norm": 0.5666823983192444,
"learning_rate": 0.00018160460646388698,
"loss": 1.9716699600219727,
"step": 33900
},
{
"epoch": 4.340224,
"grad_norm": 0.5363352298736572,
"learning_rate": 0.00018154549184624561,
"loss": 1.9823637008666992,
"step": 33910
},
{
"epoch": 4.341504,
"grad_norm": 0.6931799650192261,
"learning_rate": 0.0001814863721020248,
"loss": 2.006851005554199,
"step": 33920
},
{
"epoch": 4.342784,
"grad_norm": 0.5206031799316406,
"learning_rate": 0.00018142724724083242,
"loss": 1.9668689727783204,
"step": 33930
},
{
"epoch": 4.344064,
"grad_norm": 0.5350706577301025,
"learning_rate": 0.0001813681172722769,
"loss": 2.0019535064697265,
"step": 33940
},
{
"epoch": 4.345344,
"grad_norm": 0.5291783213615417,
"learning_rate": 0.00018130898220596785,
"loss": 2.0061113357543947,
"step": 33950
},
{
"epoch": 4.346624,
"grad_norm": 0.5512349605560303,
"learning_rate": 0.00018124984205151546,
"loss": 1.980504608154297,
"step": 33960
},
{
"epoch": 4.347904,
"grad_norm": 0.5972285866737366,
"learning_rate": 0.0001811906968185308,
"loss": 1.9810791015625,
"step": 33970
},
{
"epoch": 4.349184,
"grad_norm": 0.5083039999008179,
"learning_rate": 0.00018113154651662586,
"loss": 2.003070068359375,
"step": 33980
},
{
"epoch": 4.350464,
"grad_norm": 0.5444359183311462,
"learning_rate": 0.00018107239115541332,
"loss": 1.9921743392944335,
"step": 33990
},
{
"epoch": 4.351744,
"grad_norm": 0.6194767355918884,
"learning_rate": 0.00018101323074450678,
"loss": 1.998482894897461,
"step": 34000
},
{
"epoch": 4.353024,
"grad_norm": 0.598288357257843,
"learning_rate": 0.00018095406529352063,
"loss": 1.9889070510864257,
"step": 34010
},
{
"epoch": 4.354304,
"grad_norm": 0.5518839955329895,
"learning_rate": 0.00018089489481207,
"loss": 2.0082700729370115,
"step": 34020
},
{
"epoch": 4.355584,
"grad_norm": 0.5276957750320435,
"learning_rate": 0.00018083571930977107,
"loss": 1.9853933334350586,
"step": 34030
},
{
"epoch": 4.356864,
"grad_norm": 0.5249123573303223,
"learning_rate": 0.0001807765387962405,
"loss": 1.9675382614135741,
"step": 34040
},
{
"epoch": 4.358144,
"grad_norm": 0.5541841387748718,
"learning_rate": 0.00018071735328109608,
"loss": 1.9873579025268555,
"step": 34050
},
{
"epoch": 4.359424,
"grad_norm": 0.48741745948791504,
"learning_rate": 0.00018065816277395625,
"loss": 1.9980735778808594,
"step": 34060
},
{
"epoch": 4.360704,
"grad_norm": 0.5550099611282349,
"learning_rate": 0.0001805989672844402,
"loss": 1.968289566040039,
"step": 34070
},
{
"epoch": 4.361984,
"grad_norm": 0.5684191584587097,
"learning_rate": 0.00018053976682216811,
"loss": 1.9954418182373046,
"step": 34080
},
{
"epoch": 4.363264,
"grad_norm": 0.6188344955444336,
"learning_rate": 0.00018048056139676085,
"loss": 1.9887840270996093,
"step": 34090
},
{
"epoch": 4.364544,
"grad_norm": 0.6258160471916199,
"learning_rate": 0.0001804213510178401,
"loss": 1.9997980117797851,
"step": 34100
},
{
"epoch": 4.365824,
"grad_norm": 0.5543956160545349,
"learning_rate": 0.0001803621356950284,
"loss": 1.9964118957519532,
"step": 34110
},
{
"epoch": 4.367104,
"grad_norm": 0.6366088390350342,
"learning_rate": 0.00018030291543794897,
"loss": 1.9846967697143554,
"step": 34120
},
{
"epoch": 4.368384,
"grad_norm": 0.5422964096069336,
"learning_rate": 0.00018024369025622608,
"loss": 1.9882556915283203,
"step": 34130
},
{
"epoch": 4.369664,
"grad_norm": 0.5620670318603516,
"learning_rate": 0.00018018446015948453,
"loss": 1.9780668258666991,
"step": 34140
},
{
"epoch": 4.370944,
"grad_norm": 0.5503122806549072,
"learning_rate": 0.00018012522515735004,
"loss": 1.9974023818969726,
"step": 34150
},
{
"epoch": 4.372224,
"grad_norm": 0.667686939239502,
"learning_rate": 0.0001800659852594492,
"loss": 1.9877723693847655,
"step": 34160
},
{
"epoch": 4.373504,
"grad_norm": 0.5868141651153564,
"learning_rate": 0.00018000674047540918,
"loss": 1.9948684692382812,
"step": 34170
},
{
"epoch": 4.374784,
"grad_norm": 0.5751412510871887,
"learning_rate": 0.00017994749081485815,
"loss": 1.97469482421875,
"step": 34180
},
{
"epoch": 4.376064,
"grad_norm": 0.4889380931854248,
"learning_rate": 0.00017988823628742505,
"loss": 1.983469581604004,
"step": 34190
},
{
"epoch": 4.377344,
"grad_norm": 0.6088480949401855,
"learning_rate": 0.00017982897690273946,
"loss": 1.977452278137207,
"step": 34200
},
{
"epoch": 4.378624,
"grad_norm": 0.5315030813217163,
"learning_rate": 0.00017976971267043192,
"loss": 1.9929822921752929,
"step": 34210
},
{
"epoch": 4.379904,
"grad_norm": 0.543949544429779,
"learning_rate": 0.00017971044360013373,
"loss": 1.9920137405395508,
"step": 34220
},
{
"epoch": 4.381184,
"grad_norm": 0.5431511402130127,
"learning_rate": 0.0001796511697014768,
"loss": 1.985142707824707,
"step": 34230
},
{
"epoch": 4.382464,
"grad_norm": 0.5818638205528259,
"learning_rate": 0.00017959189098409406,
"loss": 1.9964513778686523,
"step": 34240
},
{
"epoch": 4.383744,
"grad_norm": 0.6389639377593994,
"learning_rate": 0.00017953260745761908,
"loss": 1.9942167282104493,
"step": 34250
},
{
"epoch": 4.385024,
"grad_norm": 0.5499743819236755,
"learning_rate": 0.0001794733191316863,
"loss": 2.0012651443481446,
"step": 34260
},
{
"epoch": 4.386304,
"grad_norm": 0.6033515930175781,
"learning_rate": 0.00017941402601593084,
"loss": 1.9641998291015625,
"step": 34270
},
{
"epoch": 4.387584,
"grad_norm": 0.57430499792099,
"learning_rate": 0.00017935472811998873,
"loss": 1.97576847076416,
"step": 34280
},
{
"epoch": 4.388864,
"grad_norm": 0.5246350169181824,
"learning_rate": 0.00017929542545349662,
"loss": 1.9880905151367188,
"step": 34290
},
{
"epoch": 4.390144,
"grad_norm": 0.6816380620002747,
"learning_rate": 0.00017923611802609204,
"loss": 1.9691410064697266,
"step": 34300
},
{
"epoch": 4.391424,
"grad_norm": 0.5579296946525574,
"learning_rate": 0.00017917680584741326,
"loss": 1.9790719985961913,
"step": 34310
},
{
"epoch": 4.392704,
"grad_norm": 0.5415828227996826,
"learning_rate": 0.00017911748892709937,
"loss": 1.984635353088379,
"step": 34320
},
{
"epoch": 4.393984,
"grad_norm": 0.5486041903495789,
"learning_rate": 0.00017905816727479016,
"loss": 1.9851625442504883,
"step": 34330
},
{
"epoch": 4.395264,
"grad_norm": 0.5527475476264954,
"learning_rate": 0.0001789988409001262,
"loss": 1.985497283935547,
"step": 34340
},
{
"epoch": 4.3965440000000005,
"grad_norm": 0.5042336583137512,
"learning_rate": 0.00017893950981274888,
"loss": 1.994715690612793,
"step": 34350
},
{
"epoch": 4.397824,
"grad_norm": 0.610647439956665,
"learning_rate": 0.00017888017402230034,
"loss": 1.9797185897827148,
"step": 34360
},
{
"epoch": 4.399104,
"grad_norm": 0.5837764739990234,
"learning_rate": 0.00017882083353842342,
"loss": 1.9999412536621093,
"step": 34370
},
{
"epoch": 4.400384,
"grad_norm": 0.5599245429039001,
"learning_rate": 0.00017876148837076182,
"loss": 1.997800636291504,
"step": 34380
},
{
"epoch": 4.401664,
"grad_norm": 0.5592923164367676,
"learning_rate": 0.0001787021385289599,
"loss": 1.988368606567383,
"step": 34390
},
{
"epoch": 4.402944,
"grad_norm": 0.6156344413757324,
"learning_rate": 0.0001786427840226629,
"loss": 1.979899787902832,
"step": 34400
},
{
"epoch": 4.404224,
"grad_norm": 0.609311580657959,
"learning_rate": 0.00017858342486151666,
"loss": 1.9868654251098632,
"step": 34410
},
{
"epoch": 4.405504,
"grad_norm": 0.6127653121948242,
"learning_rate": 0.00017852406105516792,
"loss": 1.9891454696655273,
"step": 34420
},
{
"epoch": 4.406784,
"grad_norm": 0.6086615324020386,
"learning_rate": 0.00017846469261326408,
"loss": 1.9956405639648438,
"step": 34430
},
{
"epoch": 4.4080639999999995,
"grad_norm": 0.5747852921485901,
"learning_rate": 0.0001784053195454535,
"loss": 1.9969406127929688,
"step": 34440
},
{
"epoch": 4.409344,
"grad_norm": 0.553615391254425,
"learning_rate": 0.00017834594186138487,
"loss": 1.993647575378418,
"step": 34450
},
{
"epoch": 4.410624,
"grad_norm": 0.624441385269165,
"learning_rate": 0.00017828655957070804,
"loss": 1.9639902114868164,
"step": 34460
},
{
"epoch": 4.411904,
"grad_norm": 0.546384871006012,
"learning_rate": 0.00017822717268307342,
"loss": 1.9685798645019532,
"step": 34470
},
{
"epoch": 4.413184,
"grad_norm": 0.6608418226242065,
"learning_rate": 0.00017816778120813222,
"loss": 1.9905109405517578,
"step": 34480
},
{
"epoch": 4.414464,
"grad_norm": 0.5136147737503052,
"learning_rate": 0.00017810838515553638,
"loss": 1.990443801879883,
"step": 34490
},
{
"epoch": 4.415744,
"grad_norm": 0.5467607378959656,
"learning_rate": 0.0001780489845349386,
"loss": 1.9827947616577148,
"step": 34500
},
{
"epoch": 4.417024,
"grad_norm": 0.5772994756698608,
"learning_rate": 0.00017798957935599217,
"loss": 1.9813961029052733,
"step": 34510
},
{
"epoch": 4.418304,
"grad_norm": 0.4884471595287323,
"learning_rate": 0.00017793016962835143,
"loss": 1.9983732223510742,
"step": 34520
},
{
"epoch": 4.419584,
"grad_norm": 0.6583099365234375,
"learning_rate": 0.00017787075536167116,
"loss": 1.9775318145751952,
"step": 34530
},
{
"epoch": 4.420864,
"grad_norm": 0.5346192717552185,
"learning_rate": 0.00017781133656560709,
"loss": 1.9820104598999024,
"step": 34540
},
{
"epoch": 4.422144,
"grad_norm": 0.5661863684654236,
"learning_rate": 0.00017775191324981556,
"loss": 1.9964462280273438,
"step": 34550
},
{
"epoch": 4.423424,
"grad_norm": 0.5083594918251038,
"learning_rate": 0.00017769248542395366,
"loss": 1.9789955139160156,
"step": 34560
},
{
"epoch": 4.424704,
"grad_norm": 0.56032395362854,
"learning_rate": 0.0001776330530976793,
"loss": 1.9823909759521485,
"step": 34570
},
{
"epoch": 4.425984,
"grad_norm": 0.5630035996437073,
"learning_rate": 0.00017757361628065093,
"loss": 1.992217445373535,
"step": 34580
},
{
"epoch": 4.427264,
"grad_norm": 0.5994723439216614,
"learning_rate": 0.000177514174982528,
"loss": 1.983785057067871,
"step": 34590
},
{
"epoch": 4.428544,
"grad_norm": 0.5148761868476868,
"learning_rate": 0.00017745472921297043,
"loss": 1.9943147659301759,
"step": 34600
},
{
"epoch": 4.429824,
"grad_norm": 0.5327091217041016,
"learning_rate": 0.00017739527898163906,
"loss": 1.975392723083496,
"step": 34610
},
{
"epoch": 4.431104,
"grad_norm": 0.5408801436424255,
"learning_rate": 0.00017733582429819538,
"loss": 1.979996681213379,
"step": 34620
},
{
"epoch": 4.432384,
"grad_norm": 0.6325792670249939,
"learning_rate": 0.00017727636517230152,
"loss": 2.018503952026367,
"step": 34630
},
{
"epoch": 4.433664,
"grad_norm": 0.6835004091262817,
"learning_rate": 0.00017721690161362047,
"loss": 1.9957134246826171,
"step": 34640
},
{
"epoch": 4.434944,
"grad_norm": 0.5096126794815063,
"learning_rate": 0.0001771574336318159,
"loss": 1.984715461730957,
"step": 34650
},
{
"epoch": 4.436224,
"grad_norm": 0.5483854413032532,
"learning_rate": 0.0001770979612365521,
"loss": 1.9851861953735352,
"step": 34660
},
{
"epoch": 4.437504,
"grad_norm": 0.5634472966194153,
"learning_rate": 0.00017703848443749426,
"loss": 1.997150993347168,
"step": 34670
},
{
"epoch": 4.438784,
"grad_norm": 0.6631827354431152,
"learning_rate": 0.00017697900324430813,
"loss": 1.9835248947143556,
"step": 34680
},
{
"epoch": 4.440064,
"grad_norm": 0.5200762152671814,
"learning_rate": 0.00017691951766666018,
"loss": 1.9785295486450196,
"step": 34690
},
{
"epoch": 4.441344,
"grad_norm": 0.5406728386878967,
"learning_rate": 0.00017686002771421773,
"loss": 1.9862791061401368,
"step": 34700
},
{
"epoch": 4.442624,
"grad_norm": 0.5652700662612915,
"learning_rate": 0.00017680053339664867,
"loss": 1.9647504806518554,
"step": 34710
},
{
"epoch": 4.443904,
"grad_norm": 0.6459764838218689,
"learning_rate": 0.0001767410347236217,
"loss": 1.9751506805419923,
"step": 34720
},
{
"epoch": 4.445184,
"grad_norm": 0.5559481382369995,
"learning_rate": 0.00017668153170480617,
"loss": 2.0000701904296876,
"step": 34730
},
{
"epoch": 4.446464,
"grad_norm": 0.5489159226417542,
"learning_rate": 0.00017662202434987206,
"loss": 1.9957593917846679,
"step": 34740
},
{
"epoch": 4.447744,
"grad_norm": 0.5389276742935181,
"learning_rate": 0.0001765625126684902,
"loss": 1.9969293594360351,
"step": 34750
},
{
"epoch": 4.449024,
"grad_norm": 0.5074173212051392,
"learning_rate": 0.00017650299667033208,
"loss": 1.9883460998535156,
"step": 34760
},
{
"epoch": 4.450304,
"grad_norm": 0.5433664321899414,
"learning_rate": 0.00017644347636506982,
"loss": 1.971017074584961,
"step": 34770
},
{
"epoch": 4.451584,
"grad_norm": 0.557510256767273,
"learning_rate": 0.00017638395176237638,
"loss": 1.9819528579711914,
"step": 34780
},
{
"epoch": 4.452864,
"grad_norm": 0.5010910630226135,
"learning_rate": 0.00017632442287192525,
"loss": 1.9809906005859375,
"step": 34790
},
{
"epoch": 4.454144,
"grad_norm": 0.6046055555343628,
"learning_rate": 0.00017626488970339075,
"loss": 1.9743171691894532,
"step": 34800
},
{
"epoch": 4.455424,
"grad_norm": 0.6656924486160278,
"learning_rate": 0.0001762053522664478,
"loss": 1.9771373748779297,
"step": 34810
},
{
"epoch": 4.456704,
"grad_norm": 0.5329189896583557,
"learning_rate": 0.00017614581057077207,
"loss": 1.9876890182495117,
"step": 34820
},
{
"epoch": 4.457984,
"grad_norm": 0.49296873807907104,
"learning_rate": 0.00017608626462603996,
"loss": 1.9755582809448242,
"step": 34830
},
{
"epoch": 4.459264,
"grad_norm": 0.5341708660125732,
"learning_rate": 0.00017602671444192847,
"loss": 1.989487648010254,
"step": 34840
},
{
"epoch": 4.460544,
"grad_norm": 0.4903120994567871,
"learning_rate": 0.00017596716002811537,
"loss": 1.9795791625976562,
"step": 34850
},
{
"epoch": 4.461824,
"grad_norm": 0.5120388865470886,
"learning_rate": 0.000175907601394279,
"loss": 1.9796350479125977,
"step": 34860
},
{
"epoch": 4.463104,
"grad_norm": 0.523037314414978,
"learning_rate": 0.00017584803855009846,
"loss": 1.9906787872314453,
"step": 34870
},
{
"epoch": 4.464384,
"grad_norm": 0.5821148753166199,
"learning_rate": 0.00017578847150525367,
"loss": 1.9949838638305664,
"step": 34880
},
{
"epoch": 4.465664,
"grad_norm": 0.5770034193992615,
"learning_rate": 0.00017572890026942498,
"loss": 1.9867748260498046,
"step": 34890
},
{
"epoch": 4.466944,
"grad_norm": 0.5120547413825989,
"learning_rate": 0.00017566932485229354,
"loss": 1.9762218475341797,
"step": 34900
},
{
"epoch": 4.468224,
"grad_norm": 0.5931008458137512,
"learning_rate": 0.00017560974526354127,
"loss": 1.9820425033569335,
"step": 34910
},
{
"epoch": 4.469504,
"grad_norm": 0.5534963011741638,
"learning_rate": 0.00017555016151285055,
"loss": 1.998911476135254,
"step": 34920
},
{
"epoch": 4.470784,
"grad_norm": 0.571474015712738,
"learning_rate": 0.0001754905736099047,
"loss": 1.9791528701782226,
"step": 34930
},
{
"epoch": 4.472064,
"grad_norm": 0.6006047129631042,
"learning_rate": 0.00017543098156438752,
"loss": 1.9915401458740234,
"step": 34940
},
{
"epoch": 4.473344,
"grad_norm": 0.5826103091239929,
"learning_rate": 0.00017537138538598353,
"loss": 1.966348648071289,
"step": 34950
},
{
"epoch": 4.474624,
"grad_norm": 0.6054190397262573,
"learning_rate": 0.00017531178508437795,
"loss": 1.9963603973388673,
"step": 34960
},
{
"epoch": 4.475904,
"grad_norm": 0.5901658535003662,
"learning_rate": 0.0001752521806692566,
"loss": 1.973241424560547,
"step": 34970
},
{
"epoch": 4.477184,
"grad_norm": 0.5658732056617737,
"learning_rate": 0.0001751925721503061,
"loss": 1.9744939804077148,
"step": 34980
},
{
"epoch": 4.478464,
"grad_norm": 0.5967923998832703,
"learning_rate": 0.00017513295953721364,
"loss": 1.9915510177612306,
"step": 34990
},
{
"epoch": 4.479744,
"grad_norm": 0.5293209552764893,
"learning_rate": 0.00017507334283966707,
"loss": 1.977126121520996,
"step": 35000
},
{
"epoch": 4.481024,
"grad_norm": 0.5954993367195129,
"learning_rate": 0.00017501372206735503,
"loss": 1.974633026123047,
"step": 35010
},
{
"epoch": 4.482304,
"grad_norm": 0.5039696097373962,
"learning_rate": 0.00017495409722996654,
"loss": 1.9698060989379882,
"step": 35020
},
{
"epoch": 4.4835840000000005,
"grad_norm": 0.5870632529258728,
"learning_rate": 0.0001748944683371916,
"loss": 1.9890886306762696,
"step": 35030
},
{
"epoch": 4.484864,
"grad_norm": 0.609430193901062,
"learning_rate": 0.00017483483539872067,
"loss": 1.9668119430541993,
"step": 35040
},
{
"epoch": 4.486144,
"grad_norm": 0.5428464412689209,
"learning_rate": 0.00017477519842424495,
"loss": 1.9747596740722657,
"step": 35050
},
{
"epoch": 4.487424,
"grad_norm": 0.6130753755569458,
"learning_rate": 0.00017471555742345628,
"loss": 1.957578468322754,
"step": 35060
},
{
"epoch": 4.488704,
"grad_norm": 0.5869508385658264,
"learning_rate": 0.0001746559124060472,
"loss": 1.9638090133666992,
"step": 35070
},
{
"epoch": 4.489984,
"grad_norm": 0.57317054271698,
"learning_rate": 0.0001745962633817107,
"loss": 1.9847122192382813,
"step": 35080
},
{
"epoch": 4.491264,
"grad_norm": 0.5729394555091858,
"learning_rate": 0.00017453661036014075,
"loss": 1.9878597259521484,
"step": 35090
},
{
"epoch": 4.492544,
"grad_norm": 0.6074166893959045,
"learning_rate": 0.00017447695335103168,
"loss": 1.9688051223754883,
"step": 35100
},
{
"epoch": 4.493824,
"grad_norm": 0.5757483839988708,
"learning_rate": 0.00017441729236407862,
"loss": 1.9763423919677734,
"step": 35110
},
{
"epoch": 4.4951039999999995,
"grad_norm": 0.5711780190467834,
"learning_rate": 0.0001743576274089773,
"loss": 1.987339973449707,
"step": 35120
},
{
"epoch": 4.496384,
"grad_norm": 0.5370378494262695,
"learning_rate": 0.00017429795849542408,
"loss": 1.9679380416870118,
"step": 35130
},
{
"epoch": 4.497664,
"grad_norm": 0.5513519048690796,
"learning_rate": 0.00017423828563311602,
"loss": 1.9986581802368164,
"step": 35140
},
{
"epoch": 4.498944,
"grad_norm": 0.49918246269226074,
"learning_rate": 0.00017417860883175073,
"loss": 1.9722042083740234,
"step": 35150
},
{
"epoch": 4.500224,
"grad_norm": 0.5581119060516357,
"learning_rate": 0.0001741189281010266,
"loss": 1.9951316833496093,
"step": 35160
},
{
"epoch": 4.501504,
"grad_norm": 0.5419240593910217,
"learning_rate": 0.00017405924345064255,
"loss": 1.9568548202514648,
"step": 35170
},
{
"epoch": 4.502784,
"grad_norm": 0.5395339131355286,
"learning_rate": 0.0001739995548902981,
"loss": 1.9981374740600586,
"step": 35180
},
{
"epoch": 4.504064,
"grad_norm": 0.6799718737602234,
"learning_rate": 0.00017393986242969357,
"loss": 1.9841026306152343,
"step": 35190
},
{
"epoch": 4.505344,
"grad_norm": 0.5972551703453064,
"learning_rate": 0.00017388016607852968,
"loss": 1.9751358032226562,
"step": 35200
},
{
"epoch": 4.506624,
"grad_norm": 0.6718701124191284,
"learning_rate": 0.00017382046584650806,
"loss": 2.0015537261962892,
"step": 35210
},
{
"epoch": 4.507904,
"grad_norm": 0.5515751242637634,
"learning_rate": 0.0001737607617433307,
"loss": 1.987485694885254,
"step": 35220
},
{
"epoch": 4.509184,
"grad_norm": 0.6002400517463684,
"learning_rate": 0.00017370105377870047,
"loss": 1.986793327331543,
"step": 35230
},
{
"epoch": 4.510464,
"grad_norm": 0.5281581282615662,
"learning_rate": 0.00017364134196232062,
"loss": 1.9957992553710937,
"step": 35240
},
{
"epoch": 4.511744,
"grad_norm": 0.5505828857421875,
"learning_rate": 0.0001735816263038952,
"loss": 1.9851430892944335,
"step": 35250
},
{
"epoch": 4.513024,
"grad_norm": 0.6447466015815735,
"learning_rate": 0.0001735219068131288,
"loss": 1.985336685180664,
"step": 35260
},
{
"epoch": 4.514304,
"grad_norm": 0.6226052641868591,
"learning_rate": 0.00017346218349972677,
"loss": 1.9801296234130858,
"step": 35270
},
{
"epoch": 4.5155840000000005,
"grad_norm": 0.46576884388923645,
"learning_rate": 0.00017340245637339482,
"loss": 1.972369384765625,
"step": 35280
},
{
"epoch": 4.516864,
"grad_norm": 0.5133728384971619,
"learning_rate": 0.0001733427254438396,
"loss": 1.9791824340820312,
"step": 35290
},
{
"epoch": 4.518144,
"grad_norm": 0.57735675573349,
"learning_rate": 0.0001732829907207681,
"loss": 1.9968873977661132,
"step": 35300
},
{
"epoch": 4.519424,
"grad_norm": 0.576205849647522,
"learning_rate": 0.00017322325221388798,
"loss": 2.0014991760253906,
"step": 35310
},
{
"epoch": 4.520704,
"grad_norm": 0.5162070989608765,
"learning_rate": 0.00017316350993290774,
"loss": 2.005424690246582,
"step": 35320
},
{
"epoch": 4.521984,
"grad_norm": 0.7662791609764099,
"learning_rate": 0.00017310376388753626,
"loss": 1.986952018737793,
"step": 35330
},
{
"epoch": 4.523264,
"grad_norm": 0.5868723392486572,
"learning_rate": 0.00017304401408748304,
"loss": 1.9752798080444336,
"step": 35340
},
{
"epoch": 4.524544,
"grad_norm": 0.5373040437698364,
"learning_rate": 0.00017298426054245834,
"loss": 1.9720220565795898,
"step": 35350
},
{
"epoch": 4.525824,
"grad_norm": 0.7360947728157043,
"learning_rate": 0.00017292450326217293,
"loss": 1.9655765533447265,
"step": 35360
},
{
"epoch": 4.527104,
"grad_norm": 0.5355858206748962,
"learning_rate": 0.00017286474225633816,
"loss": 1.9618034362792969,
"step": 35370
},
{
"epoch": 4.528384,
"grad_norm": 0.5897169709205627,
"learning_rate": 0.00017280497753466593,
"loss": 1.9652278900146485,
"step": 35380
},
{
"epoch": 4.529664,
"grad_norm": 0.5661860108375549,
"learning_rate": 0.00017274520910686903,
"loss": 1.9850175857543946,
"step": 35390
},
{
"epoch": 4.530944,
"grad_norm": 0.5212974548339844,
"learning_rate": 0.00017268543698266054,
"loss": 1.9863178253173828,
"step": 35400
},
{
"epoch": 4.532224,
"grad_norm": 0.5268967747688293,
"learning_rate": 0.00017262566117175432,
"loss": 2.001449775695801,
"step": 35410
},
{
"epoch": 4.533504,
"grad_norm": 0.5728533864021301,
"learning_rate": 0.00017256588168386466,
"loss": 1.9800605773925781,
"step": 35420
},
{
"epoch": 4.534784,
"grad_norm": 0.5975796580314636,
"learning_rate": 0.00017250609852870663,
"loss": 1.9828550338745117,
"step": 35430
},
{
"epoch": 4.536064,
"grad_norm": 0.5381984710693359,
"learning_rate": 0.00017244631171599582,
"loss": 1.9694856643676757,
"step": 35440
},
{
"epoch": 4.537344,
"grad_norm": 0.5312972068786621,
"learning_rate": 0.00017238652125544843,
"loss": 1.971841812133789,
"step": 35450
},
{
"epoch": 4.538624,
"grad_norm": 0.47393307089805603,
"learning_rate": 0.00017232672715678118,
"loss": 1.9763164520263672,
"step": 35460
},
{
"epoch": 4.539904,
"grad_norm": 0.5315333604812622,
"learning_rate": 0.0001722669294297115,
"loss": 1.988376998901367,
"step": 35470
},
{
"epoch": 4.541184,
"grad_norm": 0.5470532774925232,
"learning_rate": 0.00017220712808395733,
"loss": 1.9837001800537108,
"step": 35480
},
{
"epoch": 4.542464,
"grad_norm": 0.5680336356163025,
"learning_rate": 0.00017214732312923717,
"loss": 1.996934700012207,
"step": 35490
},
{
"epoch": 4.543744,
"grad_norm": 0.6714974045753479,
"learning_rate": 0.00017208751457527022,
"loss": 1.993208122253418,
"step": 35500
},
{
"epoch": 4.545024,
"grad_norm": 0.5633406639099121,
"learning_rate": 0.00017202770243177615,
"loss": 1.9919679641723633,
"step": 35510
},
{
"epoch": 4.546304,
