Files
100M_low_1000_6910/trainer_state.json
ModelHub XC 76e92de654 初始化项目,由ModelHub XC社区提供模型
Model: craa/100M_low_1000_6910
Source: Original Platform
2026-08-13 19:27:39 +08:00

13856 lines
340 KiB
JSON

{
"best_metric": 3.3003323078155518,
"best_model_checkpoint": "/scratch/cl5625/exceptions/models/100M_low_1000_6910/checkpoint-90000",
"epoch": 10.0,
"eval_steps": 1000,
"global_step": 92750,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.005390835579514825,
"grad_norm": 2.4885668754577637,
"learning_rate": 0.000276,
"loss": 9.0221,
"step": 50
},
{
"epoch": 0.01078167115902965,
"grad_norm": 2.189452648162842,
"learning_rate": 0.0005759999999999999,
"loss": 6.9446,
"step": 100
},
{
"epoch": 0.016172506738544475,
"grad_norm": 2.3896727561950684,
"learning_rate": 0.000599702104695089,
"loss": 6.5175,
"step": 150
},
{
"epoch": 0.0215633423180593,
"grad_norm": 2.172729969024658,
"learning_rate": 0.0005993783054506205,
"loss": 6.2538,
"step": 200
},
{
"epoch": 0.026954177897574125,
"grad_norm": 2.084505081176758,
"learning_rate": 0.0005990545062061521,
"loss": 6.085,
"step": 250
},
{
"epoch": 0.03234501347708895,
"grad_norm": 1.917132019996643,
"learning_rate": 0.0005987307069616836,
"loss": 5.9466,
"step": 300
},
{
"epoch": 0.03773584905660377,
"grad_norm": 1.8903628587722778,
"learning_rate": 0.0005984069077172153,
"loss": 5.8744,
"step": 350
},
{
"epoch": 0.0431266846361186,
"grad_norm": 1.6605838537216187,
"learning_rate": 0.0005980831084727469,
"loss": 5.8104,
"step": 400
},
{
"epoch": 0.04851752021563342,
"grad_norm": 1.1943660974502563,
"learning_rate": 0.0005977593092282784,
"loss": 5.7426,
"step": 450
},
{
"epoch": 0.05390835579514825,
"grad_norm": 0.8570104837417603,
"learning_rate": 0.00059743550998381,
"loss": 5.6401,
"step": 500
},
{
"epoch": 0.05929919137466307,
"grad_norm": 0.9798665046691895,
"learning_rate": 0.0005971117107393416,
"loss": 5.5972,
"step": 550
},
{
"epoch": 0.0646900269541779,
"grad_norm": 1.3670645952224731,
"learning_rate": 0.0005967879114948732,
"loss": 5.5182,
"step": 600
},
{
"epoch": 0.07008086253369272,
"grad_norm": 1.7250298261642456,
"learning_rate": 0.0005964641122504047,
"loss": 5.4522,
"step": 650
},
{
"epoch": 0.07547169811320754,
"grad_norm": 1.133657693862915,
"learning_rate": 0.0005961403130059363,
"loss": 5.3756,
"step": 700
},
{
"epoch": 0.08086253369272237,
"grad_norm": 1.604074478149414,
"learning_rate": 0.0005958165137614678,
"loss": 5.3031,
"step": 750
},
{
"epoch": 0.0862533692722372,
"grad_norm": 1.349126935005188,
"learning_rate": 0.0005954927145169995,
"loss": 5.2333,
"step": 800
},
{
"epoch": 0.09164420485175202,
"grad_norm": 0.9107061624526978,
"learning_rate": 0.0005951689152725309,
"loss": 5.2113,
"step": 850
},
{
"epoch": 0.09703504043126684,
"grad_norm": 1.1596487760543823,
"learning_rate": 0.0005948451160280626,
"loss": 5.1587,
"step": 900
},
{
"epoch": 0.10242587601078167,
"grad_norm": 1.2029236555099487,
"learning_rate": 0.0005945213167835941,
"loss": 5.0937,
"step": 950
},
{
"epoch": 0.1078167115902965,
"grad_norm": 1.1606078147888184,
"learning_rate": 0.0005941975175391257,
"loss": 5.1006,
"step": 1000
},
{
"epoch": 0.1078167115902965,
"eval_accuracy": 0.2282875403061469,
"eval_loss": 5.012904167175293,
"eval_runtime": 185.3465,
"eval_samples_per_second": 97.175,
"eval_steps_per_second": 6.075,
"step": 1000
},
{
"epoch": 0.11320754716981132,
"grad_norm": 0.9551291465759277,
"learning_rate": 0.0005938737182946572,
"loss": 5.0348,
"step": 1050
},
{
"epoch": 0.11859838274932614,
"grad_norm": 1.1001120805740356,
"learning_rate": 0.0005935499190501888,
"loss": 4.9997,
"step": 1100
},
{
"epoch": 0.12398921832884097,
"grad_norm": 1.1628352403640747,
"learning_rate": 0.0005932261198057204,
"loss": 5.0031,
"step": 1150
},
{
"epoch": 0.1293800539083558,
"grad_norm": 0.9464654326438904,
"learning_rate": 0.000592902320561252,
"loss": 4.9564,
"step": 1200
},
{
"epoch": 0.1347708894878706,
"grad_norm": 1.2740072011947632,
"learning_rate": 0.0005925785213167835,
"loss": 4.9358,
"step": 1250
},
{
"epoch": 0.14016172506738545,