"grad_norm": 0.5346744060516357,
"learning_rate": 0.00017196788670847536,
"loss": 1.993514633178711,
"step": 35520
},
{
"epoch": 4.547584,
"grad_norm": 0.6055626273155212,
"learning_rate": 0.0001719080674150886,
"loss": 1.9646009445190429,
"step": 35530
},
{
"epoch": 4.548864,
"grad_norm": 0.5964628458023071,
"learning_rate": 0.00017184824456133738,
"loss": 1.9990467071533202,
"step": 35540
},
{
"epoch": 4.5501439999999995,
"grad_norm": 0.5390342473983765,
"learning_rate": 0.00017178841815694378,
"loss": 1.9856517791748047,
"step": 35550
},
{
"epoch": 4.551424,
"grad_norm": 0.5844201445579529,
"learning_rate": 0.00017172858821163037,
"loss": 1.980255699157715,
"step": 35560
},
{
"epoch": 4.552704,
"grad_norm": 0.5061194896697998,
"learning_rate": 0.00017166875473512039,
"loss": 1.9763916015625,
"step": 35570
},
{
"epoch": 4.553984,
"grad_norm": 0.4657714366912842,
"learning_rate": 0.0001716089177371376,
"loss": 1.9779897689819337,
"step": 35580
},
{
"epoch": 4.555264,
"grad_norm": 0.5359524488449097,
"learning_rate": 0.0001715490772274063,
"loss": 1.9688194274902344,
"step": 35590
},
{
"epoch": 4.556544,
"grad_norm": 0.6591652631759644,
"learning_rate": 0.00017148923321565147,
"loss": 2.003652572631836,
"step": 35600
},
{
"epoch": 4.557824,
"grad_norm": 0.5239455103874207,
"learning_rate": 0.00017142938571159855,
"loss": 1.9933404922485352,
"step": 35610
},
{
"epoch": 4.559104,
"grad_norm": 0.6762784719467163,
"learning_rate": 0.00017136953472497352,
"loss": 1.9978443145751954,
"step": 35620
},
{
"epoch": 4.560384,
"grad_norm": 0.5514674782752991,
"learning_rate": 0.00017130968026550313,
"loss": 2.007864761352539,
"step": 35630
},
{
"epoch": 4.561664,
"grad_norm": 0.5219026207923889,
"learning_rate": 0.00017124982234291447,
"loss": 1.9754510879516602,
"step": 35640
},
{
"epoch": 4.562944,
"grad_norm": 0.5592935085296631,
"learning_rate": 0.00017118996096693534,
"loss": 1.978312110900879,
"step": 35650
},
{
"epoch": 4.564224,
"grad_norm": 0.5252106785774231,
"learning_rate": 0.00017113009614729398,
"loss": 1.9817329406738282,
"step": 35660
},
{
"epoch": 4.565504,
"grad_norm": 0.5455541014671326,
"learning_rate": 0.00017107022789371925,
"loss": 1.9881767272949218,
"step": 35670
},
{
"epoch": 4.566784,
"grad_norm": 0.4987039268016815,
"learning_rate": 0.00017101035621594065,
"loss": 1.9727397918701173,
"step": 35680
},
{
"epoch": 4.568064,
"grad_norm": 0.5759603977203369,
"learning_rate": 0.0001709504811236881,
"loss": 2.0021541595458983,
"step": 35690
},
{
"epoch": 4.569344,
"grad_norm": 0.6112000942230225,
"learning_rate": 0.00017089060262669213,
"loss": 1.9869022369384766,
"step": 35700
},
{
"epoch": 4.5706240000000005,
"grad_norm": 0.6133731007575989,
"learning_rate": 0.0001708307207346838,
"loss": 1.9684524536132812,
"step": 35710
},
{
"epoch": 4.571904,
"grad_norm": 0.5849367380142212,
"learning_rate": 0.00017077083545739483,
"loss": 1.9587434768676757,
"step": 35720
},
{
"epoch": 4.573184,
"grad_norm": 0.586126983165741,
"learning_rate": 0.00017071094680455736,
"loss": 1.9820745468139649,
"step": 35730
},
{
"epoch": 4.574464,
"grad_norm": 0.5100215077400208,
"learning_rate": 0.00017065105478590413,
"loss": 1.9677181243896484,
"step": 35740
},
{
"epoch": 4.575744,
"grad_norm": 0.5384213328361511,
"learning_rate": 0.0001705911594111684,
"loss": 1.988356590270996,
"step": 35750
},
{
"epoch": 4.577024,
"grad_norm": 0.5295168161392212,
"learning_rate": 0.00017053126069008402,
"loss": 1.9765466690063476,
"step": 35760
},
{
"epoch": 4.578304,
"grad_norm": 0.6606968641281128,
"learning_rate": 0.00017047135863238532,
"loss": 1.9835369110107421,
"step": 35770
},
{
"epoch": 4.579584,
"grad_norm": 0.6336447596549988,
"learning_rate": 0.0001704114532478073,
"loss": 1.974114227294922,
"step": 35780
},
{
"epoch": 4.580864,
"grad_norm": 0.5335739850997925,
"learning_rate": 0.00017035154454608537,
"loss": 1.995610809326172,
"step": 35790
},
{
"epoch": 4.5821439999999996,
"grad_norm": 0.5434647798538208,
"learning_rate": 0.0001702916325369555,
"loss": 1.9903488159179688,
"step": 35800
},
{
"epoch": 4.583424,
"grad_norm": 0.6172047257423401,
"learning_rate": 0.00017023171723015427,
"loss": 1.975278663635254,
"step": 35810
},
{
"epoch": 4.584704,
"grad_norm": 0.5485216975212097,
"learning_rate": 0.00017017179863541867,
"loss": 1.9915719985961915,
"step": 35820
},
{
"epoch": 4.585984,
"grad_norm": 0.5575408935546875,
"learning_rate": 0.0001701118767624864,
"loss": 1.9615428924560547,
"step": 35830
},
{
"epoch": 4.587264,
"grad_norm": 0.5187435746192932,
"learning_rate": 0.0001700519516210955,
"loss": 1.9727203369140625,
"step": 35840
},
{
"epoch": 4.588544,
"grad_norm": 0.5441880822181702,
"learning_rate": 0.0001699920232209847,
"loss": 1.99354248046875,
"step": 35850
},
{
"epoch": 4.589824,
"grad_norm": 0.49303027987480164,
"learning_rate": 0.00016993209157189324,
"loss": 1.9770641326904297,
"step": 35860
},
{
"epoch": 4.591104,
"grad_norm": 0.566764771938324,
"learning_rate": 0.00016987215668356071,
"loss": 1.9841463088989257,
"step": 35870
},
{
"epoch": 4.592384,
"grad_norm": 0.5895103812217712,
"learning_rate": 0.00016981221856572743,
"loss": 1.9696409225463867,
"step": 35880
},
{
"epoch": 4.593664,
"grad_norm": 0.5896733403205872,
"learning_rate": 0.0001697522772281342,
"loss": 1.9897871017456055,
"step": 35890
},
{
"epoch": 4.594944,
"grad_norm": 0.6402928233146667,
"learning_rate": 0.00016969233268052225,
"loss": 1.9732433319091798,
"step": 35900
},
{
"epoch": 4.596224,
"grad_norm": 0.5241419076919556,
"learning_rate": 0.0001696323849326335,
"loss": 1.9905794143676758,
"step": 35910
},
{
"epoch": 4.597504,
"grad_norm": 0.5397542715072632,
"learning_rate": 0.00016957243399421022,
"loss": 1.966273880004883,
"step": 35920
},
{
"epoch": 4.598784,
"grad_norm": 0.5472021698951721,
"learning_rate": 0.00016951247987499526,
"loss": 1.9742664337158202,
"step": 35930
},
{
"epoch": 4.600064,
"grad_norm": 0.6202974915504456,
"learning_rate": 0.000169452522584732,
"loss": 1.9929149627685547,
"step": 35940
},
{
"epoch": 4.601344,
"grad_norm": 0.5677409768104553,
"learning_rate": 0.00016939256213316435,
"loss": 1.9837671279907227,
"step": 35950
},
{
"epoch": 4.6026240000000005,
"grad_norm": 0.5749878883361816,
"learning_rate": 0.00016933259853003674,
"loss": 1.9854482650756835,
"step": 35960
},
{
"epoch": 4.603904,
"grad_norm": 0.46651163697242737,
"learning_rate": 0.000169272631785094,
"loss": 1.971517562866211,
"step": 35970
},
{
"epoch": 4.605184,
"grad_norm": 0.5653470754623413,
"learning_rate": 0.0001692126619080817,
"loss": 1.9750205993652343,
"step": 35980
},
{
"epoch": 4.606464,
"grad_norm": 0.6055237054824829,
"learning_rate": 0.0001691526889087456,
"loss": 1.97509708404541,
"step": 35990
},
{
"epoch": 4.607744,
"grad_norm": 0.6083812117576599,
"learning_rate": 0.00016909271279683225,
"loss": 1.9845268249511718,
"step": 36000
},
{
"epoch": 4.609024,
"grad_norm": 0.528068482875824,
"learning_rate": 0.0001690327335820886,
"loss": 1.988920021057129,
"step": 36010
},
{
"epoch": 4.610304,
"grad_norm": 0.5368921160697937,
"learning_rate": 0.00016897275127426207,
"loss": 1.9759828567504882,
"step": 36020
},
{
"epoch": 4.611584,
"grad_norm": 0.49787724018096924,
"learning_rate": 0.00016891276588310066,
"loss": 1.9938508987426757,
"step": 36030
},
{
"epoch": 4.612864,
"grad_norm": 0.5205381512641907,
"learning_rate": 0.00016885277741835274,
"loss": 1.9796434402465821,
"step": 36040
},
{
"epoch": 4.614144,
"grad_norm": 0.5507999062538147,
"learning_rate": 0.00016879278588976732,
"loss": 1.9763618469238282,
"step": 36050
},
{
"epoch": 4.615424,
"grad_norm": 0.538848876953125,
"learning_rate": 0.00016873279130709393,
"loss": 1.9772726058959962,
"step": 36060
},
{
"epoch": 4.616704,
"grad_norm": 0.5718254446983337,
"learning_rate": 0.0001686727936800824,
"loss": 2.013087272644043,
"step": 36070
},
{
"epoch": 4.617984,
"grad_norm": 0.45794951915740967,
"learning_rate": 0.0001686127930184832,
"loss": 1.9984432220458985,
"step": 36080
},
{
"epoch": 4.619264,
"grad_norm": 0.6059564352035522,
"learning_rate": 0.00016855278933204735,
"loss": 1.9926963806152345,
"step": 36090
},
{
"epoch": 4.620544,
"grad_norm": 0.5310030579566956,
"learning_rate": 0.00016849278263052615,
"loss": 1.9861013412475585,
"step": 36100
},
{
"epoch": 4.621824,
"grad_norm": 0.5988901853561401,
"learning_rate": 0.00016843277292367164,
"loss": 1.9702316284179688,
"step": 36110
},
{
"epoch": 4.623104,
"grad_norm": 0.5477018356323242,
"learning_rate": 0.00016837276022123616,
"loss": 1.988865280151367,
"step": 36120
},
{
"epoch": 4.624384,
"grad_norm": 0.5558459758758545,
"learning_rate": 0.00016831274453297262,
"loss": 1.9629695892333985,
"step": 36130
},
{
"epoch": 4.625664,
"grad_norm": 0.5527434349060059,
"learning_rate": 0.0001682527258686344,
"loss": 2.0021005630493165,
"step": 36140
},
{
"epoch": 4.626944,
"grad_norm": 0.49802812933921814,
"learning_rate": 0.00016819270423797545,
"loss": 1.9708026885986327,
"step": 36150
},
{
"epoch": 4.628224,
"grad_norm": 0.5235414505004883,
"learning_rate": 0.00016813267965074997,
"loss": 1.9872283935546875,
"step": 36160
},
{
"epoch": 4.629504,
"grad_norm": 0.587867021560669,
"learning_rate": 0.00016807265211671286,
"loss": 1.9642206192016602,
"step": 36170
},
{
"epoch": 4.630784,
"grad_norm": 0.6100273728370667,
"learning_rate": 0.00016801262164561938,
"loss": 1.965578269958496,
"step": 36180
},
{
"epoch": 4.632064,
"grad_norm": 0.5849409699440002,
"learning_rate": 0.00016795258824722545,
"loss": 1.9760587692260743,
"step": 36190
},
{
"epoch": 4.633344,
"grad_norm": 0.5430747270584106,
"learning_rate": 0.00016789255193128716,
"loss": 1.9808311462402344,
"step": 36200
},
{
"epoch": 4.634624,
"grad_norm": 0.5065959692001343,
"learning_rate": 0.00016783251270756139,
"loss": 1.9835012435913086,
"step": 36210
},
{
"epoch": 4.635904,
"grad_norm": 0.5681312680244446,
"learning_rate": 0.0001677724705858052,
"loss": 1.9900894165039062,
"step": 36220
},
{
"epoch": 4.6371839999999995,
"grad_norm": 0.5940834879875183,
"learning_rate": 0.00016771242557577634,
"loss": 1.99401912689209,
"step": 36230
},
{
"epoch": 4.638464,
"grad_norm": 0.5626266598701477,
"learning_rate": 0.00016765237768723298,
"loss": 1.9767919540405274,
"step": 36240
},
{
"epoch": 4.639744,
"grad_norm": 0.5788188576698303,
"learning_rate": 0.0001675923269299337,
"loss": 1.9973836898803712,
"step": 36250
},
{
"epoch": 4.641024,
"grad_norm": 0.6053589582443237,
"learning_rate": 0.00016753227331363759,
"loss": 1.9949832916259767,
"step": 36260
},
{
"epoch": 4.642304,
"grad_norm": 0.5874267220497131,
"learning_rate": 0.00016747221684810417,
"loss": 1.9856956481933594,
"step": 36270
},
{
"epoch": 4.643584,
"grad_norm": 0.5451436042785645,
"learning_rate": 0.00016741215754309348,
"loss": 1.9839361190795899,
"step": 36280
},
{
"epoch": 4.644864,
"grad_norm": 0.5083765983581543,
"learning_rate": 0.000167352095408366,
"loss": 1.9861085891723633,
"step": 36290
},
{
"epoch": 4.646144,
"grad_norm": 0.5157638192176819,
"learning_rate": 0.0001672920304536826,
"loss": 1.9568124771118165,
"step": 36300
},
{
"epoch": 4.647424,
"grad_norm": 0.591122567653656,
"learning_rate": 0.00016723196268880473,
"loss": 1.9843204498291016,
"step": 36310
},
{
"epoch": 4.648704,
"grad_norm": 0.5233339667320251,
"learning_rate": 0.00016717189212349422,
"loss": 1.9786342620849608,
"step": 36320
},
{
"epoch": 4.649984,
"grad_norm": 0.5225774645805359,
"learning_rate": 0.00016711181876751335,
"loss": 2.001178741455078,
"step": 36330
},
{
"epoch": 4.651264,
"grad_norm": 0.5435526967048645,
"learning_rate": 0.00016705174263062484,
"loss": 1.9760160446166992,
"step": 36340
},
{
"epoch": 4.652544,
"grad_norm": 0.5470631122589111,
"learning_rate": 0.00016699166372259198,
"loss": 1.994298553466797,
"step": 36350
},
{
"epoch": 4.653824,
"grad_norm": 0.6677800416946411,
"learning_rate": 0.00016693158205317835,
"loss": 1.9685657501220704,
"step": 36360
},
{
"epoch": 4.655104,
"grad_norm": 0.6090927720069885,
"learning_rate": 0.00016687149763214813,
"loss": 1.9798606872558593,
"step": 36370
},
{
"epoch": 4.656384,
"grad_norm": 0.5031696557998657,
"learning_rate": 0.0001668114104692658,
"loss": 1.9893949508666993,
"step": 36380
},
{
"epoch": 4.6576640000000005,
"grad_norm": 0.521969735622406,
"learning_rate": 0.00016675132057429636,
"loss": 1.9835393905639649,
"step": 36390
},
{
"epoch": 4.658944,
"grad_norm": 0.6254401803016663,
"learning_rate": 0.0001666912279570053,
"loss": 2.0096778869628906,
"step": 36400
},
{
"epoch": 4.660224,
"grad_norm": 0.5933216214179993,
"learning_rate": 0.00016663113262715843,
"loss": 1.9984706878662108,
"step": 36410
},
{
"epoch": 4.661504,
"grad_norm": 0.5622230172157288,
"learning_rate": 0.00016657103459452214,
"loss": 1.9857000350952148,
"step": 36420
},
{
"epoch": 4.662784,
"grad_norm": 0.6188202500343323,
"learning_rate": 0.0001665109338688632,
"loss": 1.9888565063476562,
"step": 36430
},
{
"epoch": 4.664064,
"grad_norm": 0.5308187007904053,
"learning_rate": 0.00016645083045994875,
"loss": 1.9786457061767577,
"step": 36440
},
{
"epoch": 4.665344,
"grad_norm": 0.5242956876754761,
"learning_rate": 0.00016639072437754646,
"loss": 1.9862606048583984,
"step": 36450
},
{
"epoch": 4.666624,
"grad_norm": 0.5086528062820435,
"learning_rate": 0.00016633061563142438,
"loss": 1.984237289428711,
"step": 36460
},
{
"epoch": 4.667904,
"grad_norm": 0.5671142339706421,
"learning_rate": 0.00016627050423135103,
"loss": 1.9829334259033202,
"step": 36470
},
{
"epoch": 4.669184,
"grad_norm": 0.5350069999694824,
"learning_rate": 0.0001662103901870954,
"loss": 1.9884746551513672,
"step": 36480
},
{
"epoch": 4.670464,
"grad_norm": 0.5590651035308838,
"learning_rate": 0.0001661502735084268,
"loss": 1.9943134307861328,
"step": 36490
},
{
"epoch": 4.671744,
"grad_norm": 0.5233832597732544,
"learning_rate": 0.00016609015420511496,
"loss": 1.9843509674072266,
"step": 36500
},
{
"epoch": 4.673024,
"grad_norm": 0.5429251194000244,
"learning_rate": 0.00016603003228693022,
"loss": 1.9855207443237304,
"step": 36510
},
{
"epoch": 4.674304,
"grad_norm": 0.5338137149810791,
"learning_rate": 0.0001659699077636431,
"loss": 1.9927095413208007,
"step": 36520
},
{
"epoch": 4.675584,
"grad_norm": 0.538873016834259,
"learning_rate": 0.0001659097806450248,
"loss": 1.984290313720703,
"step": 36530
},
{
"epoch": 4.676864,
"grad_norm": 0.5624364018440247,
"learning_rate": 0.00016584965094084675,
"loss": 1.966472053527832,
"step": 36540
},
{
"epoch": 4.678144,
"grad_norm": 0.5536797046661377,
"learning_rate": 0.00016578951866088085,
"loss": 1.9732385635375977,
"step": 36550
},
{
"epoch": 4.679424,
"grad_norm": 0.5085057020187378,
"learning_rate": 0.00016572938381489942,
"loss": 1.9967426300048827,
"step": 36560
},
{
"epoch": 4.680704,
"grad_norm": 0.531596302986145,
"learning_rate": 0.0001656692464126752,
"loss": 1.9925098419189453,
"step": 36570
},
{
"epoch": 4.681984,
"grad_norm": 0.5825409293174744,
"learning_rate": 0.0001656091064639814,
"loss": 1.9805885314941407,
"step": 36580
},
{
"epoch": 4.683264,
"grad_norm": 0.49799415469169617,
"learning_rate": 0.00016554896397859157,
"loss": 1.9813636779785155,
"step": 36590
},
{
"epoch": 4.684544,
"grad_norm": 0.5369275212287903,
"learning_rate": 0.00016548881896627968,
"loss": 1.9751935958862306,
"step": 36600
},
{
"epoch": 4.685824,
"grad_norm": 0.5287046432495117,
"learning_rate": 0.0001654286714368202,
"loss": 1.9704582214355468,
"step": 36610
},
{
"epoch": 4.687104,
"grad_norm": 0.539780855178833,
"learning_rate": 0.00016536852139998777,
"loss": 2.0098703384399412,
"step": 36620
},
{
"epoch": 4.688384,
"grad_norm": 0.5841095447540283,
"learning_rate": 0.00016530836886555778,
"loss": 1.981536865234375,
"step": 36630
},
{
"epoch": 4.6896640000000005,
"grad_norm": 0.6065785884857178,
"learning_rate": 0.00016524821384330578,
"loss": 1.9860206604003907,
"step": 36640
},
{
"epoch": 4.690944,
"grad_norm": 0.5410844087600708,
"learning_rate": 0.00016518805634300777,
"loss": 2.0117197036743164,
"step": 36650
},
{
"epoch": 4.6922239999999995,
"grad_norm": 0.5077944993972778,
"learning_rate": 0.00016512789637444024,
"loss": 1.9990810394287108,
"step": 36660
},
{
"epoch": 4.693504,
"grad_norm": 0.5866582989692688,
"learning_rate": 0.00016506773394737995,
"loss": 1.9937889099121093,
"step": 36670
},
{
"epoch": 4.694784,
"grad_norm": 0.6490481495857239,
"learning_rate": 0.00016500756907160419,
"loss": 1.9979381561279297,
"step": 36680
},
{
"epoch": 4.696064,
"grad_norm": 0.6414040327072144,
"learning_rate": 0.00016494740175689054,
"loss": 1.9867414474487304,
"step": 36690
},
{
"epoch": 4.697344,
"grad_norm": 0.6493069529533386,
"learning_rate": 0.000164887232013017,
"loss": 1.9763008117675782,
"step": 36700
},
{
"epoch": 4.698624,
"grad_norm": 0.5509941577911377,
"learning_rate": 0.00016482705984976207,
"loss": 1.9965223312377929,
"step": 36710
},
{
"epoch": 4.699904,
"grad_norm": 0.5187804698944092,
"learning_rate": 0.00016476688527690452,
"loss": 1.9856864929199218,
"step": 36720
},
{
"epoch": 4.701184,
"grad_norm": 0.5766897797584534,
"learning_rate": 0.00016470670830422355,
"loss": 1.96602783203125,
"step": 36730
},
{
"epoch": 4.702464,
"grad_norm": 0.6088153719902039,
"learning_rate": 0.00016464652894149874,
"loss": 1.9885231018066407,
"step": 36740
},
{
"epoch": 4.703744,
"grad_norm": 0.592566967010498,
"learning_rate": 0.00016458634719851003,
"loss": 1.9537927627563476,
"step": 36750
},
{
"epoch": 4.705024,
"grad_norm": 0.5999495387077332,
"learning_rate": 0.0001645261630850379,
"loss": 1.9742780685424806,
"step": 36760
},
{
"epoch": 4.706304,
"grad_norm": 0.5415589809417725,
"learning_rate": 0.000164465976610863,
"loss": 1.996878433227539,
"step": 36770
},
{
"epoch": 4.707584,
"grad_norm": 0.5936171412467957,
"learning_rate": 0.0001644057877857665,
"loss": 1.9897672653198242,
"step": 36780
},
{
"epoch": 4.708864,
"grad_norm": 0.5302193760871887,
"learning_rate": 0.00016434559661953,
"loss": 1.986495590209961,
"step": 36790
},
{
"epoch": 4.710144,
"grad_norm": 0.5972662568092346,
"learning_rate": 0.0001642854031219352,
"loss": 1.9666475296020507,
"step": 36800
},
{
"epoch": 4.711424,
"grad_norm": 0.5793780088424683,
"learning_rate": 0.00016422520730276457,
"loss": 1.9789392471313476,
"step": 36810
},
{
"epoch": 4.7127040000000004,
"grad_norm": 0.5495673418045044,
"learning_rate": 0.0001641650091718007,
"loss": 1.9920309066772461,
"step": 36820
},
{
"epoch": 4.713984,
"grad_norm": 0.599382758140564,
"learning_rate": 0.0001641048087388266,
"loss": 1.973811149597168,
"step": 36830
},
{
"epoch": 4.715264,
"grad_norm": 0.6737948060035706,
"learning_rate": 0.0001640446060136257,
"loss": 1.9743953704833985,
"step": 36840
},
{
"epoch": 4.716544,
"grad_norm": 0.5568981170654297,
"learning_rate": 0.0001639844010059817,
"loss": 1.9893709182739259,
"step": 36850
},
{
"epoch": 4.717824,
"grad_norm": 0.558271050453186,
"learning_rate": 0.00016392419372567883,
"loss": 1.98876953125,
"step": 36860
},
{
"epoch": 4.719104,
"grad_norm": 0.532891035079956,
"learning_rate": 0.0001638639841825016,
"loss": 1.9814987182617188,
"step": 36870
},
{
"epoch": 4.720384,
"grad_norm": 0.5633194446563721,
"learning_rate": 0.00016380377238623493,
"loss": 1.9800920486450195,
"step": 36880
},
{
"epoch": 4.721664,
"grad_norm": 0.5881326794624329,
"learning_rate": 0.00016374355834666396,
"loss": 1.9701086044311524,
"step": 36890
},
{
"epoch": 4.722944,
"grad_norm": 0.548471212387085,
"learning_rate": 0.00016368334207357435,
"loss": 1.9844608306884766,
"step": 36900
},
{
"epoch": 4.7242239999999995,
"grad_norm": 0.5658273696899414,
"learning_rate": 0.00016362312357675216,
"loss": 2.0059104919433595,
"step": 36910
},
{
"epoch": 4.725504,
"grad_norm": 0.47138580679893494,
"learning_rate": 0.00016356290286598365,
"loss": 1.9856452941894531,
"step": 36920
},
{
"epoch": 4.726784,
"grad_norm": 0.624931275844574,
"learning_rate": 0.0001635026799510555,
"loss": 1.9721492767333983,
"step": 36930
},
{
"epoch": 4.728064,
"grad_norm": 0.48462560772895813,
"learning_rate": 0.00016344245484175492,
"loss": 1.9780189514160156,
"step": 36940
},
{
"epoch": 4.729344,
"grad_norm": 0.6067259311676025,
"learning_rate": 0.00016338222754786906,
"loss": 1.978681755065918,
"step": 36950
},
{
"epoch": 4.730624,
"grad_norm": 0.5201408863067627,
"learning_rate": 0.0001633219980791859,
"loss": 1.9788103103637695,
"step": 36960
},
{
"epoch": 4.731904,
"grad_norm": 0.49294692277908325,
"learning_rate": 0.00016326176644549355,
"loss": 1.9784858703613282,
"step": 36970
},
{
"epoch": 4.733184,
"grad_norm": 0.5918685793876648,
"learning_rate": 0.00016320153265658038,
"loss": 1.9847665786743165,
"step": 36980
},
{
"epoch": 4.734464,
"grad_norm": 0.5251066088676453,
"learning_rate": 0.0001631412967222353,
"loss": 1.9979801177978516,
"step": 36990
},
{
"epoch": 4.735744,
"grad_norm": 0.5682145357131958,
"learning_rate": 0.00016308105865224748,
"loss": 1.9739809036254883,
"step": 37000
},
{
"epoch": 4.737024,
"grad_norm": 0.5583365559577942,
"learning_rate": 0.00016302081845640638,
"loss": 1.9868297576904297,
"step": 37010
},
{
"epoch": 4.738304,
"grad_norm": 0.5032923817634583,
"learning_rate": 0.0001629605761445019,
"loss": 1.9972209930419922,
"step": 37020
},
{
"epoch": 4.739584,
"grad_norm": 0.5414678454399109,
"learning_rate": 0.0001629003317263242,
"loss": 1.9599113464355469,
"step": 37030
},
{
"epoch": 4.740864,
"grad_norm": 0.5711342692375183,
"learning_rate": 0.00016284008521166396,
"loss": 1.9849123001098632,
"step": 37040
},
{
"epoch": 4.742144,
"grad_norm": 0.5903234481811523,
"learning_rate": 0.000162779836610312,
"loss": 1.98837890625,
"step": 37050
},
{
"epoch": 4.743424,
"grad_norm": 0.5203123092651367,
"learning_rate": 0.00016271958593205954,
"loss": 1.982809829711914,
"step": 37060
},
{
"epoch": 4.7447040000000005,
"grad_norm": 0.5850188136100769,
"learning_rate": 0.00016265933318669818,
"loss": 2.0049657821655273,
"step": 37070
},
{
"epoch": 4.745984,
"grad_norm": 0.6288490891456604,
"learning_rate": 0.00016259907838401974,
"loss": 1.9800453186035156,
"step": 37080
},
{
"epoch": 4.747264,
"grad_norm": 0.5624448657035828,
"learning_rate": 0.0001625388215338166,
"loss": 1.9969280242919922,
"step": 37090
},
{
"epoch": 4.748544,
"grad_norm": 0.6016015410423279,
"learning_rate": 0.00016247856264588122,
"loss": 1.980293083190918,
"step": 37100
},
{
"epoch": 4.749824,
"grad_norm": 0.639859676361084,
"learning_rate": 0.0001624183017300066,
"loss": 1.9813600540161134,
"step": 37110
},
{
"epoch": 4.751104,
"grad_norm": 0.5475626587867737,
"learning_rate": 0.00016235803879598588,
"loss": 2.003792572021484,
"step": 37120
},
{
"epoch": 4.752384,
"grad_norm": 0.5734973549842834,
"learning_rate": 0.00016229777385361261,
"loss": 2.0032758712768555,