"grad_norm": 1.0887484550476074,
"learning_rate": 0.0005922547220723151,
"loss": 4.8907,
"step": 1300
},
{
"epoch": 0.14555256064690028,
"grad_norm": 1.0212318897247314,
"learning_rate": 0.0005919309228278468,
"loss": 4.8418,
"step": 1350
},
{
"epoch": 0.1509433962264151,
"grad_norm": 0.9643754363059998,
"learning_rate": 0.0005916071235833783,
"loss": 4.8537,
"step": 1400
},
{
"epoch": 0.15633423180592992,
"grad_norm": 1.0317773818969727,
"learning_rate": 0.0005912833243389097,
"loss": 4.8049,
"step": 1450
},
{
"epoch": 0.16172506738544473,
"grad_norm": 0.8656252026557922,
"learning_rate": 0.0005909595250944414,
"loss": 4.8053,
"step": 1500
},
{
"epoch": 0.16711590296495957,
"grad_norm": 1.1420655250549316,
"learning_rate": 0.000590635725849973,
"loss": 4.791,
"step": 1550
},
{
"epoch": 0.1725067385444744,
"grad_norm": 0.965045690536499,
"learning_rate": 0.0005903119266055045,
"loss": 4.7651,
"step": 1600
},
{
"epoch": 0.1778975741239892,
"grad_norm": 1.3466740846633911,
"learning_rate": 0.0005899881273610361,
"loss": 4.7348,
"step": 1650
},
{
"epoch": 0.18328840970350405,
"grad_norm": 1.048919439315796,
"learning_rate": 0.0005896643281165677,
"loss": 4.7195,
"step": 1700
},
{
"epoch": 0.18867924528301888,
"grad_norm": 0.885158121585846,
"learning_rate": 0.0005893405288720993,
"loss": 4.6704,
"step": 1750
},
{
"epoch": 0.1940700808625337,
"grad_norm": 1.0743467807769775,
"learning_rate": 0.0005890167296276308,
"loss": 4.6967,
"step": 1800
},
{
"epoch": 0.19946091644204852,
"grad_norm": 1.347840666770935,
"learning_rate": 0.0005886929303831624,
"loss": 4.6479,
"step": 1850
},
{
"epoch": 0.20485175202156333,
"grad_norm": 0.8371959924697876,
"learning_rate": 0.0005883691311386939,
"loss": 4.6219,
"step": 1900
},
{
"epoch": 0.21024258760107817,
"grad_norm": 0.7352165579795837,
"learning_rate": 0.0005880453318942256,
"loss": 4.6023,
"step": 1950
},
{
"epoch": 0.215633423180593,
"grad_norm": 0.9295386075973511,
"learning_rate": 0.0005877215326497571,
"loss": 4.5834,
"step": 2000
},
{
"epoch": 0.215633423180593,
"eval_accuracy": 0.2699751543441435,
"eval_loss": 4.509932994842529,
"eval_runtime": 184.5453,
"eval_samples_per_second": 97.597,
"eval_steps_per_second": 6.101,
"step": 2000
},
{
"epoch": 0.2210242587601078,
"grad_norm": 0.8622216582298279,
"learning_rate": 0.0005873977334052887,
"loss": 4.5713,
"step": 2050
},
{
"epoch": 0.22641509433962265,
"grad_norm": 0.8100547790527344,
"learning_rate": 0.0005870739341608202,
"loss": 4.5488,
"step": 2100
},
{
"epoch": 0.23180592991913745,
"grad_norm": 0.8752567768096924,
"learning_rate": 0.0005867501349163519,
"loss": 4.5261,
"step": 2150
},
{
"epoch": 0.2371967654986523,
"grad_norm": 0.7747423648834229,
"learning_rate": 0.0005864263356718833,
"loss": 4.521,
"step": 2200
},
{
"epoch": 0.24258760107816713,
"grad_norm": 0.9414746165275574,
"learning_rate": 0.000586102536427415,
"loss": 4.4952,
"step": 2250
},
{
"epoch": 0.24797843665768193,
"grad_norm": 0.8330004215240479,
"learning_rate": 0.0005857787371829465,
"loss": 4.4901,
"step": 2300
},
{
"epoch": 0.25336927223719674,
"grad_norm": 1.1575932502746582,
"learning_rate": 0.0005854549379384781,
"loss": 4.4639,
"step": 2350
},
{
"epoch": 0.2587601078167116,
"grad_norm": 0.8225982785224915,
"learning_rate": 0.0005851311386940096,
"loss": 4.4531,
"step": 2400
},
{
"epoch": 0.2641509433962264,
"grad_norm": 0.7044151425361633,
"learning_rate": 0.0005848073394495412,
"loss": 4.4394,
"step": 2450
},
{
"epoch": 0.2695417789757412,
"grad_norm": 0.9969947934150696,
"learning_rate": 0.0005844835402050728,
"loss": 4.4196,
"step": 2500
},
{
"epoch": 0.2749326145552561,
"grad_norm": 0.7618424892425537,
"learning_rate": 0.0005841597409606044,
"loss": 4.3899,
"step": 2550
},
{
"epoch": 0.2803234501347709,
"grad_norm": 0.8111974596977234,
"learning_rate": 0.000583835941716136,
"loss": 4.4015,
"step": 2600
},
{
"epoch": 0.2857142857142857,
"grad_norm": 0.7472854256629944,
"learning_rate": 0.0005835121424716675,
"loss": 4.3827,
"step": 2650
},
{
"epoch": 0.29110512129380056,
"grad_norm": 0.7100257277488708,