"step": 37130
},
{
"epoch": 4.753664,
"grad_norm": 0.5496430993080139,
"learning_rate": 0.00016223750691268078,
"loss": 1.9767940521240235,
"step": 37140
},
{
"epoch": 4.754944,
"grad_norm": 0.5611727833747864,
"learning_rate": 0.00016217723798298453,
"loss": 1.9608840942382812,
"step": 37150
},
{
"epoch": 4.756224,
"grad_norm": 0.5937264561653137,
"learning_rate": 0.00016211696707431846,
"loss": 1.9884986877441406,
"step": 37160
},
{
"epoch": 4.757504,
"grad_norm": 0.5230306386947632,
"learning_rate": 0.00016205669419647737,
"loss": 1.9889339447021483,
"step": 37170
},
{
"epoch": 4.758784,
"grad_norm": 0.6034893989562988,
"learning_rate": 0.0001619964193592564,
"loss": 1.9814685821533202,
"step": 37180
},
{
"epoch": 4.760064,
"grad_norm": 0.5799227356910706,
"learning_rate": 0.00016193614257245114,
"loss": 1.9682188034057617,
"step": 37190
},
{
"epoch": 4.761344,
"grad_norm": 0.5055465698242188,
"learning_rate": 0.00016187586384585736,
"loss": 1.989046287536621,
"step": 37200
},
{
"epoch": 4.762624,
"grad_norm": 0.5548417568206787,
"learning_rate": 0.00016181558318927115,
"loss": 1.9835445404052734,
"step": 37210
},
{
"epoch": 4.763904,
"grad_norm": 0.5431151390075684,
"learning_rate": 0.00016175530061248906,
"loss": 1.9816299438476563,
"step": 37220
},
{
"epoch": 4.765184,
"grad_norm": 0.6686404943466187,
"learning_rate": 0.00016169501612530774,
"loss": 1.9804349899291993,
"step": 37230
},
{
"epoch": 4.766464,
"grad_norm": 0.6064391732215881,
"learning_rate": 0.00016163472973752428,
"loss": 1.9833393096923828,
"step": 37240
},
{
"epoch": 4.767744,
"grad_norm": 0.6546295285224915,
"learning_rate": 0.00016157444145893605,
"loss": 1.9780990600585937,
"step": 37250
},
{
"epoch": 4.769024,
"grad_norm": 0.49512308835983276,
"learning_rate": 0.00016151415129934073,
"loss": 1.9701847076416015,
"step": 37260
},
{
"epoch": 4.770304,
"grad_norm": 0.6059166789054871,
"learning_rate": 0.00016145385926853636,
"loss": 1.977864646911621,
"step": 37270
},
{
"epoch": 4.771584,
"grad_norm": 0.5748432874679565,
"learning_rate": 0.00016139356537632116,
"loss": 2.0067512512207033,
"step": 37280
},
{
"epoch": 4.772864,
"grad_norm": 0.5810348987579346,
"learning_rate": 0.0001613332696324938,
"loss": 1.9819681167602539,
"step": 37290
},
{
"epoch": 4.774144,
"grad_norm": 0.6014289855957031,
"learning_rate": 0.0001612729720468531,
"loss": 1.990755271911621,
"step": 37300
},
{
"epoch": 4.775424,
"grad_norm": 0.641228437423706,
"learning_rate": 0.00016121267262919823,
"loss": 1.9814952850341796,
"step": 37310
},
{
"epoch": 4.776704,
"grad_norm": 0.6819351315498352,
"learning_rate": 0.0001611523713893288,
"loss": 1.9687055587768554,
"step": 37320
},
{
"epoch": 4.777984,
"grad_norm": 0.553289532661438,
"learning_rate": 0.0001610920683370445,
"loss": 1.9811298370361328,
"step": 37330
},
{
"epoch": 4.7792639999999995,
"grad_norm": 0.5484339594841003,
"learning_rate": 0.0001610317634821454,
"loss": 1.9742219924926758,
"step": 37340
},
{
"epoch": 4.780544,
"grad_norm": 0.5434049367904663,
"learning_rate": 0.00016097145683443198,
"loss": 1.9760320663452149,
"step": 37350
},
{
"epoch": 4.781824,
"grad_norm": 0.6029630899429321,
"learning_rate": 0.0001609111484037048,
"loss": 1.9954109191894531,
"step": 37360
},
{
"epoch": 4.783104,
"grad_norm": 0.5112810134887695,
"learning_rate": 0.0001608508381997648,
"loss": 1.9569334030151366,
"step": 37370
},
{
"epoch": 4.784384,
"grad_norm": 0.522830605506897,
"learning_rate": 0.00016079052623241334,
"loss": 1.9773149490356445,
"step": 37380
},
{
"epoch": 4.785664,
"grad_norm": 0.5461567044258118,
"learning_rate": 0.00016073021251145192,
"loss": 1.984024429321289,
"step": 37390
},
{
"epoch": 4.786944,
"grad_norm": 0.5329024195671082,
"learning_rate": 0.00016066989704668225,
"loss": 1.9789205551147462,
"step": 37400
},
{
"epoch": 4.788224,
"grad_norm": 0.5485737323760986,
"learning_rate": 0.00016060957984790654,
"loss": 1.9893648147583007,
"step": 37410
},
{
"epoch": 4.789504,
"grad_norm": 0.5688614249229431,
"learning_rate": 0.00016054926092492707,
"loss": 1.990738868713379,
"step": 37420
},
{
"epoch": 4.790784,
"grad_norm": 0.48053276538848877,
"learning_rate": 0.0001604889402875466,
"loss": 1.980750846862793,
"step": 37430
},
{
"epoch": 4.792064,
"grad_norm": 0.5579006671905518,
"learning_rate": 0.00016042861794556804,
"loss": 1.9915275573730469,
"step": 37440
},
{
"epoch": 4.793344,
"grad_norm": 0.5961301922798157,
"learning_rate": 0.0001603682939087946,
"loss": 1.9767297744750976,
"step": 37450
},
{
"epoch": 4.794624,
"grad_norm": 0.5394458770751953,
"learning_rate": 0.00016030796818702978,
"loss": 1.9915121078491211,
"step": 37460
},
{
"epoch": 4.795904,
"grad_norm": 0.5379886627197266,
"learning_rate": 0.00016024764079007728,
"loss": 2.0094104766845704,
"step": 37470
},
{
"epoch": 4.797184,
"grad_norm": 0.4974423348903656,
"learning_rate": 0.00016018731172774123,
"loss": 1.9804956436157226,
"step": 37480
},
{
"epoch": 4.798464,
"grad_norm": 0.6299752593040466,
"learning_rate": 0.00016012698100982588,
"loss": 1.9811914443969727,
"step": 37490
},
{
"epoch": 4.7997440000000005,
"grad_norm": 0.5416081547737122,
"learning_rate": 0.00016006664864613582,
"loss": 1.9797634124755858,
"step": 37500
},
{
"epoch": 4.801024,
"grad_norm": 0.5690661072731018,
"learning_rate": 0.00016000631464647597,
"loss": 1.9854196548461913,
"step": 37510
},
{
"epoch": 4.802304,
"grad_norm": 0.5845214128494263,
"learning_rate": 0.0001599459790206513,
"loss": 1.9897222518920898,
"step": 37520
},
{
"epoch": 4.803584,
"grad_norm": 0.5444427132606506,
"learning_rate": 0.00015988564177846728,
"loss": 1.989141845703125,
"step": 37530
},
{
"epoch": 4.804864,
"grad_norm": 0.6150582432746887,
"learning_rate": 0.00015982530292972953,
"loss": 2.0008062362670898,
"step": 37540
},
{
"epoch": 4.806144,
"grad_norm": 0.6454845666885376,
"learning_rate": 0.0001597649624842439,
"loss": 1.994460678100586,
"step": 37550
},
{
"epoch": 4.807424,
"grad_norm": 0.5306639075279236,
"learning_rate": 0.00015970462045181662,
"loss": 1.9890781402587892,
"step": 37560
},
{
"epoch": 4.808704,
"grad_norm": 0.5867375135421753,
"learning_rate": 0.00015964427684225414,
"loss": 2.0038778305053713,
"step": 37570
},
{
"epoch": 4.809984,
"grad_norm": 0.6516774892807007,
"learning_rate": 0.000159583931665363,
"loss": 1.9850542068481445,
"step": 37580
},
{
"epoch": 4.8112639999999995,
"grad_norm": 0.5468631982803345,
"learning_rate": 0.00015952358493095023,
"loss": 1.995686912536621,
"step": 37590
},
{
"epoch": 4.812544,
"grad_norm": 0.6014682650566101,
"learning_rate": 0.00015946323664882292,
"loss": 1.9921457290649414,
"step": 37600
},
{
"epoch": 4.813824,
"grad_norm": 0.5903257727622986,
"learning_rate": 0.00015940288682878863,
"loss": 1.991041374206543,
"step": 37610
},
{
"epoch": 4.815104,
"grad_norm": 0.5550824999809265,
"learning_rate": 0.00015934253548065494,
"loss": 2.008447265625,
"step": 37620
},
{
"epoch": 4.816384,
"grad_norm": 0.5849685668945312,
"learning_rate": 0.00015928218261422985,
"loss": 1.9968027114868163,
"step": 37630
},
{
"epoch": 4.817664,
"grad_norm": 0.5668292045593262,
"learning_rate": 0.00015922182823932147,
"loss": 1.9782878875732421,
"step": 37640
},
{
"epoch": 4.818944,
"grad_norm": 0.5129431486129761,
"learning_rate": 0.00015916147236573824,
"loss": 2.0009475708007813,
"step": 37650
},
{
"epoch": 4.820224,
"grad_norm": 0.6342671513557434,
"learning_rate": 0.00015910111500328885,
"loss": 1.9944002151489257,
"step": 37660
},
{
"epoch": 4.821504,
"grad_norm": 0.577076256275177,
"learning_rate": 0.0001590407561617822,
"loss": 1.9859691619873048,
"step": 37670
},
{
"epoch": 4.822784,
"grad_norm": 0.4880009591579437,
"learning_rate": 0.0001589803958510274,
"loss": 1.983834457397461,
"step": 37680
},
{
"epoch": 4.824064,
"grad_norm": 0.55623859167099,
"learning_rate": 0.00015892003408083392,
"loss": 1.9913875579833984,
"step": 37690
},
{
"epoch": 4.825344,
"grad_norm": 0.518153190612793,
"learning_rate": 0.0001588596708610113,
"loss": 1.977010154724121,
"step": 37700
},
{
"epoch": 4.826624,
"grad_norm": 0.524726390838623,
"learning_rate": 0.00015879930620136943,
"loss": 1.9696340560913086,
"step": 37710
},
{
"epoch": 4.827904,
"grad_norm": 0.5662818551063538,
"learning_rate": 0.0001587389401117184,
"loss": 1.967978286743164,
"step": 37720
},
{
"epoch": 4.829184,
"grad_norm": 0.5958290696144104,
"learning_rate": 0.00015867857260186858,
"loss": 1.9708053588867187,
"step": 37730
},
{
"epoch": 4.830464,
"grad_norm": 0.5853819251060486,
"learning_rate": 0.00015861820368163045,
"loss": 1.9750112533569335,
"step": 37740
},
{
"epoch": 4.8317440000000005,
"grad_norm": 0.5536706447601318,
"learning_rate": 0.00015855783336081485,
"loss": 1.9848987579345703,
"step": 37750
},
{
"epoch": 4.833024,
"grad_norm": 0.4776443839073181,
"learning_rate": 0.00015849746164923277,
"loss": 1.9738258361816405,
"step": 37760
},
{
"epoch": 4.834304,
"grad_norm": 0.6043407917022705,
"learning_rate": 0.00015843708855669544,
"loss": 1.9815210342407226,
"step": 37770
},
{
"epoch": 4.835584,
"grad_norm": 0.5316645503044128,
"learning_rate": 0.00015837671409301435,
"loss": 1.9783969879150392,
"step": 37780
},
{
"epoch": 4.836864,
"grad_norm": 0.5724020004272461,
"learning_rate": 0.0001583163382680012,
"loss": 1.9691783905029296,
"step": 37790
},
{
"epoch": 4.838144,
"grad_norm": 0.4933875799179077,
"learning_rate": 0.00015825596109146785,
"loss": 1.9673341751098632,
"step": 37800
},
{
"epoch": 4.839424,
"grad_norm": 0.5399433374404907,
"learning_rate": 0.00015819558257322648,
"loss": 1.9861661911010742,
"step": 37810
},
{
"epoch": 4.840704,
"grad_norm": 0.5197950005531311,
"learning_rate": 0.0001581352027230894,
"loss": 1.9830266952514648,
"step": 37820
},
{
"epoch": 4.841984,
"grad_norm": 0.5308870077133179,
"learning_rate": 0.00015807482155086916,
"loss": 1.9848098754882812,
"step": 37830
},
{
"epoch": 4.843264,
"grad_norm": 0.46351516246795654,
"learning_rate": 0.00015801443906637857,
"loss": 1.9891773223876954,
"step": 37840
},
{
"epoch": 4.844544,
"grad_norm": 0.5871397256851196,
"learning_rate": 0.00015795405527943064,
"loss": 2.001987838745117,
"step": 37850
},
{
"epoch": 4.845824,
"grad_norm": 0.5311822891235352,
"learning_rate": 0.00015789367019983848,
"loss": 1.973931121826172,
"step": 37860
},
{
"epoch": 4.847104,
"grad_norm": 0.5201824307441711,
"learning_rate": 0.0001578332838374156,
"loss": 1.9838836669921875,
"step": 37870
},
{
"epoch": 4.848384,
"grad_norm": 0.6097989678382874,
"learning_rate": 0.00015777289620197554,
"loss": 1.9921062469482422,
"step": 37880
},
{
"epoch": 4.849664,
"grad_norm": 0.52428138256073,
"learning_rate": 0.00015771250730333222,
"loss": 1.986372184753418,
"step": 37890
},
{
"epoch": 4.850944,
"grad_norm": 0.521881639957428,
"learning_rate": 0.0001576521171512996,
"loss": 2.004850959777832,
"step": 37900
},
{
"epoch": 4.852224,
"grad_norm": 0.5171951055526733,
"learning_rate": 0.00015759172575569196,
"loss": 2.0053630828857423,
"step": 37910
},
{
"epoch": 4.853504,
"grad_norm": 0.4849284589290619,
"learning_rate": 0.00015753133312632368,
"loss": 1.9598728179931642,
"step": 37920
},
{
"epoch": 4.854784,
"grad_norm": 0.6076167225837708,
"learning_rate": 0.00015747093927300944,
"loss": 1.9736194610595703,
"step": 37930
},
{
"epoch": 4.856064,
"grad_norm": 0.6175825595855713,
"learning_rate": 0.00015741054420556407,
"loss": 1.966608428955078,
"step": 37940
},
{
"epoch": 4.857344,
"grad_norm": 0.6035305857658386,
"learning_rate": 0.00015735014793380266,
"loss": 1.9779632568359375,
"step": 37950
},
{
"epoch": 4.858624,
"grad_norm": 0.5814414024353027,
"learning_rate": 0.00015728975046754034,
"loss": 1.9899335861206056,
"step": 37960
},
{
"epoch": 4.859904,
"grad_norm": 0.6161844730377197,
"learning_rate": 0.00015722935181659262,
"loss": 1.983595085144043,
"step": 37970
},
{
"epoch": 4.861184,
"grad_norm": 0.5764700174331665,
"learning_rate": 0.00015716895199077504,
"loss": 1.9826309204101562,
"step": 37980
},
{
"epoch": 4.862464,
"grad_norm": 0.6036893725395203,
"learning_rate": 0.00015710855099990346,
"loss": 1.9940835952758789,
"step": 37990
},
{
"epoch": 4.863744,
"grad_norm": 0.5411235690116882,
"learning_rate": 0.0001570481488537939,
"loss": 1.9843908309936524,
"step": 38000
},
{
"epoch": 4.865024,
"grad_norm": 0.6100060939788818,
"learning_rate": 0.00015698774556226246,
"loss": 2.000869369506836,
"step": 38010
},
{
"epoch": 4.8663039999999995,
"grad_norm": 0.5393717885017395,
"learning_rate": 0.0001569273411351256,
"loss": 1.9928974151611327,
"step": 38020
},
{
"epoch": 4.867584,
"grad_norm": 0.6006694436073303,
"learning_rate": 0.00015686693558219985,
"loss": 1.9808277130126952,
"step": 38030
},
{
"epoch": 4.868864,
"grad_norm": 0.5432482957839966,
"learning_rate": 0.0001568065289133019,
"loss": 1.9971471786499024,
"step": 38040
},
{
"epoch": 4.870144,
"grad_norm": 0.49338510632514954,
"learning_rate": 0.00015674612113824874,
"loss": 1.9802448272705078,
"step": 38050
},
{
"epoch": 4.871424,
"grad_norm": 0.6227922439575195,
"learning_rate": 0.00015668571226685741,
"loss": 1.9920333862304687,
"step": 38060
},
{
"epoch": 4.872704,
"grad_norm": 0.5397454500198364,
"learning_rate": 0.00015662530230894526,
"loss": 1.9835517883300782,
"step": 38070
},
{
"epoch": 4.873984,
"grad_norm": 0.592204213142395,
"learning_rate": 0.00015656489127432972,
"loss": 1.998987579345703,
"step": 38080
},
{
"epoch": 4.875264,
"grad_norm": 0.5539932250976562,
"learning_rate": 0.0001565044791728284,
"loss": 1.9835775375366211,
"step": 38090
},
{
"epoch": 4.876544,
"grad_norm": 0.6041468977928162,
"learning_rate": 0.00015644406601425905,
"loss": 2.0074649810791017,
"step": 38100
},
{
"epoch": 4.877824,
"grad_norm": 0.5886192321777344,
"learning_rate": 0.00015638365180843972,
"loss": 1.9860185623168944,
"step": 38110
},
{
"epoch": 4.879104,
"grad_norm": 0.7107323408126831,
"learning_rate": 0.00015632323656518854,
"loss": 1.971036148071289,
"step": 38120
},
{
"epoch": 4.880384,
"grad_norm": 0.5186880230903625,
"learning_rate": 0.00015626282029432384,
"loss": 1.9956499099731446,
"step": 38130
},
{
"epoch": 4.881664,
"grad_norm": 0.5344050526618958,
"learning_rate": 0.00015620240300566408,
"loss": 1.9743724822998048,
"step": 38140
},
{
"epoch": 4.882944,
"grad_norm": 0.5441474318504333,
"learning_rate": 0.00015614198470902794,
"loss": 1.9984556198120118,
"step": 38150
},
{
"epoch": 4.884224,
"grad_norm": 0.5330116152763367,
"learning_rate": 0.00015608156541423415,
"loss": 1.9779165267944336,
"step": 38160
},
{
"epoch": 4.885504,
"grad_norm": 0.6072548627853394,
"learning_rate": 0.00015602114513110177,
"loss": 1.9871341705322265,
"step": 38170
},
{
"epoch": 4.8867840000000005,
"grad_norm": 0.5603382587432861,
"learning_rate": 0.00015596072386944989,
"loss": 1.9955076217651366,
"step": 38180
},
{
"epoch": 4.888064,
"grad_norm": 0.5648460388183594,
"learning_rate": 0.00015590030163909778,
"loss": 1.9949302673339844,
"step": 38190
},
{
"epoch": 4.889344,
"grad_norm": 0.47358283400535583,
"learning_rate": 0.00015583987844986492,
"loss": 1.9905818939208983,
"step": 38200
},
{
"epoch": 4.890624,
"grad_norm": 0.545353353023529,
"learning_rate": 0.00015577945431157093,
"loss": 1.9904468536376954,
"step": 38210
},
{
"epoch": 4.891904,
"grad_norm": 0.5645648837089539,
"learning_rate": 0.00015571902923403554,
"loss": 1.9899795532226563,
"step": 38220
},
{
"epoch": 4.893184,
"grad_norm": 0.5779556632041931,
"learning_rate": 0.00015565860322707867,
"loss": 1.9769765853881835,
"step": 38230
},
{
"epoch": 4.894464,
"grad_norm": 0.5693933367729187,
"learning_rate": 0.00015559817630052042,
"loss": 1.9653324127197265,
"step": 38240
},
{
"epoch": 4.895744,
"grad_norm": 0.5564526319503784,
"learning_rate": 0.00015553774846418093,
"loss": 1.9976757049560547,
"step": 38250
},
{
"epoch": 4.897024,
"grad_norm": 0.5258886814117432,
"learning_rate": 0.00015547731972788062,
"loss": 1.979925537109375,
"step": 38260
},
{
"epoch": 4.8983039999999995,
"grad_norm": 0.5153602361679077,
"learning_rate": 0.00015541689010143995,
"loss": 1.9904891967773437,
"step": 38270
},
{
"epoch": 4.899584,
"grad_norm": 0.5984225869178772,
"learning_rate": 0.0001553564595946796,
"loss": 1.98701171875,
"step": 38280
},
{
"epoch": 4.900864,
"grad_norm": 0.6170544624328613,
"learning_rate": 0.00015529602821742037,
"loss": 1.9944042205810546,
"step": 38290
},
{
"epoch": 4.902144,
"grad_norm": 0.5214157700538635,
"learning_rate": 0.00015523559597948328,
"loss": 1.9978761672973633,
"step": 38300
},
{
"epoch": 4.903424,
"grad_norm": 0.5899656414985657,
"learning_rate": 0.0001551751628906892,
"loss": 1.9824264526367188,
"step": 38310
},
{
"epoch": 4.904704,
"grad_norm": 0.4780838191509247,
"learning_rate": 0.00015511472896085948,
"loss": 1.9836315155029296,
"step": 38320
},
{
"epoch": 4.905984,
"grad_norm": 0.5547693371772766,
"learning_rate": 0.00015505429419981545,
"loss": 1.9839319229125976,
"step": 38330
},
{
"epoch": 4.907264,
"grad_norm": 0.560765266418457,
"learning_rate": 0.0001549938586173786,
"loss": 1.9636600494384766,
"step": 38340
},
{
"epoch": 4.908544,
"grad_norm": 0.5573650598526001,
"learning_rate": 0.00015493342222337055,
"loss": 1.9816709518432618,
"step": 38350
},
{
"epoch": 4.909824,
"grad_norm": 0.5275610685348511,
"learning_rate": 0.0001548729850276131,
"loss": 1.9866239547729492,
"step": 38360
},
{
"epoch": 4.911104,
"grad_norm": 0.5318896770477295,
"learning_rate": 0.00015481254703992803,
"loss": 1.9919950485229492,
"step": 38370
},
{
"epoch": 4.912384,
"grad_norm": 0.5803588628768921,
"learning_rate": 0.0001547521082701374,
"loss": 1.9905529022216797,
"step": 38380
},
{
"epoch": 4.913664,
"grad_norm": 0.5411370992660522,
"learning_rate": 0.00015469166872806334,
"loss": 1.9780118942260743,
"step": 38390
},
{
"epoch": 4.914944,
"grad_norm": 0.5451182126998901,
"learning_rate": 0.00015463122842352808,
"loss": 1.9979238510131836,
"step": 38400
},
{
"epoch": 4.916224,
"grad_norm": 0.5661409497261047,
"learning_rate": 0.0001545707873663541,
"loss": 1.9755594253540039,
"step": 38410
},
{
"epoch": 4.917504,
"grad_norm": 0.7197934985160828,
"learning_rate": 0.00015451034556636384,
"loss": 1.9833791732788086,
"step": 38420
},
{
"epoch": 4.9187840000000005,
"grad_norm": 0.5379369258880615,
"learning_rate": 0.00015444990303337992,
"loss": 1.9861717224121094,
"step": 38430
},
{
"epoch": 4.920064,
"grad_norm": 0.5379326939582825,
"learning_rate": 0.00015438945977722516,
"loss": 1.9699920654296874,
"step": 38440
},
{
"epoch": 4.921344,
"grad_norm": 0.6528730392456055,
"learning_rate": 0.00015432901580772232,
"loss": 1.9866209030151367,
"step": 38450
},
{
"epoch": 4.922624,
"grad_norm": 0.5279343128204346,
"learning_rate": 0.00015426857113469452,
"loss": 1.9842660903930665,
"step": 38460
},
{
"epoch": 4.923904,
"grad_norm": 0.5071518421173096,
"learning_rate": 0.00015420812576796473,
"loss": 1.9946937561035156,
"step": 38470
},
{
"epoch": 4.925184,
"grad_norm": 0.5544239282608032,
"learning_rate": 0.0001541476797173562,
"loss": 1.9890066146850587,
"step": 38480
},
{
"epoch": 4.926464,
"grad_norm": 0.5938829779624939,
"learning_rate": 0.00015408723299269227,
"loss": 1.9813892364501953,
"step": 38490
},
{
"epoch": 4.927744,
"grad_norm": 0.5887882709503174,
"learning_rate": 0.00015402678560379632,
"loss": 1.9920442581176758,
"step": 38500
},
{
"epoch": 4.929024,
"grad_norm": 0.5088275074958801,
"learning_rate": 0.000153966337560492,
"loss": 1.9807819366455077,
"step": 38510
},
{
"epoch": 4.930304,
"grad_norm": 0.5767085552215576,
"learning_rate": 0.0001539058888726028,
"loss": 1.9811033248901366,
"step": 38520
},
{
"epoch": 4.931584,
"grad_norm": 0.571368396282196,
"learning_rate": 0.00015384543954995264,
"loss": 1.9918119430541992,
"step": 38530
},
{
"epoch": 4.932864,
"grad_norm": 0.5651844143867493,
"learning_rate": 0.0001537849896023652,
"loss": 2.000454139709473,
"step": 38540
},
{
"epoch": 4.934144,
"grad_norm": 0.5153692960739136,
"learning_rate": 0.00015372453903966454,
"loss": 1.9783634185791015,
"step": 38550
},
{
"epoch": 4.935424,
"grad_norm": 0.5616461038589478,
"learning_rate": 0.00015366408787167472,
"loss": 1.9830533981323242,
"step": 38560
},
{
"epoch": 4.936704,
"grad_norm": 0.6770117282867432,
"learning_rate": 0.00015360363610821985,
"loss": 1.967795181274414,
"step": 38570
},
{
"epoch": 4.937984,
"grad_norm": 0.5687888860702515,
"learning_rate": 0.0001535431837591242,
"loss": 1.966420555114746,
"step": 38580
},
{
"epoch": 4.939264,
"grad_norm": 0.5690419673919678,
"learning_rate": 0.0001534827308342121,
"loss": 1.993541145324707,
"step": 38590
},
{
"epoch": 4.940544,
"grad_norm": 0.5705564022064209,
"learning_rate": 0.00015342227734330794,
"loss": 2.0127492904663087,
"step": 38600
},
{
"epoch": 4.941824,
"grad_norm": 0.60486900806427,
"learning_rate": 0.00015336182329623633,
"loss": 1.9817153930664062,
"step": 38610
},
{
"epoch": 4.943104,
"grad_norm": 0.5235728621482849,
"learning_rate": 0.0001533013687028219,
"loss": 1.9966365814208984,
"step": 38620
},
{
"epoch": 4.944384,
"grad_norm": 0.5731178522109985,
"learning_rate": 0.00015324091357288926,
"loss": 1.9726491928100587,
"step": 38630
},
{
"epoch": 4.945664,
"grad_norm": 0.5211519598960876,
"learning_rate": 0.00015318045791626333,
"loss": 1.9730669021606446,
"step": 38640
},
{
"epoch": 4.946944,
"grad_norm": 0.5838186144828796,
"learning_rate": 0.0001531200017427689,
"loss": 1.9638963699340821,
"step": 38650
},
{
"epoch": 4.948224,
"grad_norm": 0.5275238156318665,
"learning_rate": 0.00015305954506223099,
"loss": 1.9751209259033202,
"step": 38660
},
{
"epoch": 4.949504,
"grad_norm": 0.5415754318237305,
"learning_rate": 0.00015299908788447462,
"loss": 1.9925893783569335,
"step": 38670
},
{
"epoch": 4.950784,
"grad_norm": 0.5102791786193848,
"learning_rate": 0.00015293863021932487,
"loss": 1.9648414611816407,
"step": 38680
},
{
"epoch": 4.952064,
"grad_norm": 0.6883862018585205,
"learning_rate": 0.00015287817207660704,
"loss": 1.9775564193725585,
"step": 38690
},
{
"epoch": 4.9533439999999995,
"grad_norm": 0.5129445791244507,
"learning_rate": 0.00015281771346614643,
"loss": 1.992006492614746,
"step": 38700
},
{
"epoch": 4.954624,
"grad_norm": 0.5979087352752686,
"learning_rate": 0.0001527572543977683,
"loss": 1.9978168487548829,
"step": 38710
},
{
"epoch": 4.955904,
"grad_norm": 0.5311994552612305,
"learning_rate": 0.00015269679488129812,
"loss": 1.9816572189331054,
"step": 38720
},
{
"epoch": 4.957184,
"grad_norm": 0.565992534160614,
"learning_rate": 0.00015263633492656139,
"loss": 1.9796485900878906,
"step": 38730
},
{
"epoch": 4.958464,
"grad_norm": 0.5734325647354126,
"learning_rate": 0.00015257587454338374,
"loss": 1.9745121002197266,
"step": 38740
},
{
"epoch": 4.959744,
"grad_norm": 0.5435142517089844,
"learning_rate": 0.0001525154137415908,
"loss": 1.9891155242919922,
"step": 38750
},
{
"epoch": 4.961024,
"grad_norm": 0.5707155466079712,
"learning_rate": 0.00015245495253100828,
"loss": 1.9773555755615235,
"step": 38760
},
{
"epoch": 4.962304,
"grad_norm": 0.5502570271492004,
"learning_rate": 0.000152394490921462,
"loss": 2.0045707702636717,
"step": 38770
},
{
"epoch": 4.963584,
"grad_norm": 0.6464360952377319,
"learning_rate": 0.0001523340289227777,
"loss": 1.9971561431884766,
"step": 38780
},
{
"epoch": 4.964864,
"grad_norm": 0.561187744140625,
"learning_rate": 0.0001522735665447814,
"loss": 1.9830976486206056,
"step": 38790
},
{
"epoch": 4.966144,
"grad_norm": 0.503157377243042,
"learning_rate": 0.00015221310379729904,
"loss": 1.9829216003417969,
"step": 38800
},
{
"epoch": 4.967424,
"grad_norm": 0.5248684883117676,
"learning_rate": 0.00015215264069015668,
"loss": 1.9835119247436523,
"step": 38810
},
{
"epoch": 4.968704,
"grad_norm": 0.5382105112075806,
"learning_rate": 0.00015209217723318035,
"loss": 1.979224967956543,
"step": 38820
},
{
"epoch": 4.969984,
"grad_norm": 0.5205886363983154,
"learning_rate": 0.00015203171343619625,
"loss": 1.9952116012573242,
"step": 38830
},
{
"epoch": 4.971264,
"grad_norm": 1.0143158435821533,
"learning_rate": 0.0001519712493090306,
"loss": 1.9701288223266602,
"step": 38840
},
{
"epoch": 4.972544,
"grad_norm": 0.5539652705192566,
"learning_rate": 0.0001519107848615096,
"loss": 1.9698051452636718,
"step": 38850
},
{
"epoch": 4.9738240000000005,
"grad_norm": 0.5053827166557312,
"learning_rate": 0.00015185032010345956,
"loss": 2.0002426147460937,
"step": 38860
},
{
"epoch": 4.975104,
"grad_norm": 0.4895821213722229,
"learning_rate": 0.00015178985504470696,
"loss": 1.9694843292236328,
"step": 38870
},
{
"epoch": 4.976384,
"grad_norm": 0.5244887471199036,
"learning_rate": 0.00015172938969507807,
"loss": 1.9792964935302735,
"step": 38880
},
{
"epoch": 4.977664,
"grad_norm": 0.5090240240097046,
"learning_rate": 0.00015166892406439937,
"loss": 1.9676008224487305,
"step": 38890
},
{
"epoch": 4.978944,
"grad_norm": 0.4834288954734802,
"learning_rate": 0.0001516084581624974,
"loss": 1.9918012619018555,
"step": 38900
},
{
"epoch": 4.980224,
"grad_norm": 0.6063424944877625,
"learning_rate": 0.0001515479919991987,
"loss": 1.9861129760742187,
"step": 38910
},
{
"epoch": 4.981504,
"grad_norm": 0.5674892663955688,
"learning_rate": 0.00015148752558432986,
"loss": 1.981055450439453,
"step": 38920
},
{
"epoch": 4.982784,
"grad_norm": 0.618547797203064,
"learning_rate": 0.00015142705892771754,
"loss": 1.9782644271850587,
"step": 38930
},
{
"epoch": 4.984064,
"grad_norm": 0.6015236973762512,
"learning_rate": 0.00015136659203918835,
"loss": 2.000592041015625,
"step": 38940
},
{
"epoch": 4.9853439999999996,
"grad_norm": 0.5674916505813599,
"learning_rate": 0.000151306124928569,
"loss": 2.0258649826049804,
"step": 38950
},
{
"epoch": 4.986624,
"grad_norm": 0.5386903882026672,
"learning_rate": 0.00015124565760568626,
"loss": 1.9819757461547851,
"step": 38960
},
{
"epoch": 4.987904,
"grad_norm": 0.5692916512489319,
"learning_rate": 0.00015118519008036694,
"loss": 1.9823396682739258,
"step": 38970
},
{
"epoch": 4.989184,
"grad_norm": 0.7221786975860596,
"learning_rate": 0.00015112472236243778,
"loss": 2.000724029541016,
"step": 38980
},
{
"epoch": 4.990464,
"grad_norm": 0.5891557335853577,
"learning_rate": 0.0001510642544617257,
"loss": 1.9757539749145507,
"step": 38990
},
{
"epoch": 4.991744,
"grad_norm": 0.5427893400192261,
"learning_rate": 0.00015100378638805746,
"loss": 1.9839298248291015,
"step": 39000
},
{
"epoch": 4.993024,
"grad_norm": 0.5004027485847473,
"learning_rate": 0.00015094331815126004,
"loss": 1.979203224182129,
"step": 39010
},
{
"epoch": 4.994304,
"grad_norm": 0.6043548583984375,
"learning_rate": 0.0001508828497611604,
"loss": 1.9880680084228515,
"step": 39020
},
{
"epoch": 4.995584,
"grad_norm": 0.4793684780597687,
"learning_rate": 0.00015082238122758542,
"loss": 1.9913110733032227,
"step": 39030
},
{
"epoch": 4.996864,
"grad_norm": 0.5623679757118225,
"learning_rate": 0.0001507619125603621,
"loss": 1.976661491394043,
"step": 39040
},
{
"epoch": 4.998144,
"grad_norm": 0.5763385891914368,
"learning_rate": 0.0001507014437693174,
"loss": 2.0018510818481445,
"step": 39050
},
{
"epoch": 4.999424,
"grad_norm": 0.5472203493118286,
"learning_rate": 0.00015064097486427842,
"loss": 1.9944067001342773,
"step": 39060
},
{
"epoch": 5.0,
"eval_loss": 1.9901835918426514,
"eval_runtime": 18.6149,
"eval_samples_per_second": 537.205,
"eval_steps_per_second": 67.151,
"step": 39065
},
{
"epoch": 5.00064,
"grad_norm": 0.6185895800590515,
"learning_rate": 0.00015058050585507212,
"loss": 1.9628395080566405,
"step": 39070
},
{
"epoch": 5.00192,
"grad_norm": 0.6103888154029846,
"learning_rate": 0.00015052003675152559,
"loss": 1.989004135131836,
"step": 39080
},
{
"epoch": 5.0032,
"grad_norm": 0.577446460723877,
"learning_rate": 0.00015045956756346585,
"loss": 1.9980476379394532,
"step": 39090
},
{
"epoch": 5.00448,
"grad_norm": 0.5932673215866089,
"learning_rate": 0.00015039909830072006,
"loss": 1.98687744140625,
"step": 39100
},
{
"epoch": 5.00576,
"grad_norm": 0.5585435628890991,
"learning_rate": 0.00015033862897311526,
"loss": 1.981728744506836,
"step": 39110
},
{
"epoch": 5.00704,
"grad_norm": 0.5685187578201294,
"learning_rate": 0.0001502781595904785,
"loss": 1.9845190048217773,
"step": 39120
},
{
"epoch": 5.00832,
"grad_norm": 0.5641683340072632,
"learning_rate": 0.00015021769016263702,
"loss": 1.972390365600586,
"step": 39130
},
{
"epoch": 5.0096,
"grad_norm": 0.5885679125785828,
"learning_rate": 0.00015015722069941783,
"loss": 1.9769290924072265,
"step": 39140
},
{
"epoch": 5.01088,
"grad_norm": 0.5615965127944946,
"learning_rate": 0.00015009675121064816,
"loss": 1.9809673309326172,
"step": 39150
},
{
"epoch": 5.01216,
"grad_norm": 0.5219846367835999,
"learning_rate": 0.00015003628170615502,
"loss": 1.972451400756836,
"step": 39160
},
{
"epoch": 5.01344,
"grad_norm": 0.5846970081329346,
"learning_rate": 0.0001499758121957656,
"loss": 1.9852188110351563,
"step": 39170
},
{
"epoch": 5.01472,
"grad_norm": 0.5330633521080017,
"learning_rate": 0.00014991534268930706,
"loss": 2.0026166915893553,
"step": 39180
},
{
"epoch": 5.016,
"grad_norm": 0.5415689945220947,
"learning_rate": 0.0001498548731966065,
"loss": 1.9913503646850585,
"step": 39190
},
{
"epoch": 5.01728,
"grad_norm": 0.6682746410369873,
"learning_rate": 0.00014979440372749105,
"loss": 1.9838417053222657,
"step": 39200
},
{
"epoch": 5.01856,
"grad_norm": 0.5136408805847168,
"learning_rate": 0.00014973393429178785,
"loss": 2.0018194198608397,
"step": 39210
},
{
"epoch": 5.01984,
"grad_norm": 0.5736105442047119,
"learning_rate": 0.00014967346489932405,
"loss": 1.9928335189819335,
"step": 39220
},
{
"epoch": 5.02112,
"grad_norm": 0.5319226980209351,
"learning_rate": 0.00014961299555992666,
"loss": 1.9917778015136718,
"step": 39230
},
{
"epoch": 5.0224,
"grad_norm": 0.5689748525619507,
"learning_rate": 0.00014955252628342285,
"loss": 1.975779914855957,
"step": 39240
},
{
"epoch": 5.02368,
"grad_norm": 0.5306477546691895,
"learning_rate": 0.0001494920570796398,
"loss": 1.9737434387207031,
"step": 39250
},
{
"epoch": 5.02496,
"grad_norm": 0.5307998061180115,
"learning_rate": 0.00014943158795840443,
"loss": 1.9994359970092774,
"step": 39260
},
{
"epoch": 5.02624,
"grad_norm": 0.5305598378181458,
"learning_rate": 0.00014937111892954392,
"loss": 1.98242244720459,
"step": 39270
},
{
"epoch": 5.02752,
"grad_norm": 0.554580569267273,
"learning_rate": 0.00014931065000288528,
"loss": 1.9741186141967773,
"step": 39280
},
{
"epoch": 5.0288,
"grad_norm": 0.5110889077186584,
"learning_rate": 0.0001492501811882556,
"loss": 2.019338607788086,
"step": 39290
},
{
"epoch": 5.03008,
"grad_norm": 0.540069043636322,
"learning_rate": 0.00014918971249548184,
"loss": 2.0003427505493163,
"step": 39300
},
{
"epoch": 5.03136,
"grad_norm": 0.6115997433662415,
"learning_rate": 0.00014912924393439104,
"loss": 1.965072250366211,
"step": 39310
},
{
"epoch": 5.03264,
"grad_norm": 1.5892024040222168,
"learning_rate": 0.00014906877551481014,
"loss": 2.0125198364257812,
"step": 39320
},
{
"epoch": 5.03392,
"grad_norm": 0.49498775601387024,
"learning_rate": 0.00014900830724656614,
"loss": 2.00412540435791,
"step": 39330
},
{
"epoch": 5.0352,
"grad_norm": 0.5577797889709473,
"learning_rate": 0.00014894783913948592,
"loss": 1.967217254638672,
"step": 39340
},
{
"epoch": 5.03648,
"grad_norm": 0.5913921594619751,
"learning_rate": 0.00014888737120339647,
"loss": 1.9749202728271484,
"step": 39350
},
{
"epoch": 5.03776,
"grad_norm": 0.5605696439743042,
"learning_rate": 0.00014882690344812454,
"loss": 1.981511116027832,
"step": 39360
},
{
"epoch": 5.03904,
"grad_norm": 0.6057311296463013,
"learning_rate": 0.00014876643588349707,
"loss": 1.991476058959961,
"step": 39370
},
{
"epoch": 5.04032,
"grad_norm": 0.553248405456543,
"learning_rate": 0.00014870596851934093,
"loss": 1.9909002304077148,
"step": 39380
},
{
"epoch": 5.0416,
"grad_norm": 0.5219967365264893,
"learning_rate": 0.00014864550136548278,
"loss": 1.9541400909423827,
"step": 39390
},
{
"epoch": 5.04288,
"grad_norm": 0.5165095925331116,
"learning_rate": 0.0001485850344317494,
"loss": 1.9773054122924805,
"step": 39400
},
{
"epoch": 5.04416,
"grad_norm": 0.5454107522964478,
"learning_rate": 0.0001485245677279675,
"loss": 1.9876811981201172,
"step": 39410
},
{
"epoch": 5.04544,
"grad_norm": 0.6373261213302612,
"learning_rate": 0.00014846410126396388,
"loss": 1.9674722671508789,
"step": 39420
},
{
"epoch": 5.04672,
"grad_norm": 0.5073217153549194,
"learning_rate": 0.000148403635049565,
"loss": 1.9724964141845702,
"step": 39430
},
{
"epoch": 5.048,
"grad_norm": 0.5515549182891846,
"learning_rate": 0.0001483431690945976,
"loss": 1.973788070678711,
"step": 39440
},
{
"epoch": 5.04928,
"grad_norm": 0.5144228935241699,
"learning_rate": 0.00014828270340888817,
"loss": 1.9946626663208007,
"step": 39450
},
{
"epoch": 5.05056,
"grad_norm": 0.5261775255203247,
"learning_rate": 0.00014822223800226318,
"loss": 1.9879295349121093,
"step": 39460
},
{
"epoch": 5.05184,
"grad_norm": 0.5014375448226929,
"learning_rate": 0.00014816177288454915,
"loss": 1.9792221069335938,
"step": 39470
},
{
"epoch": 5.05312,
"grad_norm": 0.5460830926895142,
"learning_rate": 0.00014810130806557255,
"loss": 1.9869150161743163,
"step": 39480
},
{
"epoch": 5.0544,
"grad_norm": 0.4884892404079437,
"learning_rate": 0.00014804084355515962,
"loss": 1.9885519027709961,
"step": 39490
},
{
"epoch": 5.05568,
"grad_norm": 0.6247113943099976,
"learning_rate": 0.00014798037936313677,
"loss": 1.9932462692260742,
"step": 39500
},
{
"epoch": 5.05696,
"grad_norm": 0.5195327997207642,
"learning_rate": 0.0001479199154993303,
"loss": 1.9726434707641602,
"step": 39510
},
{
"epoch": 5.05824,
"grad_norm": 0.5041347146034241,
"learning_rate": 0.00014785945197356632,
"loss": 1.989242172241211,
"step": 39520
},
{
"epoch": 5.05952,
"grad_norm": 0.6097322106361389,
"learning_rate": 0.00014779898879567103,
"loss": 1.9797655105590821,
"step": 39530
},
{
"epoch": 5.0608,
"grad_norm": 0.49437838792800903,
"learning_rate": 0.00014773852597547058,
"loss": 1.9952882766723632,
"step": 39540
},
{
"epoch": 5.06208,
"grad_norm": 0.5483019351959229,
"learning_rate": 0.000147678063522791,
"loss": 1.9886039733886718,
"step": 39550
},
{
"epoch": 5.06336,
"grad_norm": 0.5274131298065186,
"learning_rate": 0.00014761760144745828,
"loss": 1.9907428741455078,
"step": 39560
},
{
"epoch": 5.06464,
"grad_norm": 0.5565429329872131,
"learning_rate": 0.00014755713975929827,
"loss": 1.9705310821533204,
"step": 39570
},
{
"epoch": 5.06592,
"grad_norm": 0.5578149557113647,
"learning_rate": 0.00014749667846813695,
"loss": 1.979681396484375,
"step": 39580
},
{
"epoch": 5.0672,
"grad_norm": 0.5130541324615479,
"learning_rate": 0.00014743621758380002,
"loss": 1.9894338607788087,
"step": 39590
},
{
"epoch": 5.06848,
"grad_norm": 0.5107623934745789,
"learning_rate": 0.00014737575711611326,
"loss": 1.9762174606323242,
"step": 39600
},
{
"epoch": 5.06976,
"grad_norm": 0.6198186874389648,
"learning_rate": 0.00014731529707490235,
"loss": 1.9787363052368163,
"step": 39610
},
{
"epoch": 5.07104,
"grad_norm": 0.6227676868438721,
"learning_rate": 0.00014725483746999283,
"loss": 1.9761425018310548,
"step": 39620
},
{
"epoch": 5.07232,
"grad_norm": 0.5233384370803833,
"learning_rate": 0.00014719437831121023,
"loss": 1.992502784729004,
"step": 39630
},
{
"epoch": 5.0736,
"grad_norm": 0.6265090107917786,
"learning_rate": 0.00014713391960838,
"loss": 1.9837095260620117,
"step": 39640
},
{
"epoch": 5.07488,
"grad_norm": 0.5254048705101013,
"learning_rate": 0.00014707346137132762,
"loss": 1.9800655364990234,
"step": 39650
},
{
"epoch": 5.07616,
"grad_norm": 0.5523412227630615,
"learning_rate": 0.00014701300360987828,
"loss": 1.9917169570922852,
"step": 39660
},
{
"epoch": 5.07744,
"grad_norm": 3.8104248046875,
"learning_rate": 0.0001469525463338572,
"loss": 1.9873939514160157,
"step": 39670
},
{
"epoch": 5.07872,
"grad_norm": 0.5574650764465332,
"learning_rate": 0.0001468920895530896,
"loss": 1.9744285583496093,
"step": 39680
},
{
"epoch": 5.08,
"grad_norm": 0.529992401599884,
"learning_rate": 0.00014683163327740045,
"loss": 1.9931236267089845,
"step": 39690
},
{
"epoch": 5.08128,
"grad_norm": 0.48209846019744873,
"learning_rate": 0.0001467711775166148,
"loss": 1.9715225219726562,
"step": 39700
},
{
"epoch": 5.08256,
"grad_norm": 0.5406467914581299,
"learning_rate": 0.00014671072228055756,
"loss": 1.9805070877075195,
"step": 39710
},
{
"epoch": 5.08384,
"grad_norm": 0.6512395739555359,
"learning_rate": 0.0001466502675790535,
"loss": 1.9851547241210938,
"step": 39720
},
{
"epoch": 5.08512,
"grad_norm": 0.5435133576393127,
"learning_rate": 0.00014658981342192736,
"loss": 1.9780803680419923,
"step": 39730
},
{
"epoch": 5.0864,
"grad_norm": 0.5553458333015442,
"learning_rate": 0.0001465293598190038,
"loss": 1.9978094100952148,
"step": 39740
},
{
"epoch": 5.08768,
"grad_norm": 0.5138853788375854,
"learning_rate": 0.00014646890678010732,
"loss": 1.9848352432250977,
"step": 39750
},
{
"epoch": 5.08896,
"grad_norm": 0.5326972603797913,
"learning_rate": 0.00014640845431506238,
"loss": 1.9862096786499024,
"step": 39760
},
{
"epoch": 5.09024,
"grad_norm": 0.5226320624351501,
"learning_rate": 0.0001463480024336934,
"loss": 1.9748647689819336,
"step": 39770
},
{
"epoch": 5.09152,
"grad_norm": 0.5063529014587402,
"learning_rate": 0.00014628755114582466,
"loss": 1.9794319152832032,
"step": 39780
},
{
"epoch": 5.0928,
"grad_norm": 0.5227596163749695,
"learning_rate": 0.00014622710046128024,
"loss": 1.9889839172363282,
"step": 39790
},
{
"epoch": 5.09408,
"grad_norm": 0.5256738066673279,
"learning_rate": 0.00014616665038988424,
"loss": 1.9793212890625,
"step": 39800
},
{
"epoch": 5.09536,
"grad_norm": 0.5670853853225708,
"learning_rate": 0.00014610620094146072,
"loss": 1.9790582656860352,
"step": 39810
},
{
"epoch": 5.09664,
"grad_norm": 0.5596614480018616,
"learning_rate": 0.0001460457521258334,
"loss": 1.9879968643188477,
"step": 39820
},
{
"epoch": 5.09792,
"grad_norm": 0.5723010301589966,
"learning_rate": 0.0001459853039528262,
"loss": 1.9991182327270507,
"step": 39830
},
{
"epoch": 5.0992,
"grad_norm": 0.4890998601913452,
"learning_rate": 0.00014592485643226275,
"loss": 2.0084213256835937,
"step": 39840
},
{
"epoch": 5.10048,
"grad_norm": 0.5837517380714417,
"learning_rate": 0.0001458644095739665,
"loss": 1.9833196640014648,
"step": 39850
},
{
"epoch": 5.10176,
"grad_norm": 0.5749513506889343,
"learning_rate": 0.00014580396338776102,
"loss": 1.982977294921875,
"step": 39860
},
{
"epoch": 5.10304,
"grad_norm": 0.5612340569496155,
"learning_rate": 0.00014574351788346966,
"loss": 1.9853507995605468,
"step": 39870
},
{
"epoch": 5.10432,
"grad_norm": 0.517298698425293,
"learning_rate": 0.00014568307307091555,
"loss": 1.9917947769165039,
"step": 39880
},
{
"epoch": 5.1056,
"grad_norm": 0.5579150915145874,
"learning_rate": 0.0001456226289599219,
"loss": 1.9697492599487305,
"step": 39890
},
{
"epoch": 5.10688,
"grad_norm": 0.6024275422096252,
"learning_rate": 0.00014556218556031165,
"loss": 1.9792884826660155,
"step": 39900
},
{
"epoch": 5.10816,
"grad_norm": 0.5712671279907227,
"learning_rate": 0.0001455017428819078,
"loss": 1.9927783966064454,
"step": 39910
},
{
"epoch": 5.10944,
"grad_norm": 0.5119901895523071,
"learning_rate": 0.00014544130093453297,
"loss": 1.9695430755615235,
"step": 39920
},
{
"epoch": 5.11072,
"grad_norm": 0.578326404094696,
"learning_rate": 0.0001453808597280099,
"loss": 2.001946449279785,
"step": 39930
},
{
"epoch": 5.112,
"grad_norm": 0.5617350339889526,
"learning_rate": 0.00014532041927216115,
"loss": 1.997459602355957,
"step": 39940
},
{
"epoch": 5.11328,
"grad_norm": 0.6018834114074707,
"learning_rate": 0.00014525997957680906,
"loss": 1.9878200531005858,
"step": 39950
},
{
"epoch": 5.11456,
"grad_norm": 0.6765103936195374,
"learning_rate": 0.00014519954065177597,
"loss": 1.9897710800170898,
"step": 39960
},
{
"epoch": 5.11584,
"grad_norm": 0.5492919087409973,
"learning_rate": 0.00014513910250688405,
"loss": 1.986349105834961,
"step": 39970
},
{
"epoch": 5.11712,
"grad_norm": 0.5318031311035156,
"learning_rate": 0.00014507866515195525,
"loss": 1.993898582458496,
"step": 39980
},
{
"epoch": 5.1184,
"grad_norm": 0.5629369616508484,
"learning_rate": 0.00014501822859681152,
"loss": 1.9986658096313477,
"step": 39990
},
{
"epoch": 5.11968,
"grad_norm": 0.5713302493095398,
"learning_rate": 0.00014495779285127476,
"loss": 1.978622055053711,
"step": 40000
},
{
"epoch": 5.12096,
"grad_norm": 0.5584504008293152,
"learning_rate": 0.00014489735792516648,
"loss": 1.9740705490112305,
"step": 40010
},
{
"epoch": 5.12224,
"grad_norm": 0.5452061891555786,
"learning_rate": 0.00014483692382830817,
"loss": 1.9851701736450196,
"step": 40020
},
{
"epoch": 5.12352,
"grad_norm": 0.588334321975708,
"learning_rate": 0.00014477649057052127,
"loss": 1.9832292556762696,
"step": 40030
},
{
"epoch": 5.1248,
"grad_norm": 0.537740170955658,
"learning_rate": 0.00014471605816162705,
"loss": 1.9764240264892579,
"step": 40040
},
{
"epoch": 5.12608,
"grad_norm": 0.4975545406341553,
"learning_rate": 0.00014465562661144653,
"loss": 1.9669593811035155,
"step": 40050
},
{
"epoch": 5.12736,
"grad_norm": 0.5449622273445129,
"learning_rate": 0.00014459519592980078,
"loss": 1.983632469177246,
"step": 40060
},
{
"epoch": 5.12864,
"grad_norm": 0.5062053203582764,
"learning_rate": 0.00014453476612651058,
"loss": 1.988994026184082,
"step": 40070
},
{
"epoch": 5.12992,
"grad_norm": 0.6204869747161865,
"learning_rate": 0.00014447433721139654,
"loss": 1.9693578720092773,
"step": 40080
},
{
"epoch": 5.1312,
"grad_norm": 0.5652826428413391,
"learning_rate": 0.00014441390919427922,
"loss": 1.9727882385253905,
"step": 40090
},
{
"epoch": 5.13248,
"grad_norm": 0.575089156627655,
"learning_rate": 0.00014435348208497912,
"loss": 1.987898826599121,
"step": 40100
},
{
"epoch": 5.13376,
"grad_norm": 0.5870574116706848,
"learning_rate": 0.00014429305589331635,
"loss": 1.987352180480957,
"step": 40110
},
{
"epoch": 5.13504,
"grad_norm": 0.548961877822876,
"learning_rate": 0.00014423263062911108,
"loss": 1.9919757843017578,
"step": 40120
},
{
"epoch": 5.1363199999999996,
"grad_norm": 0.55159991979599,
"learning_rate": 0.00014417220630218318,
"loss": 1.9878206253051758,
"step": 40130
},
{
"epoch": 5.1376,
"grad_norm": 0.5241367816925049,
"learning_rate": 0.00014411178292235256,
"loss": 1.9806354522705079,
"step": 40140
},
{
"epoch": 5.13888,
"grad_norm": 0.5616269707679749,
"learning_rate": 0.00014405136049943873,
"loss": 1.9750194549560547,
"step": 40150
},
{
"epoch": 5.14016,
"grad_norm": 0.5951315760612488,
"learning_rate": 0.00014399093904326124,
"loss": 1.980186653137207,
"step": 40160
},
{
"epoch": 5.14144,
"grad_norm": 0.5750012397766113,
"learning_rate": 0.00014393051856363944,
"loss": 1.9692890167236328,
"step": 40170
},
{
"epoch": 5.14272,
"grad_norm": 0.5260754823684692,
"learning_rate": 0.00014387009907039242,
"loss": 1.980681800842285,
"step": 40180
},
{
"epoch": 5.144,
"grad_norm": 0.71865314245224,
"learning_rate": 0.0001438096805733392,
"loss": 1.9984758377075196,
"step": 40190
},
{
"epoch": 5.14528,
"grad_norm": 0.5728130340576172,
"learning_rate": 0.0001437492630822987,
"loss": 1.9882091522216796,
"step": 40200
},
{
"epoch": 5.14656,
"grad_norm": 0.5627272725105286,
"learning_rate": 0.00014368884660708947,
"loss": 1.9902706146240234,
"step": 40210
},
{
"epoch": 5.14784,
"grad_norm": 0.5522575378417969,
"learning_rate": 0.00014362843115753012,
"loss": 1.9770435333251952,
"step": 40220
},
{
"epoch": 5.14912,
"grad_norm": 0.548160195350647,
"learning_rate": 0.000143568016743439,
"loss": 1.9995500564575195,
"step": 40230
},
{
"epoch": 5.1504,
"grad_norm": 0.5516613721847534,
"learning_rate": 0.00014350760337463424,
"loss": 1.9851423263549806,
"step": 40240
},
{
"epoch": 5.15168,
"grad_norm": 0.5399978756904602,
"learning_rate": 0.00014344719106093384,
"loss": 2.0002323150634767,
"step": 40250
},
{
"epoch": 5.15296,
"grad_norm": 0.6714414358139038,
"learning_rate": 0.00014338677981215567,
"loss": 1.984640121459961,
"step": 40260
},
{
"epoch": 5.15424,
"grad_norm": 0.5519720315933228,
"learning_rate": 0.00014332636963811747,
"loss": 1.9935543060302734,
"step": 40270
},
{
"epoch": 5.15552,
"grad_norm": 0.5444661378860474,
"learning_rate": 0.00014326596054863657,
"loss": 1.9796789169311524,
"step": 40280
},
{
"epoch": 5.1568,
"grad_norm": 0.6557137966156006,
"learning_rate": 0.00014320555255353042,
"loss": 1.9772241592407227,
"step": 40290
},
{
"epoch": 5.15808,
"grad_norm": 0.5694273710250854,
"learning_rate": 0.00014314514566261612,
"loss": 1.9712583541870117,
"step": 40300
},
{
"epoch": 5.15936,
"grad_norm": 0.49571314454078674,
"learning_rate": 0.00014308473988571056,
"loss": 1.9938528060913085,
"step": 40310
},
{
"epoch": 5.16064,
"grad_norm": 0.5856782793998718,
"learning_rate": 0.00014302433523263056,
"loss": 1.9798959732055663,
"step": 40320
},
{
"epoch": 5.16192,
"grad_norm": 0.5995221138000488,
"learning_rate": 0.00014296393171319278,
"loss": 1.973381233215332,
"step": 40330
},
{
"epoch": 5.1632,
"grad_norm": 0.627404510974884,
"learning_rate": 0.0001429035293372135,
"loss": 1.9905536651611329,
"step": 40340
},
{
"epoch": 5.16448,
"grad_norm": 0.5976679921150208,
"learning_rate": 0.00014284312811450905,
"loss": 1.9659038543701173,
"step": 40350
},
{
"epoch": 5.16576,
"grad_norm": 0.6107878684997559,
"learning_rate": 0.00014278272805489544,
"loss": 1.974735641479492,
"step": 40360
},
{
"epoch": 5.16704,
"grad_norm": 0.5923802256584167,
"learning_rate": 0.00014272232916818847,