"learning_rate": 0.0005831883432271992,
"loss": 4.3675,
"step": 2700
},
{
"epoch": 0.29649595687331537,
"grad_norm": 0.876581072807312,
"learning_rate": 0.0005828645439827307,
"loss": 4.3638,
"step": 2750
},
{
"epoch": 0.3018867924528302,
"grad_norm": 0.7523159980773926,
"learning_rate": 0.0005825407447382622,
"loss": 4.3216,
"step": 2800
},
{
"epoch": 0.30727762803234504,
"grad_norm": 0.79215008020401,
"learning_rate": 0.0005822169454937938,
"loss": 4.3475,
"step": 2850
},
{
"epoch": 0.31266846361185985,
"grad_norm": 0.7433144450187683,
"learning_rate": 0.0005818931462493254,
"loss": 4.3098,
"step": 2900
},
{
"epoch": 0.31805929919137466,
"grad_norm": 0.8124609589576721,
"learning_rate": 0.0005815693470048569,
"loss": 4.3204,
"step": 2950
},
{
"epoch": 0.32345013477088946,
"grad_norm": 0.6567305326461792,
"learning_rate": 0.0005812455477603885,
"loss": 4.3084,
"step": 3000
},
{
"epoch": 0.32345013477088946,
"eval_accuracy": 0.2988912733368747,
"eval_loss": 4.234070301055908,
"eval_runtime": 184.5549,
"eval_samples_per_second": 97.592,
"eval_steps_per_second": 6.101,
"step": 3000
},
{
"epoch": 0.3288409703504043,
"grad_norm": 0.7118846774101257,
"learning_rate": 0.0005809217485159201,
"loss": 4.3096,
"step": 3050
},
{
"epoch": 0.33423180592991913,
"grad_norm": 0.8099746704101562,
"learning_rate": 0.0005805979492714517,
"loss": 4.2873,
"step": 3100
},
{
"epoch": 0.33962264150943394,
"grad_norm": 0.8107202649116516,
"learning_rate": 0.0005802741500269832,
"loss": 4.2686,
"step": 3150
},
{
"epoch": 0.3450134770889488,
"grad_norm": 0.6466975212097168,
"learning_rate": 0.0005799503507825148,
"loss": 4.2688,
"step": 3200
},
{
"epoch": 0.3504043126684636,
"grad_norm": 0.7568036317825317,
"learning_rate": 0.0005796265515380463,
"loss": 4.2587,
"step": 3250
},
{
"epoch": 0.3557951482479784,
"grad_norm": 0.9061019420623779,
"learning_rate": 0.000579302752293578,
"loss": 4.2647,
"step": 3300
},
{
"epoch": 0.3611859838274933,
"grad_norm": 1.0397237539291382,
"learning_rate": 0.0005789789530491095,
"loss": 4.2395,
"step": 3350
},
{
"epoch": 0.3665768194070081,
"grad_norm": 0.7014638185501099,
"learning_rate": 0.0005786551538046411,
"loss": 4.2334,
"step": 3400
},
{
"epoch": 0.3719676549865229,
"grad_norm": 0.570610523223877,
"learning_rate": 0.0005783313545601726,
"loss": 4.2213,
"step": 3450
},
{
"epoch": 0.37735849056603776,
"grad_norm": 1.0071295499801636,
"learning_rate": 0.0005780075553157043,
"loss": 4.2243,
"step": 3500
},
{
"epoch": 0.38274932614555257,
"grad_norm": 0.6485183238983154,
"learning_rate": 0.0005776837560712357,
"loss": 4.2156,
"step": 3550
},
{
"epoch": 0.3881401617250674,
"grad_norm": 0.9307100772857666,
"learning_rate": 0.0005773599568267673,
"loss": 4.2087,
"step": 3600
},
{
"epoch": 0.3935309973045822,
"grad_norm": 0.6979120969772339,
"learning_rate": 0.0005770361575822989,
"loss": 4.1969,
"step": 3650
},
{
"epoch": 0.39892183288409705,
"grad_norm": 0.7178517580032349,
"learning_rate": 0.0005767123583378305,
"loss": 4.1876,
"step": 3700
},
{
"epoch": 0.40431266846361186,
"grad_norm": 0.7951653599739075,
"learning_rate": 0.000576388559093362,
"loss": 4.1893,
"step": 3750
},
{
"epoch": 0.40970350404312667,
"grad_norm": 0.6346942186355591,
"learning_rate": 0.0005760647598488936,
"loss": 4.1542,
"step": 3800
},
{
"epoch": 0.41509433962264153,
"grad_norm": 0.7404624819755554,
"learning_rate": 0.0005757409606044253,
"loss": 4.1625,
"step": 3850
},
{
"epoch": 0.42048517520215634,
"grad_norm": 0.7455494403839111,
"learning_rate": 0.0005754171613599568,
"loss": 4.1821,
"step": 3900
},
{
"epoch": 0.42587601078167114,
"grad_norm": 1.5372980833053589,
"learning_rate": 0.0005750933621154884,
"loss": 4.1502,
"step": 3950
},
{
"epoch": 0.431266846361186,
"grad_norm": 0.7405364513397217,
"learning_rate": 0.0005747695628710199,
"loss": 4.1514,
"step": 4000
},
{
"epoch": 0.431266846361186,
"eval_accuracy": 0.31312186801260067,
"eval_loss": 4.08581018447876,
"eval_runtime": 184.3526,
"eval_samples_per_second": 97.699,
"eval_steps_per_second": 6.108,
"step": 4000
},
{
"epoch": 0.4366576819407008,