"loss": 1.9949413299560548,
"step": 40370
},
{
"epoch": 5.16832,
"grad_norm": 0.5390204191207886,
"learning_rate": 0.00014266193146420382,
"loss": 1.9924118041992187,
"step": 40380
},
{
"epoch": 5.1696,
"grad_norm": 0.5415694713592529,
"learning_rate": 0.00014260153495275696,
"loss": 1.964266586303711,
"step": 40390
},
{
"epoch": 5.17088,
"grad_norm": 0.5004066824913025,
"learning_rate": 0.00014254113964366319,
"loss": 1.975148582458496,
"step": 40400
},
{
"epoch": 5.17216,
"grad_norm": 0.5363929867744446,
"learning_rate": 0.00014248074554673755,
"loss": 2.000972938537598,
"step": 40410
},
{
"epoch": 5.17344,
"grad_norm": 0.6247843503952026,
"learning_rate": 0.00014242035267179488,
"loss": 1.9891450881958008,
"step": 40420
},
{
"epoch": 5.17472,
"grad_norm": 0.5139356851577759,
"learning_rate": 0.00014235996102865,
"loss": 1.9859621047973632,
"step": 40430
},
{
"epoch": 5.176,
"grad_norm": 0.6279066205024719,
"learning_rate": 0.0001422995706271172,
"loss": 1.9923038482666016,
"step": 40440
},
{
"epoch": 5.17728,
"grad_norm": 0.545271098613739,
"learning_rate": 0.00014223918147701086,
"loss": 2.011551856994629,
"step": 40450
},
{
"epoch": 5.17856,
"grad_norm": 0.5943841934204102,
"learning_rate": 0.0001421787935881451,
"loss": 1.9750568389892578,
"step": 40460
},
{
"epoch": 5.17984,
"grad_norm": 0.5670748353004456,
"learning_rate": 0.00014211840697033368,
"loss": 1.975520706176758,
"step": 40470
},
{
"epoch": 5.18112,
"grad_norm": 0.49552613496780396,
"learning_rate": 0.00014205802163339028,
"loss": 1.9868440628051758,
"step": 40480
},
{
"epoch": 5.1824,
"grad_norm": 0.563875675201416,
"learning_rate": 0.0001419976375871284,
"loss": 1.984913444519043,
"step": 40490
},
{
"epoch": 5.18368,
"grad_norm": 0.5245984196662903,
"learning_rate": 0.00014193725484136133,
"loss": 1.9902751922607422,
"step": 40500
},
{
"epoch": 5.18496,
"grad_norm": 0.508735179901123,
"learning_rate": 0.000141876873405902,
"loss": 1.9617176055908203,
"step": 40510
},
{
"epoch": 5.18624,
"grad_norm": 0.5766108632087708,
"learning_rate": 0.00014181649329056325,
"loss": 2.0038837432861327,
"step": 40520
},
{
"epoch": 5.18752,
"grad_norm": 0.567573606967926,
"learning_rate": 0.00014175611450515779,
"loss": 1.990386390686035,
"step": 40530
},
{
"epoch": 5.1888,
"grad_norm": 0.622180700302124,
"learning_rate": 0.00014169573705949783,
"loss": 1.977279281616211,
"step": 40540
},
{
"epoch": 5.19008,
"grad_norm": 0.5349467992782593,
"learning_rate": 0.00014163536096339563,
"loss": 1.960148239135742,
"step": 40550
},
{
"epoch": 5.19136,
"grad_norm": 0.6138989925384521,
"learning_rate": 0.00014157498622666324,
"loss": 1.9653190612792968,
"step": 40560
},
{
"epoch": 5.19264,
"grad_norm": 0.5399139523506165,
"learning_rate": 0.00014151461285911226,
"loss": 1.9580705642700196,
"step": 40570
},
{
"epoch": 5.19392,
"grad_norm": 0.5594148635864258,
"learning_rate": 0.0001414542408705542,
"loss": 1.9733797073364259,
"step": 40580
},
{
"epoch": 5.1952,
"grad_norm": 0.6519774794578552,
"learning_rate": 0.00014139387027080046,
"loss": 1.9765405654907227,
"step": 40590
},
{
"epoch": 5.19648,
"grad_norm": 0.5991188287734985,
"learning_rate": 0.00014133350106966194,
"loss": 1.9832271575927733,
"step": 40600
},
{
"epoch": 5.19776,
"grad_norm": 0.5100942850112915,
"learning_rate": 0.00014127313327694956,
"loss": 1.966326904296875,
"step": 40610
},
{
"epoch": 5.19904,
"grad_norm": 0.5690757036209106,
"learning_rate": 0.00014121276690247394,
"loss": 1.976454734802246,
"step": 40620
},
{
"epoch": 5.20032,
"grad_norm": 0.6070961356163025,
"learning_rate": 0.00014115240195604545,
"loss": 1.983807373046875,
"step": 40630
},
{
"epoch": 5.2016,
"grad_norm": 0.5455453395843506,
"learning_rate": 0.00014109203844747418,
"loss": 1.9710678100585937,
"step": 40640
},
{
"epoch": 5.20288,
"grad_norm": 0.5097793340682983,
"learning_rate": 0.00014103167638657006,
"loss": 1.97595157623291,
"step": 40650
},
{
"epoch": 5.20416,
"grad_norm": 0.5934730768203735,
"learning_rate": 0.00014097131578314281,
"loss": 1.9855283737182616,
"step": 40660
},
{
"epoch": 5.20544,
"grad_norm": 0.5546727776527405,
"learning_rate": 0.00014091095664700177,
"loss": 2.0005664825439453,
"step": 40670
},
{
"epoch": 5.20672,
"grad_norm": 0.5520148277282715,
"learning_rate": 0.00014085059898795622,
"loss": 1.9759883880615234,
"step": 40680
},
{
"epoch": 5.208,
"grad_norm": 0.541770339012146,
"learning_rate": 0.00014079024281581505,
"loss": 1.9791215896606444,
"step": 40690
},
{
"epoch": 5.20928,
"grad_norm": 0.6461961269378662,
"learning_rate": 0.00014072988814038703,
"loss": 1.994040298461914,
"step": 40700
},
{
"epoch": 5.21056,
"grad_norm": 0.5671346187591553,
"learning_rate": 0.00014066953497148058,
"loss": 1.9652133941650392,
"step": 40710
},
{
"epoch": 5.21184,
"grad_norm": 0.545914351940155,
"learning_rate": 0.000140609183318904,
"loss": 1.9849449157714845,
"step": 40720
},
{
"epoch": 5.21312,
"grad_norm": 0.5209454894065857,
"learning_rate": 0.00014054883319246516,
"loss": 1.982111358642578,
"step": 40730
},
{
"epoch": 5.2144,
"grad_norm": 0.5175140500068665,
"learning_rate": 0.00014048848460197191,
"loss": 1.9846864700317384,
"step": 40740
},
{
"epoch": 5.21568,
"grad_norm": 0.5139882564544678,
"learning_rate": 0.00014042813755723161,
"loss": 1.9916515350341797,
"step": 40750
},
{
"epoch": 5.21696,
"grad_norm": 0.5873563885688782,
"learning_rate": 0.00014036779206805166,
"loss": 1.9742139816284179,
"step": 40760
},
{
"epoch": 5.21824,
"grad_norm": 0.4965938627719879,
"learning_rate": 0.0001403074481442388,
"loss": 1.9748699188232421,
"step": 40770
},
{
"epoch": 5.21952,
"grad_norm": 0.5344886183738708,
"learning_rate": 0.00014024710579559996,
"loss": 1.968344497680664,
"step": 40780
},
{
"epoch": 5.2208,
"grad_norm": 0.6017478108406067,
"learning_rate": 0.00014018676503194157,
"loss": 1.9773651123046876,
"step": 40790
},
{
"epoch": 5.22208,
"grad_norm": 0.5832433700561523,
"learning_rate": 0.00014012642586306976,
"loss": 1.9731361389160156,
"step": 40800
},
{
"epoch": 5.22336,
"grad_norm": 0.478929728269577,
"learning_rate": 0.00014006608829879053,
"loss": 1.9785823822021484,
"step": 40810
},
{
"epoch": 5.22464,
"grad_norm": 0.5456215143203735,
"learning_rate": 0.00014000575234890958,
"loss": 1.983126449584961,
"step": 40820
},
{
"epoch": 5.22592,
"grad_norm": 0.6200441122055054,
"learning_rate": 0.00013994541802323228,
"loss": 1.9686902999877929,
"step": 40830
},
{
"epoch": 5.2272,
"grad_norm": 0.5317085385322571,
"learning_rate": 0.00013988508533156386,
"loss": 1.981093406677246,
"step": 40840
},
{
"epoch": 5.22848,
"grad_norm": 0.5997294187545776,
"learning_rate": 0.0001398247542837092,
"loss": 1.9822105407714843,
"step": 40850
},
{
"epoch": 5.22976,
"grad_norm": 0.5115920901298523,
"learning_rate": 0.00013976442488947298,
"loss": 1.9917991638183594,
"step": 40860
},
{
"epoch": 5.23104,
"grad_norm": 0.5518493056297302,
"learning_rate": 0.00013970409715865944,
"loss": 1.9713827133178712,
"step": 40870
},
{
"epoch": 5.23232,
"grad_norm": 0.5216674208641052,
"learning_rate": 0.00013964377110107276,
"loss": 1.989434051513672,
"step": 40880
},
{
"epoch": 5.2336,
"grad_norm": 0.5475512146949768,
"learning_rate": 0.00013958344672651676,
"loss": 1.9777565002441406,
"step": 40890
},
{
"epoch": 5.23488,
"grad_norm": 0.6198081374168396,
"learning_rate": 0.00013952312404479493,
"loss": 2.0003461837768555,
"step": 40900
},
{
"epoch": 5.23616,
"grad_norm": 0.5381304025650024,
"learning_rate": 0.00013946280306571058,
"loss": 1.9825506210327148,
"step": 40910
},
{
"epoch": 5.23744,
"grad_norm": 0.5044442415237427,
"learning_rate": 0.00013940248379906679,
"loss": 1.988032341003418,
"step": 40920
},
{
"epoch": 5.23872,
"grad_norm": 0.605139434337616,
"learning_rate": 0.0001393421662546661,
"loss": 1.9757976531982422,
"step": 40930
},
{
"epoch": 5.24,
"grad_norm": 0.5392096638679504,
"learning_rate": 0.00013928185044231103,
"loss": 1.9560731887817382,
"step": 40940
},
{
"epoch": 5.24128,
"grad_norm": 0.6135014295578003,
"learning_rate": 0.0001392215363718038,
"loss": 2.0009540557861327,
"step": 40950
},
{
"epoch": 5.24256,
"grad_norm": 0.4734870195388794,
"learning_rate": 0.00013916122405294618,
"loss": 1.9797842025756835,
"step": 40960
},
{
"epoch": 5.24384,
"grad_norm": 0.5560034513473511,
"learning_rate": 0.0001391009134955398,
"loss": 1.9856050491333008,
"step": 40970
},
{
"epoch": 5.24512,
"grad_norm": 0.5792466998100281,
"learning_rate": 0.00013904060470938597,
"loss": 1.9872072219848633,
"step": 40980
},
{
"epoch": 5.2464,
"grad_norm": 0.5647287964820862,
"learning_rate": 0.00013898029770428573,
"loss": 1.9823156356811524,
"step": 40990
},
{
"epoch": 5.24768,
"grad_norm": 0.5788866877555847,
"learning_rate": 0.00013891999249003974,
"loss": 1.984652328491211,
"step": 41000
},
{
"epoch": 5.24896,
"grad_norm": 0.5349242091178894,
"learning_rate": 0.00013885968907644843,
"loss": 1.9737358093261719,
"step": 41010
},
{
"epoch": 5.25024,
"grad_norm": 0.5411757826805115,
"learning_rate": 0.00013879938747331206,
"loss": 1.9883398056030273,
"step": 41020
},
{
"epoch": 5.25152,
"grad_norm": 0.5418636202812195,
"learning_rate": 0.00013873908769043033,
"loss": 1.9937828063964844,
"step": 41030
},
{
"epoch": 5.2528,
"grad_norm": 0.49643632769584656,
"learning_rate": 0.00013867878973760283,
"loss": 1.9764518737792969,
"step": 41040
},
{
"epoch": 5.25408,
"grad_norm": 0.5732832551002502,
"learning_rate": 0.00013861849362462888,
"loss": 1.9954004287719727,
"step": 41050
},
{
"epoch": 5.25536,
"grad_norm": 0.5661678910255432,
"learning_rate": 0.00013855819936130733,
"loss": 2.0034248352050783,
"step": 41060
},
{
"epoch": 5.25664,
"grad_norm": 0.5144567489624023,
"learning_rate": 0.0001384979069574369,
"loss": 1.9724611282348632,
"step": 41070
},
{
"epoch": 5.25792,
"grad_norm": 0.5505245923995972,
"learning_rate": 0.000138437616422816,
"loss": 1.971287155151367,
"step": 41080
},
{
"epoch": 5.2592,
"grad_norm": 0.6913823485374451,
"learning_rate": 0.00013837732776724254,
"loss": 1.9794195175170899,
"step": 41090
},
{
"epoch": 5.26048,
"grad_norm": 0.502305269241333,
"learning_rate": 0.00013831704100051431,
"loss": 1.9862606048583984,
"step": 41100
},
{
"epoch": 5.26176,
"grad_norm": 0.6145079135894775,
"learning_rate": 0.00013825675613242874,
"loss": 1.9850124359130858,
"step": 41110
},
{
"epoch": 5.26304,
"grad_norm": 0.6269429326057434,
"learning_rate": 0.00013819647317278306,
"loss": 1.9766742706298828,
"step": 41120
},
{
"epoch": 5.26432,
"grad_norm": 0.5069852471351624,
"learning_rate": 0.00013813619213137395,
"loss": 1.9677963256835938,
"step": 41130
},
{
"epoch": 5.2656,
"grad_norm": 0.5314162969589233,
"learning_rate": 0.000138075913017998,
"loss": 1.9868083953857423,
"step": 41140
},
{
"epoch": 5.2668800000000005,
"grad_norm": 0.5171386003494263,
"learning_rate": 0.00013801563584245138,
"loss": 1.9863117218017579,
"step": 41150
},
{
"epoch": 5.26816,
"grad_norm": 0.5576849579811096,
"learning_rate": 0.00013795536061452993,
"loss": 1.9646520614624023,
"step": 41160
},
{
"epoch": 5.26944,
"grad_norm": 0.5731988549232483,
"learning_rate": 0.0001378950873440292,
"loss": 1.9740877151489258,
"step": 41170
},
{
"epoch": 5.27072,
"grad_norm": 0.5636196136474609,
"learning_rate": 0.00013783481604074454,
"loss": 1.965505599975586,
"step": 41180
},
{
"epoch": 5.272,
"grad_norm": 0.5637430548667908,
"learning_rate": 0.00013777454671447072,
"loss": 1.964957046508789,
"step": 41190
},
{
"epoch": 5.27328,
"grad_norm": 0.5193994045257568,
"learning_rate": 0.00013771427937500248,
"loss": 1.9718774795532226,
"step": 41200
},
{
"epoch": 5.27456,
"grad_norm": 0.4872111976146698,
"learning_rate": 0.00013765401403213404,
"loss": 1.9735729217529296,
"step": 41210
},
{
"epoch": 5.27584,
"grad_norm": 0.5604969263076782,
"learning_rate": 0.0001375937506956593,
"loss": 1.996666145324707,
"step": 41220
},
{
"epoch": 5.27712,
"grad_norm": 0.5323799848556519,
"learning_rate": 0.00013753348937537192,
"loss": 2.0031400680541993,
"step": 41230
},
{
"epoch": 5.2783999999999995,
"grad_norm": 0.5509191751480103,
"learning_rate": 0.00013747323008106525,
"loss": 1.9796882629394532,
"step": 41240
},
{
"epoch": 5.27968,
"grad_norm": 0.5598781704902649,
"learning_rate": 0.00013741297282253223,
"loss": 1.976577377319336,
"step": 41250
},
{
"epoch": 5.28096,
"grad_norm": 0.5135860443115234,
"learning_rate": 0.0001373527176095655,
"loss": 1.9931276321411133,
"step": 41260
},
{
"epoch": 5.28224,
"grad_norm": 0.5743052363395691,
"learning_rate": 0.00013729246445195732,
"loss": 1.965625,
"step": 41270
},
{
"epoch": 5.28352,
"grad_norm": 0.5317575335502625,
"learning_rate": 0.0001372322133594998,
"loss": 1.994522476196289,
"step": 41280
},
{
"epoch": 5.2848,
"grad_norm": 0.531608521938324,
"learning_rate": 0.00013717196434198436,
"loss": 1.9874383926391601,
"step": 41290
},
{
"epoch": 5.28608,
"grad_norm": 0.5175894498825073,
"learning_rate": 0.0001371117174092025,
"loss": 1.9892309188842774,
"step": 41300
},
{
"epoch": 5.28736,
"grad_norm": 0.5338895320892334,
"learning_rate": 0.0001370514725709451,
"loss": 1.9741422653198242,
"step": 41310
},
{
"epoch": 5.28864,
"grad_norm": 0.5617733001708984,
"learning_rate": 0.00013699122983700275,
"loss": 1.988671112060547,
"step": 41320
},
{
"epoch": 5.28992,
"grad_norm": 0.5422617197036743,
"learning_rate": 0.0001369309892171658,
"loss": 1.9791154861450195,
"step": 41330
},
{
"epoch": 5.2912,
"grad_norm": 0.5600425601005554,
"learning_rate": 0.00013687075072122407,
"loss": 1.9971279144287108,
"step": 41340
},
{
"epoch": 5.29248,
"grad_norm": 0.48067471385002136,
"learning_rate": 0.0001368105143589673,
"loss": 1.9809841156005858,
"step": 41350
},
{
"epoch": 5.29376,
"grad_norm": 0.5189995169639587,
"learning_rate": 0.00013675028014018464,
"loss": 1.9744823455810547,
"step": 41360
},
{
"epoch": 5.29504,
"grad_norm": 0.5661554932594299,
"learning_rate": 0.000136690048074665,
"loss": 1.9987991333007813,
"step": 41370
},
{
"epoch": 5.29632,
"grad_norm": 0.5942940711975098,
"learning_rate": 0.000136629818172197,
"loss": 1.9930763244628906,
"step": 41380
},
{
"epoch": 5.2976,
"grad_norm": 0.590764045715332,
"learning_rate": 0.00013656959044256866,
"loss": 2.004757308959961,
"step": 41390
},
{
"epoch": 5.29888,
"grad_norm": 0.5370355844497681,
"learning_rate": 0.00013650936489556793,
"loss": 1.9968799591064452,
"step": 41400
},
{
"epoch": 5.30016,
"grad_norm": 0.5596864819526672,
"learning_rate": 0.00013644914154098238,
"loss": 1.9749874114990233,
"step": 41410
},
{
"epoch": 5.30144,
"grad_norm": 0.5401380062103271,
"learning_rate": 0.00013638892038859898,
"loss": 1.9818048477172852,
"step": 41420
},
{
"epoch": 5.30272,
"grad_norm": 0.6114886403083801,
"learning_rate": 0.00013632870144820457,
"loss": 1.9852470397949218,
"step": 41430
},
{
"epoch": 5.304,
"grad_norm": 0.5634302496910095,
"learning_rate": 0.00013626848472958564,
"loss": 1.987173080444336,
"step": 41440
},
{
"epoch": 5.30528,
"grad_norm": 0.5364611148834229,
"learning_rate": 0.00013620827024252808,
"loss": 1.976517105102539,
"step": 41450
},
{
"epoch": 5.30656,
"grad_norm": 0.5797374844551086,
"learning_rate": 0.00013614805799681766,
"loss": 1.9988370895385743,
"step": 41460
},
{
"epoch": 5.30784,
"grad_norm": 0.5378036499023438,
"learning_rate": 0.00013608784800223976,
"loss": 1.9837087631225585,
"step": 41470
},
{
"epoch": 5.30912,
"grad_norm": 0.5967125296592712,
"learning_rate": 0.00013602764026857927,
"loss": 1.9630762100219727,
"step": 41480
},
{
"epoch": 5.3104,
"grad_norm": 0.793624758720398,
"learning_rate": 0.00013596743480562078,
"loss": 1.986646842956543,
"step": 41490
},
{
"epoch": 5.31168,
"grad_norm": 0.5153248310089111,
"learning_rate": 0.00013590723162314853,
"loss": 1.9850374221801759,
"step": 41500
},
{
"epoch": 5.31296,
"grad_norm": 0.625209391117096,
"learning_rate": 0.00013584703073094638,
"loss": 1.996918296813965,
"step": 41510
},
{
"epoch": 5.31424,
"grad_norm": 0.585951030254364,
"learning_rate": 0.00013578683213879777,
"loss": 1.9977087020874023,
"step": 41520
},
{
"epoch": 5.31552,
"grad_norm": 0.6587132215499878,
"learning_rate": 0.00013572663585648587,
"loss": 1.9868583679199219,
"step": 41530
},
{
"epoch": 5.3168,
"grad_norm": 0.5574489235877991,
"learning_rate": 0.00013566644189379334,
"loss": 1.9875570297241212,
"step": 41540
},
{
"epoch": 5.31808,
"grad_norm": 0.6591134667396545,
"learning_rate": 0.0001356062502605026,
"loss": 1.9837024688720704,
"step": 41550
},
{
"epoch": 5.31936,
"grad_norm": 0.6572789549827576,
"learning_rate": 0.00013554606096639554,
"loss": 1.9721830368041993,
"step": 41560
},
{
"epoch": 5.32064,
"grad_norm": 0.5816863179206848,
"learning_rate": 0.00013548587402125388,
"loss": 1.9871910095214844,
"step": 41570
},
{
"epoch": 5.32192,
"grad_norm": 0.6315808296203613,
"learning_rate": 0.00013542568943485868,
"loss": 1.9822914123535156,
"step": 41580
},
{
"epoch": 5.3232,
"grad_norm": 0.46478450298309326,
"learning_rate": 0.00013536550721699088,
"loss": 1.9868824005126953,
"step": 41590
},
{
"epoch": 5.32448,
"grad_norm": 0.6518846154212952,
"learning_rate": 0.00013530532737743086,
"loss": 1.9668336868286134,
"step": 41600
},
{
"epoch": 5.32576,
"grad_norm": 0.5457481741905212,
"learning_rate": 0.00013524514992595876,
"loss": 1.9905780792236327,
"step": 41610
},
{
"epoch": 5.32704,
"grad_norm": 0.6036630868911743,
"learning_rate": 0.00013518497487235415,
"loss": 1.9800395965576172,
"step": 41620
},
{
"epoch": 5.32832,
"grad_norm": 0.5176271796226501,
"learning_rate": 0.00013512480222639634,
"loss": 1.9863203048706055,
"step": 41630
},
{
"epoch": 5.3296,
"grad_norm": 0.6342214345932007,
"learning_rate": 0.0001350646319978643,
"loss": 1.9735530853271483,
"step": 41640
},
{
"epoch": 5.33088,
"grad_norm": 0.5800303220748901,
"learning_rate": 0.00013500446419653642,
"loss": 1.987746810913086,
"step": 41650
},
{
"epoch": 5.33216,
"grad_norm": 0.5853630900382996,
"learning_rate": 0.00013494429883219083,
"loss": 2.0032562255859374,
"step": 41660
},
{
"epoch": 5.33344,
"grad_norm": 0.5422623753547668,
"learning_rate": 0.0001348841359146053,
"loss": 1.976350975036621,
"step": 41670
},
{
"epoch": 5.33472,
"grad_norm": 0.6151648759841919,
"learning_rate": 0.00013482397545355702,
"loss": 1.9816465377807617,
"step": 41680
},
{
"epoch": 5.336,
"grad_norm": 0.590786337852478,
"learning_rate": 0.00013476381745882297,
"loss": 1.9854684829711915,
"step": 41690
},
{
"epoch": 5.33728,
"grad_norm": 0.5752931237220764,
"learning_rate": 0.00013470366194017966,
"loss": 1.9830503463745117,
"step": 41700
},
{
"epoch": 5.33856,
"grad_norm": 0.5542811751365662,
"learning_rate": 0.00013464350890740324,
"loss": 1.9848127365112305,
"step": 41710
},
{
"epoch": 5.33984,
"grad_norm": 0.6969866156578064,
"learning_rate": 0.00013458335837026927,
"loss": 2.0015079498291017,
"step": 41720
},
{
"epoch": 5.34112,
"grad_norm": 0.6784240007400513,
"learning_rate": 0.00013452321033855316,
"loss": 1.9969789505004882,
"step": 41730
},
{
"epoch": 5.3424,
"grad_norm": 0.555943489074707,
"learning_rate": 0.00013446306482202983,
"loss": 1.9974437713623048,
"step": 41740
},
{
"epoch": 5.34368,
"grad_norm": 0.5375508666038513,
"learning_rate": 0.00013440292183047363,
"loss": 1.9894983291625976,
"step": 41750
},
{
"epoch": 5.34496,
"grad_norm": 0.5872993469238281,
"learning_rate": 0.00013434278137365872,
"loss": 1.9767759323120118,
"step": 41760
},
{
"epoch": 5.34624,
"grad_norm": 0.4987710416316986,
"learning_rate": 0.00013428264346135877,
"loss": 1.9788856506347656,
"step": 41770
},
{
"epoch": 5.34752,
"grad_norm": 0.5719423890113831,
"learning_rate": 0.00013422250810334694,
"loss": 1.962596321105957,
"step": 41780
},
{
"epoch": 5.3488,
"grad_norm": 0.5370954275131226,
"learning_rate": 0.0001341623753093961,
"loss": 1.9899978637695312,
"step": 41790
},
{
"epoch": 5.35008,
"grad_norm": 0.5505043268203735,
"learning_rate": 0.00013410224508927874,
"loss": 1.983302116394043,
"step": 41800
},
{
"epoch": 5.35136,
"grad_norm": 0.6430127620697021,
"learning_rate": 0.0001340421174527667,
"loss": 1.9874677658081055,
"step": 41810
},
{
"epoch": 5.35264,
"grad_norm": 0.5549014806747437,
"learning_rate": 0.0001339819924096317,
"loss": 1.9797426223754884,
"step": 41820
},
{
"epoch": 5.3539200000000005,
"grad_norm": 0.48545536398887634,
"learning_rate": 0.00013392186996964478,
"loss": 1.987848663330078,
"step": 41830
},
{
"epoch": 5.3552,
"grad_norm": 0.49766141176223755,
"learning_rate": 0.00013386175014257678,
"loss": 1.9840032577514648,
"step": 41840
},
{
"epoch": 5.35648,
"grad_norm": 0.5071004629135132,
"learning_rate": 0.0001338016329381979,
"loss": 1.9783273696899415,
"step": 41850
},
{
"epoch": 5.35776,
"grad_norm": 0.5669161081314087,
"learning_rate": 0.00013374151836627806,
"loss": 1.9850311279296875,
"step": 41860
},
{
"epoch": 5.35904,
"grad_norm": 0.5111224055290222,
"learning_rate": 0.00013368140643658677,
"loss": 1.9945789337158204,
"step": 41870
},
{
"epoch": 5.36032,
"grad_norm": 0.6820855140686035,
"learning_rate": 0.00013362129715889298,
"loss": 1.9525671005249023,
"step": 41880
},
{
"epoch": 5.3616,
"grad_norm": 0.5566214919090271,
"learning_rate": 0.00013356119054296525,
"loss": 1.997808837890625,
"step": 41890
},
{
"epoch": 5.36288,
"grad_norm": 0.4983486235141754,
"learning_rate": 0.00013350108659857185,
"loss": 1.9829071044921875,
"step": 41900
},
{
"epoch": 5.36416,
"grad_norm": 0.5923448801040649,
"learning_rate": 0.00013344098533548036,
"loss": 1.9881202697753906,
"step": 41910
},
{
"epoch": 5.3654399999999995,
"grad_norm": 0.542529821395874,
"learning_rate": 0.00013338088676345816,
"loss": 2.0010839462280274,
"step": 41920
},
{
"epoch": 5.36672,
"grad_norm": 0.5152354836463928,
"learning_rate": 0.00013332079089227214,
"loss": 1.9676273345947266,
"step": 41930
},
{
"epoch": 5.368,
"grad_norm": 0.5243868827819824,
"learning_rate": 0.00013326069773168865,
"loss": 1.9854520797729491,
"step": 41940
},
{
"epoch": 5.36928,
"grad_norm": 0.569597065448761,
"learning_rate": 0.0001332006072914736,
"loss": 1.9843923568725585,
"step": 41950
},
{
"epoch": 5.37056,
"grad_norm": 0.5500484704971313,
"learning_rate": 0.00013314051958139263,
"loss": 2.005801773071289,
"step": 41960
},
{
"epoch": 5.37184,
"grad_norm": 0.511417031288147,
"learning_rate": 0.0001330804346112108,
"loss": 1.9727169036865235,
"step": 41970
},
{
"epoch": 5.37312,
"grad_norm": 0.5581918358802795,
"learning_rate": 0.0001330203523906927,
"loss": 1.9793724060058593,
"step": 41980
},
{
"epoch": 5.3744,
"grad_norm": 0.5457528233528137,
"learning_rate": 0.00013296027292960258,
"loss": 1.9967533111572267,
"step": 41990
},
{
"epoch": 5.37568,
"grad_norm": 0.5484282970428467,
"learning_rate": 0.0001329001962377042,
"loss": 2.003395843505859,
"step": 42000
},
{
"epoch": 5.37696,