"grad_norm": 0.7655566930770874,
"learning_rate": 0.0005744457636265515,
"loss": 4.1594,
"step": 4050
},
{
"epoch": 0.4420485175202156,
"grad_norm": 0.6172462701797485,
"learning_rate": 0.0005741219643820831,
"loss": 4.1542,
"step": 4100
},
{
"epoch": 0.4474393530997305,
"grad_norm": 0.6381276249885559,
"learning_rate": 0.0005737981651376146,
"loss": 4.137,
"step": 4150
},
{
"epoch": 0.4528301886792453,
"grad_norm": 0.6830670237541199,
"learning_rate": 0.0005734743658931462,
"loss": 4.1449,
"step": 4200
},
{
"epoch": 0.4582210242587601,
"grad_norm": 0.5589593052864075,
"learning_rate": 0.0005731505666486778,
"loss": 4.1287,
"step": 4250
},
{
"epoch": 0.4636118598382749,
"grad_norm": 0.7087463736534119,
"learning_rate": 0.0005728267674042093,
"loss": 4.1346,
"step": 4300
},
{
"epoch": 0.46900269541778977,
"grad_norm": 0.6437616348266602,
"learning_rate": 0.0005725029681597409,
"loss": 4.1202,
"step": 4350
},
{
"epoch": 0.4743935309973046,
"grad_norm": 0.7208832502365112,
"learning_rate": 0.0005721791689152725,
"loss": 4.1055,
"step": 4400
},
{
"epoch": 0.4797843665768194,
"grad_norm": 0.6060395836830139,
"learning_rate": 0.0005718553696708041,
"loss": 4.1123,
"step": 4450
},
{
"epoch": 0.48517520215633425,
"grad_norm": 0.5975813865661621,
"learning_rate": 0.0005715315704263356,
"loss": 4.1078,
"step": 4500
},
{
"epoch": 0.49056603773584906,
"grad_norm": 0.7575790286064148,
"learning_rate": 0.0005712077711818672,
"loss": 4.1149,
"step": 4550
},
{
"epoch": 0.49595687331536387,
"grad_norm": 0.8036497235298157,
"learning_rate": 0.0005708839719373987,
"loss": 4.1082,
"step": 4600
},
{
"epoch": 0.5013477088948787,
"grad_norm": 0.6828065514564514,
"learning_rate": 0.0005705601726929304,
"loss": 4.0686,
"step": 4650
},
{
"epoch": 0.5067385444743935,
"grad_norm": 0.6462509036064148,
"learning_rate": 0.0005702363734484619,
"loss": 4.0673,
"step": 4700
},
{
"epoch": 0.5121293800539084,
"grad_norm": 0.6186754107475281,
"learning_rate": 0.0005699125742039935,
"loss": 4.0721,
"step": 4750
},
{
"epoch": 0.5175202156334232,
"grad_norm": 0.6847729086875916,
"learning_rate": 0.000569588774959525,
"loss": 4.0707,
"step": 4800
},
{
"epoch": 0.522911051212938,
"grad_norm": 0.6802958846092224,
"learning_rate": 0.0005692649757150567,
"loss": 4.0655,
"step": 4850
},
{
"epoch": 0.5283018867924528,
"grad_norm": 0.6418774724006653,
"learning_rate": 0.0005689411764705881,
"loss": 4.0666,
"step": 4900
},
{
"epoch": 0.5336927223719676,
"grad_norm": 0.7241053581237793,
"learning_rate": 0.0005686173772261197,
"loss": 4.0583,
"step": 4950
},
{
"epoch": 0.5390835579514824,
"grad_norm": 0.7215654253959656,
"learning_rate": 0.0005682935779816514,
"loss": 4.0572,
"step": 5000
},
{
"epoch": 0.5390835579514824,
"eval_accuracy": 0.32137209909012987,
"eval_loss": 3.990335702896118,
"eval_runtime": 184.5721,
"eval_samples_per_second": 97.582,
"eval_steps_per_second": 6.101,
"step": 5000
},
{
"epoch": 0.5444743935309974,
"grad_norm": 0.700507402420044,
"learning_rate": 0.0005679697787371829,
"loss": 4.066,
"step": 5050
},
{
"epoch": 0.5498652291105122,
"grad_norm": 0.5913358330726624,
"learning_rate": 0.0005676459794927145,
"loss": 4.0516,
"step": 5100
},
{
"epoch": 0.555256064690027,
"grad_norm": 0.5954362154006958,
"learning_rate": 0.000567322180248246,
"loss": 4.0433,
"step": 5150
},
{
"epoch": 0.5606469002695418,
"grad_norm": 0.6123712658882141,
"learning_rate": 0.0005669983810037777,
"loss": 4.0523,
"step": 5200
},
{
"epoch": 0.5660377358490566,
"grad_norm": 0.5665382146835327,
"learning_rate": 0.0005666745817593092,
"loss": 4.0282,
"step": 5250
},
{
"epoch": 0.5714285714285714,
"grad_norm": 0.5454007983207703,
"learning_rate": 0.0005663507825148408,
"loss": 4.0273,
"step": 5300
},
{
"epoch": 0.5768194070080862,
"grad_norm": 0.7059000134468079,
"learning_rate": 0.0005660269832703723,
"loss": 4.022,
"step": 5350
},
{
"epoch": 0.5822102425876011,
"grad_norm": 0.6376985311508179,
"learning_rate": 0.0005657031840259039,
"loss": 4.0252,
"step": 5400
},
{
"epoch": 0.5876010781671159,
"grad_norm": 0.7689996957778931,
"learning_rate": 0.0005653793847814355,