"grad_norm": 0.5372671484947205,
"learning_rate": 0.00013284012232476074,
"loss": 1.99564208984375,
"step": 42010
},
{
"epoch": 5.37824,
"grad_norm": 0.7000912427902222,
"learning_rate": 0.00013278005120053512,
"loss": 1.9873809814453125,
"step": 42020
},
{
"epoch": 5.37952,
"grad_norm": 0.5784828066825867,
"learning_rate": 0.0001327199828747898,
"loss": 1.975860595703125,
"step": 42030
},
{
"epoch": 5.3808,
"grad_norm": 0.6562021970748901,
"learning_rate": 0.00013265991735728655,
"loss": 1.9766521453857422,
"step": 42040
},
{
"epoch": 5.38208,
"grad_norm": 0.5688745379447937,
"learning_rate": 0.00013259985465778701,
"loss": 1.9763147354125976,
"step": 42050
},
{
"epoch": 5.38336,
"grad_norm": 0.5956392288208008,
"learning_rate": 0.0001325397947860521,
"loss": 1.9824779510498047,
"step": 42060
},
{
"epoch": 5.38464,
"grad_norm": 0.5860964059829712,
"learning_rate": 0.00013247973775184244,
"loss": 1.9935476303100585,
"step": 42070
},
{
"epoch": 5.38592,
"grad_norm": 0.5184463858604431,
"learning_rate": 0.00013241968356491807,
"loss": 1.991524887084961,
"step": 42080
},
{
"epoch": 5.3872,
"grad_norm": 0.5186555981636047,
"learning_rate": 0.00013235963223503865,
"loss": 1.9914512634277344,
"step": 42090
},
{
"epoch": 5.38848,
"grad_norm": 0.5447313785552979,
"learning_rate": 0.00013229958377196344,
"loss": 1.9820980072021483,
"step": 42100
},
{
"epoch": 5.38976,
"grad_norm": 0.5571222901344299,
"learning_rate": 0.00013223953818545102,
"loss": 1.9802343368530273,
"step": 42110
},
{
"epoch": 5.39104,
"grad_norm": 0.5370756983757019,
"learning_rate": 0.00013217949548525966,
"loss": 1.9871055603027343,
"step": 42120
},
{
"epoch": 5.39232,
"grad_norm": 0.5395784974098206,
"learning_rate": 0.0001321194556811472,
"loss": 1.9738758087158204,
"step": 42130
},
{
"epoch": 5.3936,
"grad_norm": 0.5368621349334717,
"learning_rate": 0.00013205941878287087,
"loss": 1.985044288635254,
"step": 42140
},
{
"epoch": 5.39488,
"grad_norm": 0.546603262424469,
"learning_rate": 0.00013199938480018748,
"loss": 2.0017009735107423,
"step": 42150
},
{
"epoch": 5.39616,
"grad_norm": 0.5311381816864014,
"learning_rate": 0.0001319393537428535,
"loss": 1.9789604187011718,
"step": 42160
},
{
"epoch": 5.39744,
"grad_norm": 0.5148146152496338,
"learning_rate": 0.0001318793256206247,
"loss": 1.9912275314331054,
"step": 42170
},
{
"epoch": 5.39872,
"grad_norm": 0.5618871450424194,
"learning_rate": 0.0001318193004432565,
"loss": 1.99395809173584,
"step": 42180
},
{
"epoch": 5.4,
"grad_norm": 0.5344218611717224,
"learning_rate": 0.00013175927822050377,
"loss": 1.9580873489379882,
"step": 42190
},
{
"epoch": 5.40128,
"grad_norm": 0.5136785507202148,
"learning_rate": 0.00013169925896212116,
"loss": 1.9612302780151367,
"step": 42200
},
{
"epoch": 5.40256,
"grad_norm": 0.6413950324058533,
"learning_rate": 0.00013163924267786236,
"loss": 1.9747426986694336,
"step": 42210
},
{
"epoch": 5.40384,
"grad_norm": 0.5182486772537231,
"learning_rate": 0.00013157922937748106,
"loss": 1.9677410125732422,
"step": 42220
},
{
"epoch": 5.40512,
"grad_norm": 0.530019998550415,
"learning_rate": 0.00013151921907073017,
"loss": 1.9809322357177734,
"step": 42230
},
{
"epoch": 5.4064,
"grad_norm": 0.5610824823379517,
"learning_rate": 0.00013145921176736214,
"loss": 1.9813684463500976,
"step": 42240
},
{
"epoch": 5.40768,
"grad_norm": 0.6297482252120972,
"learning_rate": 0.00013139920747712904,
"loss": 1.9961540222167968,
"step": 42250
},
{
"epoch": 5.40896,
"grad_norm": 0.5692352652549744,
"learning_rate": 0.00013133920620978247,
"loss": 1.9881837844848633,
"step": 42260
},
{
"epoch": 5.41024,
"grad_norm": 0.5268560647964478,
"learning_rate": 0.00013127920797507335,
"loss": 1.988850975036621,
"step": 42270
},
{
"epoch": 5.41152,
"grad_norm": 0.5636352896690369,
"learning_rate": 0.00013121921278275227,
"loss": 1.995048713684082,
"step": 42280
},
{
"epoch": 5.4128,
"grad_norm": 0.536689043045044,
"learning_rate": 0.00013115922064256934,
"loss": 1.9744501113891602,
"step": 42290
},
{
"epoch": 5.41408,
"grad_norm": 0.5223714113235474,
"learning_rate": 0.000131099231564274,
"loss": 1.983205795288086,
"step": 42300
},
{
"epoch": 5.41536,
"grad_norm": 0.5272925496101379,
"learning_rate": 0.00013103924555761534,
"loss": 1.9593936920166015,
"step": 42310
},
{
"epoch": 5.41664,
"grad_norm": 0.5660974383354187,
"learning_rate": 0.000130979262632342,
"loss": 1.9742008209228517,
"step": 42320
},
{
"epoch": 5.41792,
"grad_norm": 0.5386303663253784,
"learning_rate": 0.00013091928279820194,
"loss": 1.9738498687744142,
"step": 42330
},
{
"epoch": 5.4192,
"grad_norm": 0.530667781829834,
"learning_rate": 0.00013085930606494277,
"loss": 1.997178077697754,
"step": 42340
},
{
"epoch": 5.42048,
"grad_norm": 0.5362817645072937,
"learning_rate": 0.0001307993324423115,
"loss": 1.9797508239746093,
"step": 42350
},
{
"epoch": 5.42176,
"grad_norm": 0.5410745739936829,
"learning_rate": 0.00013073936194005475,
"loss": 1.980545997619629,
"step": 42360
},
{
"epoch": 5.42304,
"grad_norm": 0.5476128458976746,
"learning_rate": 0.00013067939456791843,
"loss": 1.982173538208008,
"step": 42370
},
{
"epoch": 5.42432,
"grad_norm": 0.5028052926063538,
"learning_rate": 0.0001306194303356482,
"loss": 1.994675636291504,
"step": 42380
},
{
"epoch": 5.4256,
"grad_norm": 0.543493926525116,
"learning_rate": 0.00013055946925298899,
"loss": 1.9975576400756836,
"step": 42390
},
{
"epoch": 5.42688,
"grad_norm": 0.6385568380355835,
"learning_rate": 0.00013049951132968534,
"loss": 1.9856590270996093,
"step": 42400
},
{
"epoch": 5.42816,
"grad_norm": 0.584900975227356,
"learning_rate": 0.0001304395565754812,
"loss": 1.9822347640991211,
"step": 42410
},
{
"epoch": 5.42944,
"grad_norm": 0.5094760060310364,
"learning_rate": 0.00013037960500012017,
"loss": 1.984293556213379,
"step": 42420
},
{
"epoch": 5.43072,
"grad_norm": 0.5932551622390747,
"learning_rate": 0.00013031965661334507,
"loss": 1.9834859848022461,
"step": 42430
},
{
"epoch": 5.432,
"grad_norm": 0.49106287956237793,
"learning_rate": 0.00013025971142489843,
"loss": 1.995701789855957,
"step": 42440
},
{
"epoch": 5.43328,
"grad_norm": 0.5897199511528015,
"learning_rate": 0.0001301997694445221,
"loss": 1.9877338409423828,
"step": 42450
},
{
"epoch": 5.43456,
"grad_norm": 0.6547048091888428,
"learning_rate": 0.0001301398306819576,
"loss": 1.9811365127563476,
"step": 42460
},
{
"epoch": 5.43584,
"grad_norm": 0.6407286524772644,
"learning_rate": 0.00013007989514694568,
"loss": 1.9855857849121095,
"step": 42470
},
{
"epoch": 5.43712,
"grad_norm": 0.6334633231163025,
"learning_rate": 0.00013001996284922674,
"loss": 1.9707197189331054,
"step": 42480
},
{
"epoch": 5.4384,
"grad_norm": 0.5733006596565247,
"learning_rate": 0.0001299600337985407,
"loss": 1.9843164443969727,
"step": 42490
},
{
"epoch": 5.43968,
"grad_norm": 0.5275271534919739,
"learning_rate": 0.00012990010800462674,
"loss": 1.9828315734863282,
"step": 42500
},
{
"epoch": 5.4409600000000005,
"grad_norm": 0.6423556804656982,
"learning_rate": 0.00012984018547722363,
"loss": 1.9886388778686523,
"step": 42510
},
{
"epoch": 5.44224,
"grad_norm": 0.521068274974823,
"learning_rate": 0.00012978026622606975,
"loss": 1.9765953063964843,
"step": 42520
},
{
"epoch": 5.44352,
"grad_norm": 0.5579521656036377,
"learning_rate": 0.0001297203502609026,
"loss": 1.9853191375732422,
"step": 42530
},
{
"epoch": 5.4448,
"grad_norm": 0.5345422625541687,
"learning_rate": 0.00012966043759145952,
"loss": 1.9920236587524414,
"step": 42540
},
{
"epoch": 5.44608,
"grad_norm": 0.5586206912994385,
"learning_rate": 0.0001296005282274771,
"loss": 2.00451602935791,
"step": 42550
},
{
"epoch": 5.44736,
"grad_norm": 0.5866800546646118,
"learning_rate": 0.00012954062217869147,
"loss": 1.9994705200195313,
"step": 42560
},
{
"epoch": 5.44864,
"grad_norm": 0.5355567932128906,
"learning_rate": 0.00012948071945483809,
"loss": 1.9890165328979492,
"step": 42570
},
{
"epoch": 5.44992,
"grad_norm": 0.5373163223266602,
"learning_rate": 0.00012942082006565207,
"loss": 1.9741241455078125,
"step": 42580
},
{
"epoch": 5.4512,
"grad_norm": 0.6275017261505127,
"learning_rate": 0.00012936092402086793,
"loss": 1.95538330078125,
"step": 42590
},
{
"epoch": 5.4524799999999995,
"grad_norm": 0.5589377284049988,
"learning_rate": 0.00012930103133021947,
"loss": 1.979604148864746,
"step": 42600
},
{
"epoch": 5.45376,
"grad_norm": 0.5043761134147644,
"learning_rate": 0.0001292411420034402,
"loss": 1.9590951919555664,
"step": 42610
},
{
"epoch": 5.45504,
"grad_norm": 0.4999364912509918,
"learning_rate": 0.00012918125605026294,
"loss": 1.9968822479248047,
"step": 42620
},
{
"epoch": 5.45632,
"grad_norm": 0.6129090785980225,
"learning_rate": 0.00012912137348041996,
"loss": 1.9768346786499023,
"step": 42630
},
{
"epoch": 5.4576,
"grad_norm": 0.5647895932197571,
"learning_rate": 0.000129061494303643,
"loss": 1.9784723281860352,
"step": 42640
},
{
"epoch": 5.45888,
"grad_norm": 0.573650598526001,
"learning_rate": 0.00012900161852966334,
"loss": 1.997418212890625,
"step": 42650
},
{
"epoch": 5.46016,
"grad_norm": 0.5058845281600952,
"learning_rate": 0.00012894174616821149,
"loss": 1.986598014831543,
"step": 42660
},
{
"epoch": 5.46144,
"grad_norm": 0.5411257743835449,
"learning_rate": 0.00012888187722901766,
"loss": 1.9873947143554687,
"step": 42670
},
{
"epoch": 5.46272,
"grad_norm": 0.5356424450874329,
"learning_rate": 0.00012882201172181125,
"loss": 1.9847625732421874,
"step": 42680
},
{
"epoch": 5.464,
"grad_norm": 0.5442690849304199,
"learning_rate": 0.00012876214965632142,
"loss": 1.9512657165527343,
"step": 42690
},
{
"epoch": 5.46528,
"grad_norm": 0.5769009590148926,
"learning_rate": 0.0001287022910422764,
"loss": 1.9809892654418946,
"step": 42700
},
{
"epoch": 5.46656,
"grad_norm": 0.6198396682739258,
"learning_rate": 0.00012864243588940413,
"loss": 1.976656150817871,
"step": 42710
},
{
"epoch": 5.46784,
"grad_norm": 0.616519570350647,
"learning_rate": 0.00012858258420743197,
"loss": 1.9940874099731445,
"step": 42720
},
{
"epoch": 5.46912,
"grad_norm": 0.49171146750450134,
"learning_rate": 0.0001285227360060865,
"loss": 1.9729713439941405,
"step": 42730
},
{
"epoch": 5.4704,
"grad_norm": 0.58741694688797,
"learning_rate": 0.00012846289129509397,
"loss": 1.9720787048339843,
"step": 42740
},
{
"epoch": 5.47168,
"grad_norm": 0.4959277808666229,
"learning_rate": 0.00012840305008418,
"loss": 1.9861623764038085,
"step": 42750
},
{
"epoch": 5.47296,
"grad_norm": 0.6514643430709839,
"learning_rate": 0.0001283432123830695,
"loss": 1.9905845642089843,
"step": 42760
},
{
"epoch": 5.47424,
"grad_norm": 0.5936971306800842,
"learning_rate": 0.000128283378201487,
"loss": 1.981403923034668,
"step": 42770
},
{
"epoch": 5.47552,
"grad_norm": 0.5611435174942017,
"learning_rate": 0.00012822354754915647,
"loss": 1.9822219848632812,
"step": 42780
},
{
"epoch": 5.4768,
"grad_norm": 0.6029632687568665,
"learning_rate": 0.00012816372043580106,
"loss": 1.9772762298583983,
"step": 42790
},
{
"epoch": 5.47808,
"grad_norm": 0.5702844858169556,
"learning_rate": 0.00012810389687114359,
"loss": 1.9794082641601562,
"step": 42800
},
{
"epoch": 5.47936,
"grad_norm": 0.5267938375473022,
"learning_rate": 0.0001280440768649062,
"loss": 1.977566909790039,
"step": 42810
},
{
"epoch": 5.48064,
"grad_norm": 0.48588624596595764,
"learning_rate": 0.0001279842604268105,
"loss": 1.9959896087646485,
"step": 42820
},
{
"epoch": 5.48192,
"grad_norm": 0.5755043029785156,
"learning_rate": 0.00012792444756657743,
"loss": 1.9952581405639649,
"step": 42830
},
{
"epoch": 5.4832,
"grad_norm": 0.5789010524749756,
"learning_rate": 0.00012786463829392747,
"loss": 1.981414794921875,
"step": 42840
},
{
"epoch": 5.48448,
"grad_norm": 0.5904437303543091,
"learning_rate": 0.00012780483261858047,
"loss": 2.0018661499023436,
"step": 42850
},
{
"epoch": 5.48576,
"grad_norm": 0.5201811790466309,
"learning_rate": 0.00012774503055025557,
"loss": 1.9836355209350587,
"step": 42860
},
{
"epoch": 5.48704,
"grad_norm": 0.5404776930809021,
"learning_rate": 0.0001276852320986715,
"loss": 1.984324073791504,
"step": 42870
},
{
"epoch": 5.48832,
"grad_norm": 0.5876849889755249,
"learning_rate": 0.0001276254372735464,
"loss": 1.9835599899291991,
"step": 42880
},
{
"epoch": 5.4896,
"grad_norm": 0.5391837358474731,
"learning_rate": 0.00012756564608459766,
"loss": 2.0009517669677734,
"step": 42890
},
{
"epoch": 5.49088,
"grad_norm": 0.548099160194397,
"learning_rate": 0.0001275058585415422,
"loss": 1.9938699722290039,
"step": 42900
},
{
"epoch": 5.49216,
"grad_norm": 0.5481283664703369,
"learning_rate": 0.00012744607465409633,
"loss": 1.9715789794921874,
"step": 42910
},
{
"epoch": 5.49344,
"grad_norm": 0.555843710899353,
"learning_rate": 0.00012738629443197585,
"loss": 1.9791458129882813,
"step": 42920
},
{
"epoch": 5.49472,
"grad_norm": 0.613559901714325,
"learning_rate": 0.0001273265178848957,
"loss": 1.9916711807250977,
"step": 42930
},
{
"epoch": 5.496,
"grad_norm": 0.588998556137085,
"learning_rate": 0.0001272667450225705,
"loss": 1.9913293838500976,
"step": 42940
},
{
"epoch": 5.49728,
"grad_norm": 0.5430281758308411,
"learning_rate": 0.00012720697585471424,
"loss": 1.9878467559814452,
"step": 42950
},
{
"epoch": 5.49856,
"grad_norm": 0.5223708152770996,
"learning_rate": 0.0001271472103910401,
"loss": 1.990936279296875,
"step": 42960
},
{
"epoch": 5.49984,
"grad_norm": 0.6238298416137695,
"learning_rate": 0.00012708744864126084,
"loss": 1.9727264404296876,
"step": 42970
},
{
"epoch": 5.50112,
"grad_norm": 0.5289245247840881,
"learning_rate": 0.00012702769061508862,
"loss": 1.9807937622070313,
"step": 42980
},
{
"epoch": 5.5024,
"grad_norm": 0.6004960536956787,
"learning_rate": 0.00012696793632223488,
"loss": 1.9944551467895508,
"step": 42990
},
{
"epoch": 5.50368,
"grad_norm": 0.49687081575393677,
"learning_rate": 0.00012690818577241051,
"loss": 1.960506057739258,
"step": 43000
},
{
"epoch": 5.50496,
"grad_norm": 0.5790823101997375,
"learning_rate": 0.00012684843897532594,
"loss": 1.9902423858642577,
"step": 43010
},
{
"epoch": 5.50624,
"grad_norm": 0.5271921157836914,
"learning_rate": 0.00012678869594069073,
"loss": 1.974283790588379,
"step": 43020
},
{
"epoch": 5.5075199999999995,
"grad_norm": 0.5904396772384644,
"learning_rate": 0.0001267289566782139,
"loss": 1.991844367980957,
"step": 43030
},
{
"epoch": 5.5088,
"grad_norm": 0.5598679780960083,
"learning_rate": 0.00012666922119760403,
"loss": 1.9727169036865235,
"step": 43040
},
{
"epoch": 5.51008,
"grad_norm": 0.549584686756134,
"learning_rate": 0.00012660948950856898,
"loss": 1.9765899658203125,
"step": 43050
},
{
"epoch": 5.51136,
"grad_norm": 0.60740065574646,
"learning_rate": 0.0001265497616208158,
"loss": 1.9855976104736328,
"step": 43060
},
{
"epoch": 5.51264,
"grad_norm": 0.5468126535415649,
"learning_rate": 0.00012649003754405127,
"loss": 1.978692626953125,
"step": 43070
},
{
"epoch": 5.51392,
"grad_norm": 0.5490145087242126,
"learning_rate": 0.00012643031728798136,
"loss": 1.9990509033203125,
"step": 43080
},
{
"epoch": 5.5152,
"grad_norm": 0.5293770432472229,
"learning_rate": 0.00012637060086231137,
"loss": 1.9928163528442382,
"step": 43090
},
{
"epoch": 5.51648,
"grad_norm": 0.5895763039588928,
"learning_rate": 0.00012631088827674605,
"loss": 1.985493278503418,
"step": 43100
},
{
"epoch": 5.51776,
"grad_norm": 0.5370447635650635,
"learning_rate": 0.0001262511795409896,
"loss": 2.0149173736572266,
"step": 43110
},
{
"epoch": 5.51904,
"grad_norm": 0.5539367198944092,
"learning_rate": 0.00012619147466474541,
"loss": 1.9680770874023437,
"step": 43120
},
{
"epoch": 5.52032,
"grad_norm": 0.6116985082626343,
"learning_rate": 0.00012613177365771645,
"loss": 2.0006887435913088,
"step": 43130
},
{
"epoch": 5.5216,
"grad_norm": 0.5365800857543945,
"learning_rate": 0.00012607207652960494,
"loss": 2.0009613037109375,
"step": 43140
},
{
"epoch": 5.52288,
"grad_norm": 0.5279664397239685,
"learning_rate": 0.0001260123832901124,
"loss": 1.9851922988891602,
"step": 43150
},
{
"epoch": 5.52416,
"grad_norm": 0.6240684390068054,
"learning_rate": 0.0001259526939489399,
"loss": 1.9764827728271483,
"step": 43160
},
{
"epoch": 5.52544,
"grad_norm": 0.5508872866630554,
"learning_rate": 0.00012589300851578776,
"loss": 1.9774866104125977,
"step": 43170
},
{
"epoch": 5.52672,
"grad_norm": 0.5111578702926636,
"learning_rate": 0.00012583332700035572,
"loss": 1.9838226318359375,
"step": 43180
},
{
"epoch": 5.5280000000000005,
"grad_norm": 0.5133348703384399,
"learning_rate": 0.0001257736494123428,
"loss": 1.9747491836547852,
"step": 43190
},
{
"epoch": 5.52928,
"grad_norm": 0.48493191599845886,
"learning_rate": 0.00012571397576144745,
"loss": 1.9591146469116212,
"step": 43200
},
{
"epoch": 5.53056,
"grad_norm": 0.5083683729171753,
"learning_rate": 0.00012565430605736752,
"loss": 1.985081672668457,
"step": 43210
},
{
"epoch": 5.53184,
"grad_norm": 0.5935429930686951,
"learning_rate": 0.00012559464030980006,
"loss": 1.9805543899536133,
"step": 43220
},
{
"epoch": 5.53312,
"grad_norm": 0.5425887107849121,
"learning_rate": 0.00012553497852844166,
"loss": 1.9806880950927734,
"step": 43230
},
{
"epoch": 5.5344,
"grad_norm": 0.5577046275138855,
"learning_rate": 0.00012547532072298815,
"loss": 1.9826868057250977,
"step": 43240
},
{
"epoch": 5.53568,
"grad_norm": 0.5794188380241394,
"learning_rate": 0.0001254156669031348,
"loss": 1.982406234741211,
"step": 43250
},
{
"epoch": 5.53696,
"grad_norm": 0.5767436027526855,
"learning_rate": 0.0001253560170785761,
"loss": 1.985041618347168,
"step": 43260
},
{
"epoch": 5.53824,
"grad_norm": 0.5795595049858093,
"learning_rate": 0.00012529637125900602,
"loss": 1.9792238235473634,
"step": 43270
},
{
"epoch": 5.5395199999999996,
"grad_norm": 0.5892584919929504,
"learning_rate": 0.00012523672945411786,
"loss": 1.9816059112548827,
"step": 43280
},
{
"epoch": 5.5408,
"grad_norm": 0.594740629196167,
"learning_rate": 0.00012517709167360423,
"loss": 1.9862995147705078,
"step": 43290
},
{
"epoch": 5.54208,
"grad_norm": 0.6121344566345215,
"learning_rate": 0.000125117457927157,
"loss": 1.9872638702392578,
"step": 43300
},
{
"epoch": 5.54336,
"grad_norm": 0.5967824459075928,
"learning_rate": 0.00012505782822446764,
"loss": 1.9846179962158204,
"step": 43310
},
{
"epoch": 5.54464,
"grad_norm": 0.5640577077865601,
"learning_rate": 0.00012499820257522667,
"loss": 1.9854454040527343,
"step": 43320
},
{
"epoch": 5.54592,
"grad_norm": 0.5184594392776489,
"learning_rate": 0.0001249385809891241,
"loss": 1.9860681533813476,
"step": 43330
},
{
"epoch": 5.5472,
"grad_norm": 0.5958163738250732,
"learning_rate": 0.00012487896347584937,
"loss": 1.9844581604003906,
"step": 43340
},
{
"epoch": 5.54848,
"grad_norm": 0.5555460453033447,
"learning_rate": 0.000124819350045091,
"loss": 1.9790166854858398,
"step": 43350
},
{
"epoch": 5.54976,
"grad_norm": 0.5916967988014221,
"learning_rate": 0.00012475974070653712,
"loss": 1.9919532775878905,
"step": 43360
},
{
"epoch": 5.55104,
"grad_norm": 0.5260958671569824,
"learning_rate": 0.00012470013546987509,
"loss": 1.96513671875,
"step": 43370
},
{
"epoch": 5.55232,
"grad_norm": 0.538943350315094,
"learning_rate": 0.00012464053434479142,
"loss": 1.9787694931030273,
"step": 43380
},
{
"epoch": 5.5536,
"grad_norm": 0.6128352284431458,
"learning_rate": 0.00012458093734097223,
"loss": 1.9958230972290039,
"step": 43390
},
{
"epoch": 5.55488,
"grad_norm": 0.6461018919944763,
"learning_rate": 0.0001245213444681029,
"loss": 1.984263038635254,
"step": 43400
},
{
"epoch": 5.55616,
"grad_norm": 0.6027620434761047,
"learning_rate": 0.00012446175573586804,
"loss": 1.9954885482788085,
"step": 43410
},
{
"epoch": 5.55744,
"grad_norm": 0.593879222869873,
"learning_rate": 0.00012440217115395165,
"loss": 2.005280303955078,
"step": 43420
},
{
"epoch": 5.55872,
"grad_norm": 0.5401463508605957,
"learning_rate": 0.00012434259073203704,
"loss": 2.0000343322753906,
"step": 43430
},
{
"epoch": 5.5600000000000005,
"grad_norm": 0.6177568435668945,
"learning_rate": 0.0001242830144798069,
"loss": 2.0004384994506834,
"step": 43440
},
{
"epoch": 5.56128,
"grad_norm": 0.5807085633277893,
"learning_rate": 0.00012422344240694313,
"loss": 1.9914360046386719,
"step": 43450
},
{
"epoch": 5.5625599999999995,
"grad_norm": 0.5654975175857544,
"learning_rate": 0.0001241638745231271,
"loss": 1.9913894653320312,
"step": 43460
},
{
"epoch": 5.56384,
"grad_norm": 0.5794891715049744,
"learning_rate": 0.00012410431083803933,
"loss": 1.976278305053711,
"step": 43470
},
{
"epoch": 5.56512,
"grad_norm": 0.4785745441913605,
"learning_rate": 0.0001240447513613598,
"loss": 1.9747161865234375,
"step": 43480
},
{
"epoch": 5.5664,
"grad_norm": 0.5875653028488159,
"learning_rate": 0.0001239851961027677,
"loss": 1.9984664916992188,
"step": 43490
},
{
"epoch": 5.56768,
"grad_norm": 0.5351356267929077,
"learning_rate": 0.00012392564507194165,
"loss": 1.9875604629516601,
"step": 43500
},
{
"epoch": 5.56896,
"grad_norm": 0.561576783657074,
"learning_rate": 0.00012386609827855945,
"loss": 1.9731199264526367,
"step": 43510
},
{
"epoch": 5.57024,
"grad_norm": 0.6075373291969299,
"learning_rate": 0.00012380655573229836,
"loss": 1.9910007476806642,
"step": 43520
},
{
"epoch": 5.57152,
"grad_norm": 0.5152136087417603,
"learning_rate": 0.00012374701744283475,
"loss": 1.9782018661499023,
"step": 43530
},
{
"epoch": 5.5728,
"grad_norm": 0.5507935881614685,
"learning_rate": 0.00012368748341984458,
"loss": 1.9851539611816407,
"step": 43540
},
{
"epoch": 5.57408,
"grad_norm": 0.577321469783783,
"learning_rate": 0.0001236279536730028,
"loss": 1.983145523071289,
"step": 43550
},
{
"epoch": 5.57536,
"grad_norm": 0.6597135066986084,
"learning_rate": 0.00012356842821198386,
"loss": 1.9956960678100586,
"step": 43560
},
{
"epoch": 5.57664,
"grad_norm": 0.5625806450843811,
"learning_rate": 0.00012350890704646158,
"loss": 1.9897554397583008,
"step": 43570
},
{
"epoch": 5.57792,
"grad_norm": 0.6256997585296631,
"learning_rate": 0.00012344939018610882,
"loss": 1.999744987487793,
"step": 43580
},
{
"epoch": 5.5792,
"grad_norm": 0.5985603332519531,
"learning_rate": 0.00012338987764059795,
"loss": 1.9878057479858398,
"step": 43590
},
{
"epoch": 5.58048,
"grad_norm": 0.5251756906509399,
"learning_rate": 0.00012333036941960068,
"loss": 1.986051368713379,
"step": 43600
},
{
"epoch": 5.58176,
"grad_norm": 0.6119303703308105,
"learning_rate": 0.00012327086553278775,
"loss": 1.993195152282715,
"step": 43610
},
{
"epoch": 5.5830400000000004,
"grad_norm": 0.49255499243736267,
"learning_rate": 0.00012321136598982945,
"loss": 1.969487190246582,
"step": 43620
},
{
"epoch": 5.58432,
"grad_norm": 0.5893344283103943,
"learning_rate": 0.00012315187080039533,
"loss": 1.9847057342529297,
"step": 43630
},
{
"epoch": 5.5856,
"grad_norm": 0.5411562323570251,
"learning_rate": 0.0001230923799741541,
"loss": 1.9762943267822266,
"step": 43640
},
{
"epoch": 5.58688,
"grad_norm": 0.5760587453842163,
"learning_rate": 0.00012303289352077386,
"loss": 1.9896160125732423,
"step": 43650
},
{
"epoch": 5.58816,
"grad_norm": 0.5041560530662537,
"learning_rate": 0.000122973411449922,
"loss": 1.9900367736816407,
"step": 43660
},
{
"epoch": 5.58944,
"grad_norm": 0.5801728963851929,
"learning_rate": 0.00012291393377126525,
"loss": 1.9793540954589843,
"step": 43670
},
{
"epoch": 5.59072,
"grad_norm": 0.5786896347999573,
"learning_rate": 0.0001228544604944694,
"loss": 1.9897022247314453,
"step": 43680
},
{
"epoch": 5.592,
"grad_norm": 0.711625874042511,
"learning_rate": 0.0001227949916291998,
"loss": 1.9556003570556642,
"step": 43690
},
{
"epoch": 5.59328,
"grad_norm": 0.5756287574768066,
"learning_rate": 0.00012273552718512098,
"loss": 1.9835685729980468,
"step": 43700
},
{
"epoch": 5.5945599999999995,
"grad_norm": 0.48640769720077515,
"learning_rate": 0.0001226760671718966,
"loss": 1.9591850280761718,
"step": 43710
},
{
"epoch": 5.59584,
"grad_norm": 0.5523165464401245,
"learning_rate": 0.00012261661159918984,
"loss": 1.9997575759887696,
"step": 43720
},
{
"epoch": 5.59712,
"grad_norm": 0.503327488899231,
"learning_rate": 0.0001225571604766631,
"loss": 1.991920280456543,
"step": 43730
},
{
"epoch": 5.5984,
"grad_norm": 0.5423598289489746,
"learning_rate": 0.0001224977138139779,
"loss": 1.9877286911010743,
"step": 43740
},
{
"epoch": 5.59968,
"grad_norm": 0.6281557679176331,
"learning_rate": 0.00012243827162079521,
"loss": 1.9730823516845704,
"step": 43750
},
{
"epoch": 5.60096,
"grad_norm": 0.6227759122848511,
"learning_rate": 0.00012237883390677515,
"loss": 1.974014663696289,
"step": 43760
},
{
"epoch": 5.60224,
"grad_norm": 0.5385846495628357,
"learning_rate": 0.0001223194006815773,
"loss": 1.9909061431884765,
"step": 43770
},
{
"epoch": 5.60352,
"grad_norm": 0.5344575643539429,
"learning_rate": 0.0001222599719548602,
"loss": 1.9781526565551757,
"step": 43780
},
{
"epoch": 5.6048,
"grad_norm": 0.5791848301887512,
"learning_rate": 0.000122200547736282,
"loss": 1.9783771514892579,
"step": 43790
},
{
"epoch": 5.60608,
"grad_norm": 0.6039301156997681,
"learning_rate": 0.0001221411280354999,
"loss": 1.964688491821289,
"step": 43800
},
{
"epoch": 5.60736,
"grad_norm": 0.5714719295501709,
"learning_rate": 0.00012208171286217043,
"loss": 1.9690006256103516,
"step": 43810
},
{
"epoch": 5.60864,
"grad_norm": 0.6743038296699524,
"learning_rate": 0.00012202230222594934,
"loss": 1.988062858581543,
"step": 43820
},
{
"epoch": 5.60992,
"grad_norm": 0.6070483326911926,
"learning_rate": 0.00012196289613649178,
"loss": 1.9861572265625,
"step": 43830
},
{
"epoch": 5.6112,
"grad_norm": 0.5014244318008423,
"learning_rate": 0.0001219034946034519,
"loss": 1.9987524032592774,
"step": 43840
},
{
"epoch": 5.61248,
"grad_norm": 0.5042184591293335,
"learning_rate": 0.0001218440976364834,
"loss": 1.9657289505004882,
"step": 43850
},
{
"epoch": 5.61376,
"grad_norm": 0.6088996529579163,
"learning_rate": 0.00012178470524523913,
"loss": 1.9824615478515626,
"step": 43860
},
{
"epoch": 5.6150400000000005,
"grad_norm": 0.5464186072349548,
"learning_rate": 0.0001217253174393711,
"loss": 1.9862804412841797,
"step": 43870
},
{
"epoch": 5.61632,
"grad_norm": 0.5586224794387817,
"learning_rate": 0.00012166593422853063,
"loss": 1.9796472549438477,
"step": 43880
},
{
"epoch": 5.6176,
"grad_norm": 0.5465304851531982,
"learning_rate": 0.00012160655562236837,
"loss": 1.9873022079467773,
"step": 43890
},
{
"epoch": 5.61888,
"grad_norm": 0.5460226535797119,
"learning_rate": 0.00012154718163053421,
"loss": 1.9819259643554688,
"step": 43900
},
{
"epoch": 5.62016,
"grad_norm": 0.6158143281936646,
"learning_rate": 0.00012148781226267715,
"loss": 1.9980255126953126,
"step": 43910
},
{
"epoch": 5.62144,
"grad_norm": 0.5176824331283569,
"learning_rate": 0.0001214284475284456,
"loss": 1.9766815185546875,
"step": 43920
},
{
"epoch": 5.62272,
"grad_norm": 0.5180767178535461,
"learning_rate": 0.00012136908743748714,
"loss": 1.9696922302246094,
"step": 43930
},
{
"epoch": 5.624,
"grad_norm": 0.5925465226173401,
"learning_rate": 0.00012130973199944855,
"loss": 1.967524528503418,
"step": 43940
},
{
"epoch": 5.62528,
"grad_norm": 0.6556721329689026,
"learning_rate": 0.00012125038122397595,
"loss": 1.9877590179443358,
"step": 43950
},
{
"epoch": 5.62656,
"grad_norm": 0.5563145279884338,
"learning_rate": 0.00012119103512071474,
"loss": 1.9909889221191406,
"step": 43960
},
{
"epoch": 5.62784,
"grad_norm": 0.5005587339401245,
"learning_rate": 0.00012113169369930934,
"loss": 1.9655153274536132,
"step": 43970
},
{
"epoch": 5.62912,
"grad_norm": 0.5706337094306946,
"learning_rate": 0.00012107235696940366,
"loss": 1.9829479217529298,
"step": 43980
},
{
"epoch": 5.6304,
"grad_norm": 0.5546872615814209,
"learning_rate": 0.00012101302494064075,
"loss": 1.9848451614379883,
"step": 43990
},
{
"epoch": 5.63168,
"grad_norm": 0.574387788772583,
"learning_rate": 0.00012095369762266278,
"loss": 1.9744913101196289,
"step": 44000
},
{
"epoch": 5.63296,
"grad_norm": 0.5326042771339417,
"learning_rate": 0.00012089437502511132,
"loss": 1.9825868606567383,
"step": 44010
},
{
"epoch": 5.63424,
"grad_norm": 0.5653786659240723,
"learning_rate": 0.00012083505715762715,
"loss": 1.9967487335205079,
"step": 44020
},
{
"epoch": 5.63552,
"grad_norm": 0.6153043508529663,
"learning_rate": 0.0001207757440298502,
"loss": 1.977321243286133,
"step": 44030
},
{
"epoch": 5.6368,
"grad_norm": 0.5117276906967163,
"learning_rate": 0.00012071643565141973,
"loss": 1.9789136886596679,
"step": 44040
},
{
"epoch": 5.63808,
"grad_norm": 0.6473629474639893,
"learning_rate": 0.00012065713203197405,
"loss": 1.9867519378662108,
"step": 44050
},
{
"epoch": 5.63936,
"grad_norm": 0.5396027565002441,
"learning_rate": 0.000120597833181151,
"loss": 1.9922346115112304,
"step": 44060
},
{
"epoch": 5.64064,
"grad_norm": 0.5477069616317749,
"learning_rate": 0.0001205385391085873,
"loss": 1.9754367828369142,
"step": 44070
},
{
"epoch": 5.64192,
"grad_norm": 0.6671707034111023,
"learning_rate": 0.0001204792498239191,
"loss": 1.9924203872680664,
"step": 44080
},
{
"epoch": 5.6432,
"grad_norm": 0.5158919095993042,
"learning_rate": 0.00012041996533678185,
"loss": 1.9612640380859374,
"step": 44090
},
{
"epoch": 5.64448,
"grad_norm": 0.5805119872093201,
"learning_rate": 0.00012036068565680995,
"loss": 1.9765321731567382,
"step": 44100
},
{
"epoch": 5.64576,
"grad_norm": 0.5373415350914001,
"learning_rate": 0.00012030141079363721,
"loss": 1.9557493209838868,
"step": 44110
},
{
"epoch": 5.64704,
"grad_norm": 0.6231686472892761,
"learning_rate": 0.00012024214075689662,
"loss": 1.9635099411010741,
"step": 44120
},
{
"epoch": 5.64832,
"grad_norm": 0.5194215178489685,
"learning_rate": 0.00012018287555622044,
"loss": 2.0010929107666016,
"step": 44130
},
{
"epoch": 5.6495999999999995,
"grad_norm": 0.5520660281181335,
"learning_rate": 0.00012012361520124002,
"loss": 1.9929811477661132,
"step": 44140
},
{
"epoch": 5.65088,
"grad_norm": 0.5442034602165222,
"learning_rate": 0.000120064359701586,
"loss": 1.970256233215332,
"step": 44150
},
{
"epoch": 5.65216,
"grad_norm": 0.546295702457428,
"learning_rate": 0.00012000510906688828,
"loss": 1.982122039794922,
"step": 44160
},
{
"epoch": 5.65344,
"grad_norm": 0.523432195186615,
"learning_rate": 0.00011994586330677579,
"loss": 1.9862810134887696,
"step": 44170
},
{
"epoch": 5.65472,
"grad_norm": 0.5411070585250854,
"learning_rate": 0.00011988662243087684,
"loss": 1.9872026443481445,
"step": 44180
},
{
"epoch": 5.656,
"grad_norm": 0.5371291637420654,
"learning_rate": 0.00011982738644881898,
"loss": 1.9838180541992188,
"step": 44190
},
{
"epoch": 5.65728,
"grad_norm": 0.5135741233825684,
"learning_rate": 0.00011976815537022872,
"loss": 1.9905227661132812,
"step": 44200
},
{
"epoch": 5.65856,
"grad_norm": 0.6010394096374512,
"learning_rate": 0.00011970892920473208,
"loss": 2.001226615905762,
"step": 44210
},
{
"epoch": 5.65984,
"grad_norm": 0.4963909089565277,
"learning_rate": 0.0001196497079619541,
"loss": 1.9765230178833009,
"step": 44220
},
{
"epoch": 5.66112,
"grad_norm": 0.6319615244865417,
"learning_rate": 0.00011959049165151892,
"loss": 1.9928529739379883,
"step": 44230
},
{
"epoch": 5.6624,
"grad_norm": 0.5118665099143982,
"learning_rate": 0.00011953128028305012,
"loss": 1.984581756591797,
"step": 44240
},
{
"epoch": 5.66368,
"grad_norm": 0.4987267553806305,
"learning_rate": 0.0001194720738661704,
"loss": 1.971200942993164,
"step": 44250
},
{
"epoch": 5.66496,
"grad_norm": 0.53094083070755,
"learning_rate": 0.00011941287241050158,
"loss": 1.9800697326660157,
"step": 44260
},
{
"epoch": 5.66624,
"grad_norm": 0.5714093446731567,
"learning_rate": 0.0001193536759256647,
"loss": 1.9713449478149414,
"step": 44270
},
{
"epoch": 5.66752,
"grad_norm": 0.6253819465637207,
"learning_rate": 0.00011929448442128003,
"loss": 1.9456159591674804,
"step": 44280
},
{
"epoch": 5.6688,
"grad_norm": 0.5579882264137268,
"learning_rate": 0.00011923529790696703,
"loss": 2.016672897338867,
"step": 44290
},
{
"epoch": 5.6700800000000005,
"grad_norm": 0.5571644902229309,
"learning_rate": 0.00011917611639234426,
"loss": 1.9700908660888672,
"step": 44300
},
{
"epoch": 5.67136,
"grad_norm": 0.5862537026405334,
"learning_rate": 0.0001191169398870296,
"loss": 1.9832935333251953,
"step": 44310
},
{
"epoch": 5.67264,
"grad_norm": 0.5416548848152161,
"learning_rate": 0.00011905776840064004,
"loss": 1.994603157043457,
"step": 44320
},
{
"epoch": 5.67392,
"grad_norm": 0.5965008735656738,
"learning_rate": 0.00011899860194279177,
"loss": 1.9788105010986328,
"step": 44330
},
{
"epoch": 5.6752,
"grad_norm": 0.6559287309646606,
"learning_rate": 0.0001189394405231001,
"loss": 1.979328727722168,
"step": 44340
},
{
"epoch": 5.67648,
"grad_norm": 0.5698477029800415,
"learning_rate": 0.00011888028415117969,
"loss": 1.9944160461425782,
"step": 44350
},
{
"epoch": 5.67776,
"grad_norm": 0.49233806133270264,
"learning_rate": 0.00011882113283664416,
"loss": 1.9748605728149413,
"step": 44360
},
{
"epoch": 5.67904,
"grad_norm": 0.5298371911048889,
"learning_rate": 0.00011876198658910647,
"loss": 1.980626678466797,
"step": 44370
},
{
"epoch": 5.68032,
"grad_norm": 0.5201873183250427,
"learning_rate": 0.00011870284541817868,
"loss": 1.9673849105834962,
"step": 44380
},
{
"epoch": 5.6815999999999995,
"grad_norm": 0.6403574347496033,
"learning_rate": 0.00011864370933347214,
"loss": 1.9894193649291991,
"step": 44390
},
{
"epoch": 5.68288,
"grad_norm": 0.5389891862869263,
"learning_rate": 0.00011858457834459714,
"loss": 1.9877696990966798,
"step": 44400
},
{
"epoch": 5.68416,
"grad_norm": 0.6680673956871033,
"learning_rate": 0.00011852545246116337,
"loss": 1.9791730880737304,
"step": 44410
},
{
"epoch": 5.68544,
"grad_norm": 0.604011595249176,
"learning_rate": 0.00011846633169277963,
"loss": 1.9843769073486328,
"step": 44420
},
{
"epoch": 5.68672,
"grad_norm": 0.5669480562210083,
"learning_rate": 0.00011840721604905382,
"loss": 1.998988723754883,
"step": 44430
},
{
"epoch": 5.688,
"grad_norm": 0.5626456141471863,
"learning_rate": 0.00011834810553959301,
"loss": 1.9924798965454102,
"step": 44440
},
{
"epoch": 5.68928,
"grad_norm": 0.546132504940033,
"learning_rate": 0.00011828900017400358,
"loss": 1.964630126953125,
"step": 44450
},
{
"epoch": 5.69056,
"grad_norm": 0.5781171917915344,
"learning_rate": 0.00011822989996189088,
"loss": 1.9881515502929688,
"step": 44460
},
{
"epoch": 5.69184,
"grad_norm": 0.5588444471359253,
"learning_rate": 0.00011817080491285952,
"loss": 1.9937620162963867,
"step": 44470
},
{
"epoch": 5.69312,
"grad_norm": 0.5440167188644409,
"learning_rate": 0.00011811171503651334,
"loss": 1.997095489501953,
"step": 44480
},
{
"epoch": 5.6944,
"grad_norm": 0.5387884378433228,
"learning_rate": 0.00011805263034245526,
"loss": 1.9904386520385742,
"step": 44490
},
{
"epoch": 5.69568,
"grad_norm": 0.6314904689788818,
"learning_rate": 0.00011799355084028722,
"loss": 2.005698585510254,
"step": 44500
},
{
"epoch": 5.69696,
"grad_norm": 0.6244338750839233,
"learning_rate": 0.00011793447653961057,
"loss": 1.962590789794922,
"step": 44510
},
{
"epoch": 5.69824,
"grad_norm": 0.578115701675415,
"learning_rate": 0.00011787540745002575,
"loss": 1.9888931274414063,
"step": 44520
},
{
"epoch": 5.69952,
"grad_norm": 0.5415886044502258,
"learning_rate": 0.00011781634358113218,
"loss": 1.976822280883789,
"step": 44530
},
{
"epoch": 5.7008,
"grad_norm": 0.5215967893600464,
"learning_rate": 0.00011775728494252864,
"loss": 1.9818981170654297,
"step": 44540
},
{
"epoch": 5.7020800000000005,
"grad_norm": 0.5115883946418762,
"learning_rate": 0.000117698231543813,
"loss": 1.9732782363891601,
"step": 44550
},
{
"epoch": 5.70336,
"grad_norm": 0.5564931035041809,
"learning_rate": 0.00011763918339458214,
"loss": 1.954865074157715,
"step": 44560
},
{
"epoch": 5.70464,
"grad_norm": 0.5260511040687561,
"learning_rate": 0.00011758014050443229,
"loss": 1.9788331985473633,
"step": 44570
},
{
"epoch": 5.70592,
"grad_norm": 0.5807152390480042,
"learning_rate": 0.00011752110288295876,
"loss": 1.9963584899902345,
"step": 44580
},
{
"epoch": 5.7072,
"grad_norm": 0.540276825428009,
"learning_rate": 0.00011746207053975589,
"loss": 1.9891365051269532,
"step": 44590
},
{
"epoch": 5.70848,
"grad_norm": 0.6140125393867493,
"learning_rate": 0.00011740304348441733,
"loss": 2.009307861328125,
"step": 44600
},
{
"epoch": 5.70976,
"grad_norm": 0.541618824005127,
"learning_rate": 0.00011734402172653574,
"loss": 1.9814813613891602,
"step": 44610
},
{
"epoch": 5.71104,
"grad_norm": 0.5606712102890015,
"learning_rate": 0.00011728500527570306,
"loss": 1.9971675872802734,
"step": 44620
},
{
"epoch": 5.71232,
"grad_norm": 0.5125880241394043,
"learning_rate": 0.00011722599414151018,
"loss": 1.9951154708862304,
"step": 44630
},
{
"epoch": 5.7136,
"grad_norm": 0.5207726955413818,
"learning_rate": 0.00011716698833354726,
"loss": 1.9909067153930664,
"step": 44640
},
{
"epoch": 5.71488,
"grad_norm": 0.536038339138031,
"learning_rate": 0.00011710798786140364,
"loss": 1.979075050354004,
"step": 44650
},
{
"epoch": 5.71616,
"grad_norm": 0.5525376200675964,
"learning_rate": 0.00011704899273466762,
"loss": 1.9694101333618164,
"step": 44660
},
{
"epoch": 5.71744,
"grad_norm": 0.5448262095451355,
"learning_rate": 0.0001169900029629267,
"loss": 1.9872264862060547,
"step": 44670
},
{
"epoch": 5.71872,
"grad_norm": 0.5812845826148987,
"learning_rate": 0.00011693101855576765,
"loss": 1.9796199798583984,
"step": 44680
},
{
"epoch": 5.72,
"grad_norm": 0.5144921541213989,
"learning_rate": 0.00011687203952277614,
"loss": 1.973304557800293,
"step": 44690
},
{
"epoch": 5.72128,
"grad_norm": 0.5949316024780273,
"learning_rate": 0.00011681306587353713,
"loss": 1.9806039810180665,
"step": 44700
},
{
"epoch": 5.72256,
"grad_norm": 0.49195539951324463,
"learning_rate": 0.00011675409761763469,
"loss": 1.9881065368652344,
"step": 44710
},
{
"epoch": 5.72384,
"grad_norm": 0.48496535420417786,
"learning_rate": 0.00011669513476465195,
"loss": 1.9931032180786132,
"step": 44720
},
{
"epoch": 5.72512,
"grad_norm": 0.5091038346290588,
"learning_rate": 0.00011663617732417112,
"loss": 1.9788013458251954,
"step": 44730
},
{
"epoch": 5.7264,
"grad_norm": 0.5520978569984436,
"learning_rate": 0.00011657722530577369,
"loss": 1.9473638534545898,
"step": 44740
},
{
"epoch": 5.72768,
"grad_norm": 0.5686102509498596,
"learning_rate": 0.0001165182787190402,
"loss": 2.0024169921875,
"step": 44750
},
{
"epoch": 5.72896,
"grad_norm": 0.5838239789009094,
"learning_rate": 0.0001164593375735502,
"loss": 1.9769811630249023,
"step": 44760
},
{
"epoch": 5.73024,
"grad_norm": 0.5804700255393982,
"learning_rate": 0.00011640040187888248,
"loss": 1.9933982849121095,
"step": 44770
},
{
"epoch": 5.73152,
"grad_norm": 0.5425624847412109,
"learning_rate": 0.00011634147164461497,
"loss": 1.9738555908203126,
"step": 44780
},
{
"epoch": 5.7328,
"grad_norm": 0.5333347916603088,
"learning_rate": 0.00011628254688032451,
"loss": 1.9721805572509765,
"step": 44790
},
{
"epoch": 5.73408,
"grad_norm": 0.5649722814559937,
"learning_rate": 0.00011622362759558726,
"loss": 1.9901483535766602,
"step": 44800
},
{
"epoch": 5.73536,
"grad_norm": 0.5057212710380554,
"learning_rate": 0.00011616471379997852,
"loss": 1.986781120300293,
"step": 44810
},
{
"epoch": 5.7366399999999995,
"grad_norm": 0.5558092594146729,
"learning_rate": 0.00011610580550307243,
"loss": 1.9708751678466796,
"step": 44820
},
{
"epoch": 5.73792,
"grad_norm": 0.5248475074768066,
"learning_rate": 0.00011604690271444254,
"loss": 2.001468467712402,
"step": 44830
},
{
"epoch": 5.7392,
"grad_norm": 0.506753146648407,
"learning_rate": 0.00011598800544366131,
"loss": 1.9980134963989258,
"step": 44840
},
{
"epoch": 5.74048,
"grad_norm": 0.56104975938797,
"learning_rate": 0.00011592911370030031,
"loss": 1.9912473678588867,
"step": 44850
},
{
"epoch": 5.74176,
"grad_norm": 0.5781740546226501,
"learning_rate": 0.00011587022749393031,
"loss": 1.9731857299804687,
"step": 44860
},
{
"epoch": 5.74304,
"grad_norm": 0.6047148108482361,
"learning_rate": 0.00011581134683412115,
"loss": 1.9815412521362306,
"step": 44870
},
{
"epoch": 5.74432,
"grad_norm": 0.5581022500991821,
"learning_rate": 0.00011575247173044182,
"loss": 1.9787160873413085,
"step": 44880
},
{
"epoch": 5.7456,
"grad_norm": 0.4806927740573883,
"learning_rate": 0.00011569360219246022,
"loss": 1.9954851150512696,
"step": 44890
},
{
"epoch": 5.74688,
"grad_norm": 0.6509290933609009,
"learning_rate": 0.00011563473822974349,
"loss": 1.9996822357177735,
"step": 44900
},
{
"epoch": 5.74816,
"grad_norm": 0.587274432182312,
"learning_rate": 0.00011557587985185792,
"loss": 1.9959234237670898,
"step": 44910
},
{
"epoch": 5.74944,
"grad_norm": 0.5513741970062256,
"learning_rate": 0.00011551702706836869,
"loss": 1.9986631393432617,
"step": 44920
},
{
"epoch": 5.75072,
"grad_norm": 0.5147374868392944,
"learning_rate": 0.00011545817988884024,
"loss": 1.9679647445678712,
"step": 44930
},
{
"epoch": 5.752,
"grad_norm": 0.5074622631072998,
"learning_rate": 0.00011539933832283615,
"loss": 1.9909784317016601,
"step": 44940
},
{
"epoch": 5.75328,
"grad_norm": 0.6587070226669312,
"learning_rate": 0.00011534050237991884,
"loss": 1.9872098922729493,
"step": 44950
},
{
"epoch": 5.75456,
"grad_norm": 0.5896905064582825,
"learning_rate": 0.00011528167206965002,
"loss": 1.9892034530639648,
"step": 44960
},
{
"epoch": 5.75584,
"grad_norm": 0.5050958395004272,
"learning_rate": 0.00011522284740159046,
"loss": 1.9833921432495116,
"step": 44970
},
{
"epoch": 5.7571200000000005,
"grad_norm": 0.5492570400238037,
"learning_rate": 0.0001151640283853,
"loss": 1.9804567337036132,
"step": 44980
},
{
"epoch": 5.7584,
"grad_norm": 0.5212603807449341,
"learning_rate": 0.00011510521503033747,
"loss": 1.9968971252441405,
"step": 44990
},
{
"epoch": 5.75968,
"grad_norm": 0.5957831144332886,
"learning_rate": 0.00011504640734626093,
"loss": 1.9785430908203125,
"step": 45000
},
{
"epoch": 5.76096,
"grad_norm": 0.5273776054382324,
"learning_rate": 0.00011498760534262743,
"loss": 1.9616100311279296,
"step": 45010
},
{
"epoch": 5.76224,
"grad_norm": 0.5860339999198914,
"learning_rate": 0.000114928809028993,
"loss": 1.9617446899414062,
"step": 45020
},
{
"epoch": 5.76352,
"grad_norm": 0.6032606363296509,
"learning_rate": 0.00011487001841491298,
"loss": 1.9787763595581054,
"step": 45030
},
{
"epoch": 5.7648,
"grad_norm": 0.5983586311340332,
"learning_rate": 0.00011481123350994166,
"loss": 1.9550310134887696,
"step": 45040
},
{
"epoch": 5.76608,
"grad_norm": 0.5536155700683594,
"learning_rate": 0.00011475245432363232,
"loss": 1.982370948791504,
"step": 45050
},
{
"epoch": 5.76736,
"grad_norm": 0.5272994637489319,
"learning_rate": 0.00011469368086553747,
"loss": 1.9925317764282227,
"step": 45060
},
{
"epoch": 5.7686399999999995,
"grad_norm": 0.584190845489502,
"learning_rate": 0.00011463491314520859,
"loss": 2.001248359680176,
"step": 45070
},
{
"epoch": 5.76992,
"grad_norm": 0.551550030708313,
"learning_rate": 0.00011457615117219619,
"loss": 1.980942153930664,
"step": 45080
},
{
"epoch": 5.7712,
"grad_norm": 0.5417651534080505,
"learning_rate": 0.00011451739495604995,
"loss": 1.990875244140625,
"step": 45090
},
{
"epoch": 5.77248,
"grad_norm": 0.4959915578365326,
"learning_rate": 0.0001144586445063186,
"loss": 1.991243553161621,
"step": 45100
},
{
"epoch": 5.77376,
"grad_norm": 0.5780417919158936,
"learning_rate": 0.00011439989983254988,
"loss": 1.9733209609985352,
"step": 45110
},
{
"epoch": 5.77504,
"grad_norm": 0.6107548475265503,
"learning_rate": 0.00011434116094429059,
"loss": 1.9845096588134765,
"step": 45120
},
{
"epoch": 5.77632,
"grad_norm": 0.5434926748275757,
"learning_rate": 0.00011428242785108662,
"loss": 1.9745063781738281,
"step": 45130
},
{
"epoch": 5.7776,
"grad_norm": 0.5189668536186218,
"learning_rate": 0.00011422370056248296,
"loss": 1.9786745071411134,
"step": 45140
},
{
"epoch": 5.77888,
"grad_norm": 0.4956158399581909,
"learning_rate": 0.00011416497908802352,
"loss": 1.9727724075317383,
"step": 45150
},
{
"epoch": 5.78016,
"grad_norm": 0.5454486012458801,
"learning_rate": 0.00011410626343725142,
"loss": 1.98092041015625,
"step": 45160
},
{
"epoch": 5.78144,
"grad_norm": 0.5586042404174805,
"learning_rate": 0.00011404755361970875,
"loss": 1.9803630828857421,
"step": 45170
},
{
"epoch": 5.78272,
"grad_norm": 0.4740152657032013,
"learning_rate": 0.00011398884964493669,
"loss": 1.9866451263427733,
"step": 45180
},
{
"epoch": 5.784,
"grad_norm": 0.5406758189201355,
"learning_rate": 0.00011393015152247537,
"loss": 1.9687381744384767,
"step": 45190
},
{
"epoch": 5.78528,
"grad_norm": 0.5535482168197632,
"learning_rate": 0.00011387145926186417,
"loss": 1.9828323364257812,
"step": 45200
},
{
"epoch": 5.78656,
"grad_norm": 0.641049325466156,
"learning_rate": 0.00011381277287264127,
"loss": 1.972235870361328,
"step": 45210
},
{
"epoch": 5.78784,
"grad_norm": 0.545458972454071,
"learning_rate": 0.0001137540923643441,
"loss": 1.9925958633422851,
"step": 45220
},
{
"epoch": 5.7891200000000005,
"grad_norm": 0.5368591547012329,
"learning_rate": 0.000113695417746509,
"loss": 1.9777208328247071,
"step": 45230
},
{
"epoch": 5.7904,
"grad_norm": 0.5226603150367737,
"learning_rate": 0.00011363674902867152,
"loss": 1.9790481567382812,
"step": 45240
},
{
"epoch": 5.79168,
"grad_norm": 0.5600732564926147,
"learning_rate": 0.000113578086220366,
"loss": 1.9725765228271483,
"step": 45250
},
{
"epoch": 5.79296,
"grad_norm": 0.5785577297210693,
"learning_rate": 0.00011351942933112602,
"loss": 1.984701156616211,
"step": 45260
},
{
"epoch": 5.79424,
"grad_norm": 0.5912672281265259,