"loss": 4.0216,
"step": 5450
},
{
"epoch": 0.5929919137466307,
"grad_norm": 0.7546736598014832,
"learning_rate": 0.000565055585536967,
"loss": 4.0014,
"step": 5500
},
{
"epoch": 0.5983827493261455,
"grad_norm": 0.6733143925666809,
"learning_rate": 0.0005647317862924986,
"loss": 4.0276,
"step": 5550
},
{
"epoch": 0.6037735849056604,
"grad_norm": 0.6395922303199768,
"learning_rate": 0.0005644079870480302,
"loss": 3.974,
"step": 5600
},
{
"epoch": 0.6091644204851752,
"grad_norm": 0.5772866010665894,
"learning_rate": 0.0005640841878035617,
"loss": 4.0094,
"step": 5650
},
{
"epoch": 0.6145552560646901,
"grad_norm": 0.6236124634742737,
"learning_rate": 0.0005637603885590933,
"loss": 4.0101,
"step": 5700
},
{
"epoch": 0.6199460916442049,
"grad_norm": 0.6327507495880127,
"learning_rate": 0.0005634365893146248,
"loss": 4.0097,
"step": 5750
},
{
"epoch": 0.6253369272237197,
"grad_norm": 0.6486158967018127,
"learning_rate": 0.0005631127900701565,
"loss": 3.9994,
"step": 5800
},
{
"epoch": 0.6307277628032345,
"grad_norm": 0.6257344484329224,
"learning_rate": 0.000562788990825688,
"loss": 3.9734,
"step": 5850
},
{
"epoch": 0.6361185983827493,
"grad_norm": 0.6696811318397522,
"learning_rate": 0.0005624651915812196,
"loss": 3.9876,
"step": 5900
},
{
"epoch": 0.6415094339622641,
"grad_norm": 0.7144728302955627,
"learning_rate": 0.0005621413923367511,
"loss": 3.9944,
"step": 5950
},
{
"epoch": 0.6469002695417789,
"grad_norm": 0.6092597842216492,
"learning_rate": 0.0005618175930922828,
"loss": 4.0005,
"step": 6000
},
{
"epoch": 0.6469002695417789,
"eval_accuracy": 0.32822603190635513,
"eval_loss": 3.9149487018585205,
"eval_runtime": 184.3617,
"eval_samples_per_second": 97.694,
"eval_steps_per_second": 6.108,
"step": 6000
},
{
"epoch": 0.6522911051212938,
"grad_norm": 0.5555917620658875,
"learning_rate": 0.0005614937938478143,
"loss": 3.9873,
"step": 6050
},
{
"epoch": 0.6576819407008087,
"grad_norm": 0.6707965731620789,
"learning_rate": 0.0005611699946033459,
"loss": 3.9828,
"step": 6100
},
{
"epoch": 0.6630727762803235,
"grad_norm": 0.5630139708518982,
"learning_rate": 0.0005608461953588774,
"loss": 3.9757,
"step": 6150
},
{
"epoch": 0.6684636118598383,
"grad_norm": 0.6303977370262146,
"learning_rate": 0.000560522396114409,
"loss": 3.9702,
"step": 6200
},
{
"epoch": 0.6738544474393531,
"grad_norm": 0.5762717723846436,
"learning_rate": 0.0005601985968699405,
"loss": 3.9541,
"step": 6250
},
{
"epoch": 0.6792452830188679,
"grad_norm": 0.6994869709014893,
"learning_rate": 0.0005598747976254721,
"loss": 3.9698,
"step": 6300
},
{
"epoch": 0.6846361185983828,
"grad_norm": 0.5598124265670776,
"learning_rate": 0.0005595509983810038,
"loss": 3.9419,
"step": 6350
},
{
"epoch": 0.6900269541778976,
"grad_norm": 0.5439598560333252,
"learning_rate": 0.0005592271991365353,
"loss": 3.9549,
"step": 6400
},
{
"epoch": 0.6954177897574124,
"grad_norm": 0.5971542000770569,
"learning_rate": 0.0005589033998920669,
"loss": 3.9681,
"step": 6450
},
{
"epoch": 0.7008086253369272,
"grad_norm": 0.6659536361694336,
"learning_rate": 0.0005585796006475984,
"loss": 3.9407,
"step": 6500
},
{
"epoch": 0.706199460916442,
"grad_norm": 0.7119298577308655,
"learning_rate": 0.0005582558014031301,
"loss": 3.9646,
"step": 6550
},
{
"epoch": 0.7115902964959568,
"grad_norm": 0.5771445035934448,
"learning_rate": 0.0005579320021586616,
"loss": 3.9469,
"step": 6600
},
{
"epoch": 0.7169811320754716,
"grad_norm": 0.5822080969810486,
"learning_rate": 0.0005576082029141932,
"loss": 3.9439,
"step": 6650
},
{
"epoch": 0.7223719676549866,
"grad_norm": 0.535957396030426,
"learning_rate": 0.0005572844036697247,
"loss": 3.9286,
"step": 6700
},
{
"epoch": 0.7277628032345014,
"grad_norm": 0.5305382609367371,
"learning_rate": 0.0005569606044252563,
"loss": 3.936,
"step": 6750
},
{
"epoch": 0.7331536388140162,
"grad_norm": 0.7795604467391968,
"learning_rate": 0.0005566368051807879,
"loss": 3.9529,
"step": 6800
},
{
"epoch": 0.738544474393531,
"grad_norm": 0.6772651076316833,
"learning_rate": 0.0005563130059363194,
"loss": 3.9368,
"step": 6850
},
{
"epoch": 0.7439353099730458,
"grad_norm": 0.574789822101593,
"learning_rate": 0.000555989206691851,
"loss": 3.9362,
"step": 6900
},
{
"epoch": 0.7493261455525606,
"grad_norm": 0.5974192023277283,
"learning_rate": 0.0005556654074473826,
"loss": 3.9374,
"step": 6950
},
{
"epoch": 0.7547169811320755,
"grad_norm": 0.6801210045814514,
"learning_rate": 0.0005553416082029141,
"loss": 3.9239,
"step": 7000
},
{
"epoch": 0.7547169811320755,
"eval_accuracy": 0.33335944624403807,
"eval_loss": 3.860983371734619,
"eval_runtime": 184.1953,
"eval_samples_per_second": 97.782,
"eval_steps_per_second": 6.113,
"step": 7000
},
{
"epoch": 0.7601078167115903,
"grad_norm": 0.6873490810394287,
"learning_rate": 0.0005550178089584457,
"loss": 3.9175,
"step": 7050
},
{
"epoch": 0.7654986522911051,
"grad_norm": 0.6209083795547485,
"learning_rate": 0.0005546940097139772,
"loss": 3.9162,
"step": 7100
},
{
"epoch": 0.77088948787062,
"grad_norm": 0.6457363963127136,
"learning_rate": 0.0005543702104695089,
"loss": 3.9208,
"step": 7150
},
{
"epoch": 0.7762803234501348,
"grad_norm": 0.5128622055053711,
"learning_rate": 0.0005540464112250404,
"loss": 3.9169,
"step": 7200
},
{
"epoch": 0.7816711590296496,
"grad_norm": 0.6504648327827454,
"learning_rate": 0.000553722611980572,
"loss": 3.9306,
"step": 7250
},
{
"epoch": 0.7870619946091644,
"grad_norm": 0.5498122572898865,
"learning_rate": 0.0005533988127361035,
"loss": 3.9121,
"step": 7300
},
{
"epoch": 0.7924528301886793,
"grad_norm": 0.55793696641922,
"learning_rate": 0.0005530750134916352,
"loss": 3.9261,
"step": 7350
},
{
"epoch": 0.7978436657681941,
"grad_norm": 0.6482006907463074,
"learning_rate": 0.0005527512142471668,
"loss": 3.9,
"step": 7400
},
{
"epoch": 0.8032345013477089,
"grad_norm": 0.582915186882019,
"learning_rate": 0.0005524274150026982,
"loss": 3.9028,
"step": 7450
},
{
"epoch": 0.8086253369272237,
"grad_norm": 0.608665406703949,
"learning_rate": 0.0005521036157582299,
"loss": 3.8952,
"step": 7500
},
{
"epoch": 0.8140161725067385,
"grad_norm": 0.6810957193374634,
"learning_rate": 0.0005517798165137614,
"loss": 3.8981,
"step": 7550
},
{
"epoch": 0.8194070080862533,
"grad_norm": 0.5375815033912659,
"learning_rate": 0.000551456017269293,
"loss": 3.874,
"step": 7600
},
{
"epoch": 0.8247978436657682,
"grad_norm": 0.5809592008590698,
"learning_rate": 0.0005511322180248245,
"loss": 3.8917,
"step": 7650
},
{
"epoch": 0.8301886792452831,
"grad_norm": 0.5947631001472473,
"learning_rate": 0.0005508084187803562,
"loss": 3.9022,
"step": 7700
},
{
"epoch": 0.8355795148247979,
"grad_norm": 0.5407851934432983,
"learning_rate": 0.0005504846195358877,
"loss": 3.8966,
"step": 7750
},
{
"epoch": 0.8409703504043127,
"grad_norm": 0.5668228268623352,
"learning_rate": 0.0005501608202914193,
"loss": 3.8862,
"step": 7800
},
{
"epoch": 0.8463611859838275,
"grad_norm": 0.6631332635879517,
"learning_rate": 0.0005498370210469508,
"loss": 3.8898,
"step": 7850
},
{
"epoch": 0.8517520215633423,
"grad_norm": 0.5788032412528992,
"learning_rate": 0.0005495132218024824,
"loss": 3.8713,
"step": 7900
},
{
"epoch": 0.8571428571428571,
"grad_norm": 0.634852945804596,
"learning_rate": 0.000549189422558014,
"loss": 3.887,
"step": 7950
},
{
"epoch": 0.862533692722372,
"grad_norm": 0.5373018383979797,
"learning_rate": 0.0005488656233135456,
"loss": 3.8793,
"step": 8000
},
{
"epoch": 0.862533692722372,
"eval_accuracy": 0.33780280609121127,
"eval_loss": 3.8143656253814697,
"eval_runtime": 184.4779,
"eval_samples_per_second": 97.632,
"eval_steps_per_second": 6.104,
"step": 8000
},
{
"epoch": 0.8679245283018868,
"grad_norm": 0.5931937098503113,
"learning_rate": 0.0005485418240690771,
"loss": 3.8639,
"step": 8050
},
{
"epoch": 0.8733153638814016,
"grad_norm": 0.6077438592910767,
"learning_rate": 0.0005482180248246087,
"loss": 3.8798,
"step": 8100
},
{
"epoch": 0.8787061994609164,
"grad_norm": 0.5896062850952148,
"learning_rate": 0.0005478942255801403,
"loss": 3.8736,
"step": 8150
},
{
"epoch": 0.8840970350404312,
"grad_norm": 0.5305744409561157,
"learning_rate": 0.0005475704263356718,
"loss": 3.8645,