"learning_rate": 0.00011346077837048421,
"loss": 1.9633541107177734,
"step": 45270
},
{
"epoch": 5.79552,
"grad_norm": 0.5617584586143494,
"learning_rate": 0.00011340213334797206,
"loss": 2.0101083755493163,
"step": 45280
},
{
"epoch": 5.7968,
"grad_norm": 0.6158162951469421,
"learning_rate": 0.0001133434942731202,
"loss": 1.9858604431152345,
"step": 45290
},
{
"epoch": 5.79808,
"grad_norm": 0.5250460505485535,
"learning_rate": 0.00011328486115545838,
"loss": 1.970917320251465,
"step": 45300
},
{
"epoch": 5.79936,
"grad_norm": 0.5678423047065735,
"learning_rate": 0.00011322623400451518,
"loss": 1.973000717163086,
"step": 45310
},
{
"epoch": 5.80064,
"grad_norm": 0.6280449628829956,
"learning_rate": 0.00011316761282981837,
"loss": 1.9768505096435547,
"step": 45320
},
{
"epoch": 5.80192,
"grad_norm": 0.5354257822036743,
"learning_rate": 0.00011310899764089476,
"loss": 1.9900197982788086,
"step": 45330
},
{
"epoch": 5.8032,
"grad_norm": 0.6120757460594177,
"learning_rate": 0.00011305038844727007,
"loss": 1.9989439010620118,
"step": 45340
},
{
"epoch": 5.80448,
"grad_norm": 0.5177772045135498,
"learning_rate": 0.00011299178525846912,
"loss": 2.000916862487793,
"step": 45350
},
{
"epoch": 5.80576,
"grad_norm": 0.5001358389854431,
"learning_rate": 0.00011293318808401573,
"loss": 1.9918975830078125,
"step": 45360
},
{
"epoch": 5.80704,
"grad_norm": 0.5333527326583862,
"learning_rate": 0.00011287459693343281,
"loss": 1.9969669342041017,
"step": 45370
},
{
"epoch": 5.80832,
"grad_norm": 0.4966394007205963,
"learning_rate": 0.00011281601181624213,
"loss": 1.9774642944335938,
"step": 45380
},
{
"epoch": 5.8096,
"grad_norm": 0.550881564617157,
"learning_rate": 0.00011275743274196468,
"loss": 1.9735660552978516,
"step": 45390
},
{
"epoch": 5.81088,
"grad_norm": 0.5198240876197815,
"learning_rate": 0.00011269885972012031,
"loss": 1.9867525100708008,
"step": 45400
},
{
"epoch": 5.81216,
"grad_norm": 0.559284508228302,
"learning_rate": 0.00011264029276022799,
"loss": 1.9800422668457032,
"step": 45410
},
{
"epoch": 5.81344,
"grad_norm": 0.6312355995178223,
"learning_rate": 0.00011258173187180558,
"loss": 1.9753734588623046,
"step": 45420
},
{
"epoch": 5.81472,
"grad_norm": 0.5332842469215393,
"learning_rate": 0.00011252317706437018,
"loss": 1.9815406799316406,
"step": 45430
},
{
"epoch": 5.816,
"grad_norm": 0.6148523688316345,
"learning_rate": 0.00011246462834743762,
"loss": 1.9825111389160157,
"step": 45440
},
{
"epoch": 5.81728,
"grad_norm": 0.5562883615493774,
"learning_rate": 0.00011240608573052295,
"loss": 1.9774847030639648,
"step": 45450
},
{
"epoch": 5.81856,
"grad_norm": 0.5304770469665527,
"learning_rate": 0.0001123475492231401,
"loss": 1.9615726470947266,
"step": 45460
},
{
"epoch": 5.81984,
"grad_norm": 0.5264233946800232,
"learning_rate": 0.00011228901883480218,
"loss": 1.9846017837524415,
"step": 45470
},
{
"epoch": 5.82112,
"grad_norm": 0.568549633026123,
"learning_rate": 0.00011223049457502102,
"loss": 1.981773567199707,
"step": 45480
},
{
"epoch": 5.8224,
"grad_norm": 0.5381590127944946,
"learning_rate": 0.00011217197645330774,
"loss": 1.99804744720459,
"step": 45490
},
{
"epoch": 5.8236799999999995,
"grad_norm": 0.5650418400764465,
"learning_rate": 0.00011211346447917237,
"loss": 1.9925737380981445,
"step": 45500
},
{
"epoch": 5.82496,
"grad_norm": 0.5886473655700684,
"learning_rate": 0.00011205495866212385,
"loss": 1.9882450103759766,
"step": 45510
},
{
"epoch": 5.82624,
"grad_norm": 0.5452497005462646,
"learning_rate": 0.00011199645901167016,
"loss": 1.9893831253051757,
"step": 45520
},
{
"epoch": 5.82752,
"grad_norm": 0.6179299354553223,
"learning_rate": 0.0001119379655373184,
"loss": 2.002470779418945,
"step": 45530
},
{
"epoch": 5.8288,
"grad_norm": 0.5729301571846008,
"learning_rate": 0.00011187947824857444,
"loss": 1.992620086669922,
"step": 45540
},
{
"epoch": 5.83008,
"grad_norm": 0.6246586441993713,
"learning_rate": 0.00011182099715494337,
"loss": 1.9860965728759765,
"step": 45550
},
{
"epoch": 5.83136,
"grad_norm": 0.5626221299171448,
"learning_rate": 0.0001117625222659292,
"loss": 1.9712797164916993,
"step": 45560
},
{
"epoch": 5.83264,
"grad_norm": 0.5335031151771545,
"learning_rate": 0.00011170405359103485,
"loss": 1.9836847305297851,
"step": 45570
},
{
"epoch": 5.83392,
"grad_norm": 0.6009292602539062,
"learning_rate": 0.00011164559113976225,
"loss": 1.9907087326049804,
"step": 45580
},
{
"epoch": 5.8352,
"grad_norm": 0.550502598285675,
"learning_rate": 0.00011158713492161246,
"loss": 1.9696290969848633,
"step": 45590
},
{
"epoch": 5.83648,
"grad_norm": 0.533101499080658,
"learning_rate": 0.0001115286849460854,
"loss": 2.0090660095214843,
"step": 45600
},
{
"epoch": 5.83776,
"grad_norm": 0.5244407653808594,
"learning_rate": 0.00011147024122267993,
"loss": 1.9805353164672852,
"step": 45610
},
{
"epoch": 5.83904,
"grad_norm": 0.6165556907653809,
"learning_rate": 0.00011141180376089406,
"loss": 1.9932254791259765,
"step": 45620
},
{
"epoch": 5.84032,
"grad_norm": 0.526007354259491,
"learning_rate": 0.00011135337257022465,
"loss": 1.993150520324707,
"step": 45630
},
{
"epoch": 5.8416,
"grad_norm": 0.5896913409233093,
"learning_rate": 0.00011129494766016752,
"loss": 1.965115737915039,
"step": 45640
},
{
"epoch": 5.84288,
"grad_norm": 0.5420181751251221,
"learning_rate": 0.00011123652904021758,
"loss": 1.980116081237793,
"step": 45650
},
{
"epoch": 5.8441600000000005,
"grad_norm": 0.5326025485992432,
"learning_rate": 0.00011117811671986869,
"loss": 2.0015869140625,
"step": 45660
},
{
"epoch": 5.84544,
"grad_norm": 0.590660572052002,
"learning_rate": 0.00011111971070861361,
"loss": 1.9801996231079102,
"step": 45670
},
{
"epoch": 5.84672,
"grad_norm": 0.5863092541694641,
"learning_rate": 0.00011106131101594417,
"loss": 1.9933433532714844,
"step": 45680
},
{
"epoch": 5.848,
"grad_norm": 0.56316077709198,
"learning_rate": 0.00011100291765135107,
"loss": 1.9733230590820312,
"step": 45690
},
{
"epoch": 5.84928,
"grad_norm": 0.5843523144721985,
"learning_rate": 0.00011094453062432413,
"loss": 1.9890064239501952,
"step": 45700
},
{
"epoch": 5.85056,
"grad_norm": 0.5465300679206848,
"learning_rate": 0.00011088614994435197,
"loss": 1.9929014205932618,
"step": 45710
},
{
"epoch": 5.85184,
"grad_norm": 0.5442835688591003,
"learning_rate": 0.00011082777562092228,
"loss": 1.995138931274414,
"step": 45720
},
{
"epoch": 5.85312,
"grad_norm": 0.5561668276786804,
"learning_rate": 0.00011076940766352175,
"loss": 1.9675748825073243,
"step": 45730
},
{
"epoch": 5.8544,
"grad_norm": 0.5517189502716064,
"learning_rate": 0.00011071104608163591,
"loss": 1.971379280090332,
"step": 45740
},
{
"epoch": 5.8556799999999996,
"grad_norm": 0.6517131924629211,
"learning_rate": 0.00011065269088474932,
"loss": 1.9801050186157227,
"step": 45750
},
{
"epoch": 5.85696,
"grad_norm": 0.5294321179389954,
"learning_rate": 0.00011059434208234561,
"loss": 1.9892789840698242,
"step": 45760
},
{
"epoch": 5.85824,
"grad_norm": 0.5459432005882263,
"learning_rate": 0.00011053599968390713,
"loss": 1.9855180740356446,
"step": 45770
},
{
"epoch": 5.85952,
"grad_norm": 0.5477169752120972,
"learning_rate": 0.00011047766369891539,
"loss": 1.9775909423828124,
"step": 45780
},
{
"epoch": 5.8608,
"grad_norm": 0.531773030757904,
"learning_rate": 0.00011041933413685087,
"loss": 1.9878747940063477,
"step": 45790
},
{
"epoch": 5.86208,
"grad_norm": 0.6014275550842285,
"learning_rate": 0.00011036101100719286,
"loss": 1.9807065963745116,
"step": 45800
},
{
"epoch": 5.86336,
"grad_norm": 0.6035335063934326,
"learning_rate": 0.0001103026943194196,
"loss": 1.983103370666504,
"step": 45810
},
{
"epoch": 5.86464,
"grad_norm": 0.5293020606040955,
"learning_rate": 0.00011024438408300846,
"loss": 1.9830497741699218,
"step": 45820
},
{
"epoch": 5.86592,
"grad_norm": 0.5241860151290894,
"learning_rate": 0.00011018608030743571,
"loss": 1.997563362121582,
"step": 45830
},
{
"epoch": 5.8672,
"grad_norm": 0.5331718325614929,
"learning_rate": 0.0001101277830021764,
"loss": 1.976490592956543,
"step": 45840
},
{
"epoch": 5.86848,
"grad_norm": 0.5664361119270325,
"learning_rate": 0.00011006949217670473,
"loss": 1.9870201110839845,
"step": 45850
},
{
"epoch": 5.86976,
"grad_norm": 0.5292598605155945,
"learning_rate": 0.00011001120784049376,
"loss": 1.988079833984375,
"step": 45860
},
{
"epoch": 5.87104,
"grad_norm": 0.49678730964660645,
"learning_rate": 0.00010995293000301541,
"loss": 1.9711204528808595,
"step": 45870
},
{
"epoch": 5.87232,
"grad_norm": 0.5668963193893433,
"learning_rate": 0.00010989465867374071,
"loss": 1.977288818359375,
"step": 45880
},
{
"epoch": 5.8736,
"grad_norm": 0.5368448495864868,
"learning_rate": 0.00010983639386213965,
"loss": 1.9830488204956054,
"step": 45890
},
{
"epoch": 5.87488,
"grad_norm": 0.5478209257125854,
"learning_rate": 0.00010977813557768088,
"loss": 1.9833507537841797,
"step": 45900
},
{
"epoch": 5.8761600000000005,
"grad_norm": 0.46102043986320496,
"learning_rate": 0.00010971988382983232,
"loss": 1.9810808181762696,
"step": 45910
},
{
"epoch": 5.87744,
"grad_norm": 0.5900238752365112,
"learning_rate": 0.00010966163862806065,
"loss": 1.9943126678466796,
"step": 45920
},
{
"epoch": 5.8787199999999995,
"grad_norm": 0.5182954668998718,
"learning_rate": 0.00010960339998183146,
"loss": 1.9932270050048828,
"step": 45930
},
{
"epoch": 5.88,
"grad_norm": 0.5300226211547852,
"learning_rate": 0.00010954516790060939,
"loss": 1.9844392776489257,
"step": 45940
},
{
"epoch": 5.88128,
"grad_norm": 0.5539748668670654,
"learning_rate": 0.00010948694239385798,
"loss": 1.975348663330078,
"step": 45950
},
{
"epoch": 5.88256,
"grad_norm": 0.5445032119750977,
"learning_rate": 0.00010942872347103966,
"loss": 1.9720674514770509,
"step": 45960
},
{
"epoch": 5.88384,
"grad_norm": 0.5253638625144958,
"learning_rate": 0.00010937051114161581,
"loss": 1.9661537170410157,
"step": 45970
},
{
"epoch": 5.88512,
"grad_norm": 0.5724123120307922,
"learning_rate": 0.00010931230541504671,
"loss": 1.9988731384277343,
"step": 45980
},
{
"epoch": 5.8864,
"grad_norm": 0.6102231740951538,
"learning_rate": 0.0001092541063007917,
"loss": 1.9803321838378907,
"step": 45990
},
{
"epoch": 5.88768,
"grad_norm": 0.5670763254165649,
"learning_rate": 0.0001091959138083088,
"loss": 1.9708581924438477,
"step": 46000
},
{
"epoch": 5.88896,
"grad_norm": 0.511055052280426,
"learning_rate": 0.0001091377279470552,
"loss": 1.9795167922973633,
"step": 46010
},
{
"epoch": 5.89024,
"grad_norm": 0.5518162250518799,
"learning_rate": 0.00010907954872648686,
"loss": 1.984358787536621,
"step": 46020
},
{
"epoch": 5.89152,
"grad_norm": 0.6072937846183777,
"learning_rate": 0.00010902137615605873,
"loss": 1.987832260131836,
"step": 46030
},
{
"epoch": 5.8928,
"grad_norm": 0.5729063153266907,
"learning_rate": 0.00010896321024522463,
"loss": 1.9663116455078125,
"step": 46040
},
{
"epoch": 5.89408,
"grad_norm": 0.4984786808490753,
"learning_rate": 0.0001089050510034374,
"loss": 1.9913915634155273,
"step": 46050
},
{
"epoch": 5.89536,
"grad_norm": 0.5118082761764526,
"learning_rate": 0.00010884689844014861,
"loss": 1.9762439727783203,
"step": 46060
},
{
"epoch": 5.89664,
"grad_norm": 0.5643131136894226,
"learning_rate": 0.00010878875256480892,
"loss": 1.9908561706542969,
"step": 46070
},
{
"epoch": 5.89792,
"grad_norm": 0.4942738711833954,
"learning_rate": 0.00010873061338686787,
"loss": 1.9592121124267579,
"step": 46080
},
{
"epoch": 5.8992,
"grad_norm": 0.5165303349494934,
"learning_rate": 0.00010867248091577388,
"loss": 1.9865238189697265,
"step": 46090
},
{
"epoch": 5.90048,
"grad_norm": 0.5649989247322083,
"learning_rate": 0.0001086143551609742,
"loss": 1.9568763732910157,
"step": 46100
},
{
"epoch": 5.90176,
"grad_norm": 0.490532785654068,
"learning_rate": 0.00010855623613191511,
"loss": 1.9820899963378906,
"step": 46110
},
{
"epoch": 5.90304,
"grad_norm": 0.5794340372085571,
"learning_rate": 0.00010849812383804182,
"loss": 1.9851873397827149,
"step": 46120
},
{
"epoch": 5.90432,
"grad_norm": 0.6528133749961853,
"learning_rate": 0.00010844001828879828,
"loss": 1.9824542999267578,
"step": 46130
},
{
"epoch": 5.9056,
"grad_norm": 0.5524939298629761,
"learning_rate": 0.00010838191949362754,
"loss": 1.9941194534301758,
"step": 46140
},
{
"epoch": 5.90688,
"grad_norm": 0.5567999482154846,
"learning_rate": 0.00010832382746197145,
"loss": 1.966557502746582,
"step": 46150
},
{
"epoch": 5.90816,
"grad_norm": 0.5278201699256897,
"learning_rate": 0.00010826574220327067,
"loss": 1.9728958129882812,
"step": 46160
},
{
"epoch": 5.90944,
"grad_norm": 0.5004035830497742,
"learning_rate": 0.0001082076637269649,
"loss": 1.9802356719970704,
"step": 46170
},
{
"epoch": 5.9107199999999995,
"grad_norm": 0.5470080375671387,
"learning_rate": 0.0001081495920424928,
"loss": 1.9808481216430665,
"step": 46180
},
{
"epoch": 5.912,
"grad_norm": 0.5576385259628296,
"learning_rate": 0.00010809152715929172,
"loss": 1.9858572006225585,
"step": 46190
},
{
"epoch": 5.91328,
"grad_norm": 0.5933824777603149,
"learning_rate": 0.00010803346908679804,
"loss": 1.9786058425903321,
"step": 46200
},
{
"epoch": 5.91456,
"grad_norm": 0.5011841058731079,
"learning_rate": 0.00010797541783444697,
"loss": 1.9728206634521483,
"step": 46210
},
{
"epoch": 5.91584,
"grad_norm": 0.5848407745361328,
"learning_rate": 0.00010791737341167272,
"loss": 1.989761734008789,
"step": 46220
},
{
"epoch": 5.91712,
"grad_norm": 0.5781438946723938,
"learning_rate": 0.0001078593358279082,
"loss": 1.9732755661010741,
"step": 46230
},
{
"epoch": 5.9184,
"grad_norm": 0.5058385729789734,
"learning_rate": 0.00010780130509258543,
"loss": 1.9938703536987306,
"step": 46240
},
{
"epoch": 5.91968,
"grad_norm": 0.6043643951416016,
"learning_rate": 0.00010774328121513515,
"loss": 1.9742294311523438,
"step": 46250
},
{
"epoch": 5.92096,
"grad_norm": 0.5059100389480591,
"learning_rate": 0.00010768526420498705,
"loss": 1.995716667175293,
"step": 46260
},
{
"epoch": 5.92224,
"grad_norm": 0.6549926996231079,
"learning_rate": 0.0001076272540715697,
"loss": 1.9806327819824219,
"step": 46270
},
{
"epoch": 5.92352,
"grad_norm": 0.5881863236427307,
"learning_rate": 0.00010756925082431058,
"loss": 1.997120475769043,
"step": 46280
},
{
"epoch": 5.9248,
"grad_norm": 0.5266397595405579,
"learning_rate": 0.00010751125447263594,
"loss": 2.000605583190918,
"step": 46290
},
{
"epoch": 5.92608,
"grad_norm": 0.539055347442627,
"learning_rate": 0.0001074532650259711,
"loss": 1.9799924850463868,
"step": 46300
},
{
"epoch": 5.92736,
"grad_norm": 0.5580207705497742,
"learning_rate": 0.00010739528249374004,
"loss": 1.9890464782714843,
"step": 46310
},
{
"epoch": 5.92864,
"grad_norm": 0.6059518456459045,
"learning_rate": 0.00010733730688536584,
"loss": 1.9975244522094726,
"step": 46320
},
{
"epoch": 5.92992,
"grad_norm": 0.5649478435516357,
"learning_rate": 0.00010727933821027022,
"loss": 1.984848976135254,
"step": 46330
},
{
"epoch": 5.9312000000000005,
"grad_norm": 0.5378928184509277,
"learning_rate": 0.00010722137647787395,
"loss": 1.9768827438354493,
"step": 46340
},
{
"epoch": 5.93248,
"grad_norm": 0.5408989191055298,
"learning_rate": 0.00010716342169759667,
"loss": 1.9802236557006836,
"step": 46350
},
{
"epoch": 5.93376,
"grad_norm": 0.6308097839355469,
"learning_rate": 0.00010710547387885677,
"loss": 1.9761140823364258,
"step": 46360
},
{
"epoch": 5.93504,
"grad_norm": 0.6577153205871582,
"learning_rate": 0.00010704753303107156,
"loss": 1.9870122909545898,
"step": 46370
},
{
"epoch": 5.93632,
"grad_norm": 0.5785885453224182,
"learning_rate": 0.00010698959916365728,
"loss": 1.9906888961791993,
"step": 46380
},
{
"epoch": 5.9376,
"grad_norm": 0.5369347929954529,
"learning_rate": 0.00010693167228602891,
"loss": 1.9697481155395509,
"step": 46390
},
{
"epoch": 5.93888,
"grad_norm": 0.5440661907196045,
"learning_rate": 0.00010687375240760042,
"loss": 2.0077592849731447,
"step": 46400
},
{
"epoch": 5.94016,
"grad_norm": 0.5574685335159302,
"learning_rate": 0.00010681583953778466,
"loss": 1.9787776947021485,
"step": 46410
},
{
"epoch": 5.94144,
"grad_norm": 0.6656621694564819,
"learning_rate": 0.00010675793368599315,
"loss": 1.9853326797485351,
"step": 46420
},
{
"epoch": 5.94272,
"grad_norm": 0.6027159690856934,
"learning_rate": 0.00010670003486163644,
"loss": 1.9844270706176759,
"step": 46430
},
{
"epoch": 5.944,
"grad_norm": 0.5948783755302429,
"learning_rate": 0.00010664214307412389,
"loss": 1.9994451522827148,
"step": 46440
},
{
"epoch": 5.94528,
"grad_norm": 0.552497386932373,
"learning_rate": 0.00010658425833286382,
"loss": 1.9829038619995116,
"step": 46450
},
{
"epoch": 5.94656,
"grad_norm": 0.5735434293746948,
"learning_rate": 0.00010652638064726313,
"loss": 1.9849256515502929,
"step": 46460
},
{
"epoch": 5.94784,
"grad_norm": 0.5350366234779358,
"learning_rate": 0.00010646851002672784,
"loss": 1.9760896682739257,
"step": 46470
},
{
"epoch": 5.94912,
"grad_norm": 0.6024263501167297,
"learning_rate": 0.00010641064648066276,
"loss": 2.0014179229736326,
"step": 46480
},
{
"epoch": 5.9504,
"grad_norm": 0.5790411829948425,
"learning_rate": 0.00010635279001847139,
"loss": 2.000760078430176,
"step": 46490
},
{
"epoch": 5.95168,
"grad_norm": 0.5908792614936829,
"learning_rate": 0.00010629494064955632,
"loss": 1.966510009765625,
"step": 46500
},
{
"epoch": 5.95296,
"grad_norm": 0.5292767286300659,
"learning_rate": 0.00010623709838331887,
"loss": 2.0071983337402344,
"step": 46510
},
{
"epoch": 5.95424,
"grad_norm": 0.5480591654777527,
"learning_rate": 0.00010617926322915913,
"loss": 1.9725728988647462,
"step": 46520
},
{
"epoch": 5.95552,
"grad_norm": 0.5399124622344971,
"learning_rate": 0.0001061214351964762,
"loss": 1.9773927688598634,
"step": 46530
},
{
"epoch": 5.9568,
"grad_norm": 0.4855659008026123,
"learning_rate": 0.00010606361429466787,
"loss": 1.9959627151489259,
"step": 46540
},
{
"epoch": 5.95808,
"grad_norm": 0.6065044403076172,
"learning_rate": 0.00010600580053313093,
"loss": 1.965280532836914,
"step": 46550
},
{
"epoch": 5.95936,
"grad_norm": 0.5414953231811523,
"learning_rate": 0.00010594799392126079,
"loss": 1.962303924560547,
"step": 46560
},
{
"epoch": 5.96064,
"grad_norm": 0.55504310131073,
"learning_rate": 0.0001058901944684519,
"loss": 1.9929946899414062,
"step": 46570
},
{
"epoch": 5.96192,
"grad_norm": 0.6124005913734436,
"learning_rate": 0.0001058324021840975,
"loss": 1.9907045364379883,
"step": 46580
},
{
"epoch": 5.9632,
"grad_norm": 0.6791595220565796,
"learning_rate": 0.00010577461707758958,
"loss": 1.9737562179565429,
"step": 46590
},
{
"epoch": 5.96448,
"grad_norm": 0.5676315426826477,
"learning_rate": 0.000105716839158319,
"loss": 1.9683692932128907,
"step": 46600
},
{
"epoch": 5.9657599999999995,
"grad_norm": 0.5128874778747559,
"learning_rate": 0.0001056590684356756,
"loss": 2.0007801055908203,
"step": 46610
},
{
"epoch": 5.96704,
"grad_norm": 0.5274934768676758,
"learning_rate": 0.00010560130491904776,
"loss": 1.9905298233032227,
"step": 46620
},
{
"epoch": 5.96832,
"grad_norm": 0.5085563063621521,
"learning_rate": 0.00010554354861782291,
"loss": 1.9985103607177734,
"step": 46630
},
{
"epoch": 5.9696,
"grad_norm": 0.5893690586090088,
"learning_rate": 0.00010548579954138736,
"loss": 1.980723762512207,
"step": 46640
},
{
"epoch": 5.97088,
"grad_norm": 0.5319373607635498,
"learning_rate": 0.00010542805769912599,
"loss": 1.9893537521362306,
"step": 46650
},
{
"epoch": 5.97216,
"grad_norm": 0.5448458790779114,
"learning_rate": 0.00010537032310042265,
"loss": 1.9856779098510742,
"step": 46660
},
{
"epoch": 5.97344,
"grad_norm": 0.5361920595169067,
"learning_rate": 0.0001053125957546601,
"loss": 1.9936471939086915,
"step": 46670
},
{
"epoch": 5.97472,
"grad_norm": 0.615745484828949,
"learning_rate": 0.00010525487567121981,
"loss": 1.9735252380371093,
"step": 46680
},
{
"epoch": 5.976,
"grad_norm": 0.5108727216720581,
"learning_rate": 0.00010519716285948206,
"loss": 1.972274398803711,
"step": 46690
},
{
"epoch": 5.97728,
"grad_norm": 0.531282365322113,
"learning_rate": 0.00010513945732882601,
"loss": 1.9774908065795898,
"step": 46700
},
{
"epoch": 5.97856,
"grad_norm": 0.5269642472267151,
"learning_rate": 0.00010508175908862963,
"loss": 1.9926895141601562,
"step": 46710
},
{
"epoch": 5.97984,
"grad_norm": 0.550825297832489,
"learning_rate": 0.0001050240681482696,
"loss": 1.9636404037475585,
"step": 46720
},
{
"epoch": 5.98112,
"grad_norm": 0.6149098873138428,
"learning_rate": 0.00010496638451712154,
"loss": 1.9686792373657227,
"step": 46730
},
{
"epoch": 5.9824,
"grad_norm": 0.5422677397727966,
"learning_rate": 0.00010490870820455992,
"loss": 1.9812324523925782,
"step": 46740
},
{
"epoch": 5.98368,
"grad_norm": 0.5397686958312988,
"learning_rate": 0.00010485103921995777,
"loss": 1.9890701293945312,
"step": 46750
},
{
"epoch": 5.98496,
"grad_norm": 0.502703845500946,
"learning_rate": 0.00010479337757268724,
"loss": 1.9929964065551757,
"step": 46760
},
{
"epoch": 5.9862400000000004,
"grad_norm": 0.5306805968284607,
"learning_rate": 0.00010473572327211914,
"loss": 1.9683404922485352,
"step": 46770
},
{
"epoch": 5.98752,
"grad_norm": 0.561654806137085,
"learning_rate": 0.00010467807632762302,
"loss": 1.9840415954589843,
"step": 46780
},
{
"epoch": 5.9888,
"grad_norm": 0.5578703284263611,
"learning_rate": 0.00010462043674856731,
"loss": 1.9836523056030273,
"step": 46790
},
{
"epoch": 5.99008,
"grad_norm": 0.5089002847671509,
"learning_rate": 0.00010456280454431927,
"loss": 1.9764291763305664,
"step": 46800
},
{
"epoch": 5.99136,
"grad_norm": 0.5650126338005066,
"learning_rate": 0.00010450517972424501,
"loss": 1.9781505584716796,
"step": 46810
},
{
"epoch": 5.99264,
"grad_norm": 0.5884016752243042,
"learning_rate": 0.00010444756229770925,
"loss": 1.9961868286132813,
"step": 46820
},
{
"epoch": 5.99392,
"grad_norm": 0.5217655897140503,
"learning_rate": 0.0001043899522740756,
"loss": 1.9846002578735351,
"step": 46830
},
{
"epoch": 5.9952,
"grad_norm": 0.5891818404197693,
"learning_rate": 0.00010433234966270664,
"loss": 1.9893861770629884,
"step": 46840
},
{
"epoch": 5.99648,
"grad_norm": 0.4867234230041504,
"learning_rate": 0.00010427475447296343,
"loss": 1.9740755081176757,
"step": 46850
},
{
"epoch": 5.9977599999999995,
"grad_norm": 0.5858220458030701,
"learning_rate": 0.00010421716671420602,
"loss": 1.9803401947021484,
"step": 46860
},
{
"epoch": 5.99904,
"grad_norm": 0.5502802133560181,
"learning_rate": 0.00010415958639579332,
"loss": 1.9846601486206055,
"step": 46870
},
{
"epoch": 6.0,
"eval_loss": 1.9871164560317993,
"eval_runtime": 18.5791,
"eval_samples_per_second": 538.239,
"eval_steps_per_second": 67.28,
"step": 46878
}
],
"logging_steps": 10,
"max_steps": 78130,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 9142901090942976.0,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}