"step": 8200
},
{
"epoch": 0.889487870619946,
"grad_norm": 0.5718162655830383,
"learning_rate": 0.0005472466270912034,
"loss": 3.8731,
"step": 8250
},
{
"epoch": 0.894878706199461,
"grad_norm": 0.6724256277084351,
"learning_rate": 0.000546922827846735,
"loss": 3.8623,
"step": 8300
},
{
"epoch": 0.9002695417789758,
"grad_norm": 0.5793610215187073,
"learning_rate": 0.0005465990286022665,
"loss": 3.8641,
"step": 8350
},
{
"epoch": 0.9056603773584906,
"grad_norm": 0.5765576362609863,
"learning_rate": 0.0005462752293577981,
"loss": 3.8782,
"step": 8400
},
{
"epoch": 0.9110512129380054,
"grad_norm": 0.568999171257019,
"learning_rate": 0.0005459514301133296,
"loss": 3.8727,
"step": 8450
},
{
"epoch": 0.9164420485175202,
"grad_norm": 0.5299171805381775,
"learning_rate": 0.0005456276308688613,
"loss": 3.8685,
"step": 8500
},
{
"epoch": 0.921832884097035,
"grad_norm": 0.5593698620796204,
"learning_rate": 0.0005453038316243929,
"loss": 3.8607,
"step": 8550
},
{
"epoch": 0.9272237196765498,
"grad_norm": 0.5512341260910034,
"learning_rate": 0.0005449800323799244,
"loss": 3.871,
"step": 8600
},
{
"epoch": 0.9326145552560647,
"grad_norm": 0.5711010098457336,
"learning_rate": 0.000544656233135456,
"loss": 3.848,
"step": 8650
},
{
"epoch": 0.9380053908355795,
"grad_norm": 0.5815480351448059,
"learning_rate": 0.0005443324338909875,
"loss": 3.8527,
"step": 8700
},
{
"epoch": 0.9433962264150944,
"grad_norm": 0.6103795170783997,
"learning_rate": 0.0005440086346465192,
"loss": 3.8654,
"step": 8750
},
{
"epoch": 0.9487870619946092,
"grad_norm": 0.6336491703987122,
"learning_rate": 0.0005436848354020506,
"loss": 3.8468,
"step": 8800
},
{
"epoch": 0.954177897574124,
"grad_norm": 0.5809378623962402,
"learning_rate": 0.0005433610361575823,
"loss": 3.8531,
"step": 8850
},
{
"epoch": 0.9595687331536388,
"grad_norm": 0.5474573373794556,
"learning_rate": 0.0005430372369131138,
"loss": 3.8367,
"step": 8900
},
{
"epoch": 0.9649595687331537,
"grad_norm": 0.5602928996086121,
"learning_rate": 0.0005427134376686454,
"loss": 3.8501,
"step": 8950
},
{
"epoch": 0.9703504043126685,
"grad_norm": 0.6437180638313293,
"learning_rate": 0.0005423896384241769,
"loss": 3.8385,
"step": 9000
},
{
"epoch": 0.9703504043126685,
"eval_accuracy": 0.3411421439453015,
"eval_loss": 3.7762644290924072,
"eval_runtime": 184.2631,
"eval_samples_per_second": 97.746,
"eval_steps_per_second": 6.111,
"step": 9000
},
{
"epoch": 0.9757412398921833,
"grad_norm": 0.6614698767662048,
"learning_rate": 0.0005420658391797086,
"loss": 3.8319,
"step": 9050
},
{
"epoch": 0.9811320754716981,
"grad_norm": 0.602741539478302,
"learning_rate": 0.0005417420399352401,
"loss": 3.8436,
"step": 9100
},
{
"epoch": 0.9865229110512129,
"grad_norm": 0.6384924650192261,
"learning_rate": 0.0005414182406907717,
"loss": 3.8352,
"step": 9150
},
{
"epoch": 0.9919137466307277,
"grad_norm": 0.6380344033241272,
"learning_rate": 0.0005410944414463032,
"loss": 3.8506,
"step": 9200
},
{
"epoch": 0.9973045822102425,
"grad_norm": 0.5483061075210571,
"learning_rate": 0.0005407706422018348,
"loss": 3.8496,
"step": 9250
},
{
"epoch": 1.0026954177897573,
"grad_norm": 0.6125226616859436,
"learning_rate": 0.0005404468429573664,
"loss": 3.7955,
"step": 9300
},
{
"epoch": 1.0080862533692723,
"grad_norm": 0.5759546756744385,
"learning_rate": 0.000540123043712898,
"loss": 3.7632,
"step": 9350
},
{
"epoch": 1.013477088948787,
"grad_norm": 0.597389280796051,
"learning_rate": 0.0005397992444684295,
"loss": 3.7467,
"step": 9400
},
{
"epoch": 1.0188679245283019,
"grad_norm": 0.5881744623184204,
"learning_rate": 0.0005394754452239611,
"loss": 3.7607,
"step": 9450
},
{
"epoch": 1.0242587601078168,
"grad_norm": 0.5600555539131165,
"learning_rate": 0.0005391516459794927,
"loss": 3.7732,
"step": 9500
},
{
"epoch": 1.0296495956873315,
"grad_norm": 0.576345682144165,
"learning_rate": 0.0005388278467350242,
"loss": 3.7823,
"step": 9550
},
{
"epoch": 1.0350404312668464,
"grad_norm": 0.5923919081687927,
"learning_rate": 0.0005385040474905557,
"loss": 3.774,
"step": 9600
},
{
"epoch": 1.0404312668463611,
"grad_norm": 0.5580589771270752,
"learning_rate": 0.0005381802482460874,
"loss": 3.7841,
"step": 9650
},
{
"epoch": 1.045822102425876,
"grad_norm": 0.6432645320892334,
"learning_rate": 0.000537856449001619,
"loss": 3.7838,
"step": 9700
},
{
"epoch": 1.0512129380053907,
"grad_norm": 0.6011029481887817,
"learning_rate": 0.0005375326497571505,
"loss": 3.7544,
"step": 9750
},
{
"epoch": 1.0566037735849056,
"grad_norm": 0.5988569855690002,
"learning_rate": 0.000537208850512682,
"loss": 3.7645,
"step": 9800
},
{
"epoch": 1.0619946091644206,
"grad_norm": 0.5536928772926331,
"learning_rate": 0.0005368850512682137,
"loss": 3.7599,
"step": 9850
},
{
"epoch": 1.0673854447439353,
"grad_norm": 0.49431952834129333,
"learning_rate": 0.0005365612520237453,
"loss": 3.7743,
"step": 9900
},
{
"epoch": 1.0727762803234502,
"grad_norm": 0.5850538015365601,
"learning_rate": 0.0005362374527792768,
"loss": 3.775,
"step": 9950
},
{
"epoch": 1.0781671159029649,
"grad_norm": 0.5507252216339111,
"learning_rate": 0.0005359136535348084,
"loss": 3.7731,
"step": 10000
},
{
"epoch": 1.0781671159029649,
"eval_accuracy": 0.3448305835279397,
"eval_loss": 3.751156806945801,
"eval_runtime": 184.2345,
"eval_samples_per_second": 97.761,
"eval_steps_per_second": 6.112,
"step": 10000
},
{
"epoch": 1.0835579514824798,
"grad_norm": 0.5451328158378601,
"learning_rate": 0.0005355898542903399,
"loss": 3.7508,
"step": 10050
},
{
"epoch": 1.0889487870619945,
"grad_norm": 0.5866617560386658,
"learning_rate": 0.0005352660550458716,
"loss": 3.7558,
"step": 10100
},
{
"epoch": 1.0943396226415094,
"grad_norm": 0.6449048519134521,
"learning_rate": 0.000534942255801403,
"loss": 3.744,
"step": 10150
},
{
"epoch": 1.0997304582210243,
"grad_norm": 0.6361781358718872,
"learning_rate": 0.0005346184565569347,
"loss": 3.7595,
"step": 10200
},
{
"epoch": 1.105121293800539,
"grad_norm": 0.5851870179176331,
"learning_rate": 0.0005342946573124662,
"loss": 3.7785,
"step": 10250
},
{
"epoch": 1.110512129380054,
"grad_norm": 0.6075997352600098,
"learning_rate": 0.0005339708580679978,
"loss": 3.7803,
"step": 10300
},
{
"epoch": 1.1159029649595686,
"grad_norm": 0.5346486568450928,
"learning_rate": 0.0005336470588235293,
"loss": 3.7569,
"step": 10350
},
{
"epoch": 1.1212938005390836,
"grad_norm": 0.654135525226593,
"learning_rate": 0.000533323259579061,
"loss": 3.7587,
"step": 10400
},
{
"epoch": 1.1266846361185983,
"grad_norm": 0.5073196291923523,
"learning_rate": 0.0005329994603345925,
"loss": 3.7544,
"step": 10450
},
{
"epoch": 1.1320754716981132,
"grad_norm": 0.5539218783378601,
"learning_rate": 0.0005326756610901241,
"loss": 3.7527,
"step": 10500
},
{
"epoch": 1.137466307277628,
"grad_norm": 0.6038421392440796,
"learning_rate": 0.0005323518618456556,
"loss": 3.7532,
"step": 10550
},
{
"epoch": 1.1428571428571428,
"grad_norm": 0.5592815279960632,
"learning_rate": 0.0005320280626011872,
"loss": 3.7661,
"step": 10600
},
{
"epoch": 1.1482479784366577,
"grad_norm": 0.5629820823669434,
"learning_rate": 0.0005317042633567188,
"loss": 3.7506,
"step": 10650
},
{
"epoch": 1.1536388140161726,
"grad_norm": 0.6271799206733704,
"learning_rate": 0.0005313804641122504,
"loss": 3.7569,
"step": 10700
},
{
"epoch": 1.1590296495956873,
"grad_norm": 0.5629302859306335,
"learning_rate": 0.0005310566648677819,
"loss": 3.7517,
"step": 10750
},
{
"epoch": 1.1644204851752022,
"grad_norm": 0.5934892892837524,
"learning_rate": 0.0005307328656233135,
"loss": 3.7418,
"step": 10800
},
{
"epoch": 1.169811320754717,
"grad_norm": 0.61458420753479,
"learning_rate": 0.000530409066378845,
"loss": 3.7404,
"step": 10850
},
{
"epoch": 1.1752021563342319,
"grad_norm": 0.8293150067329407,
"learning_rate": 0.0005300852671343766,
"loss": 3.751,
"step": 10900
},
{
"epoch": 1.1805929919137466,
"grad_norm": 0.6258216500282288,
"learning_rate": 0.0005297614678899081,
"loss": 3.7443,
"step": 10950
},
{
"epoch": 1.1859838274932615,
"grad_norm": 0.547909140586853,
"learning_rate": 0.0005294376686454398,
"loss": 3.7342,
"step": 11000
},
{
"epoch": 1.1859838274932615,
"eval_accuracy": 0.3479475089206737,
"eval_loss": 3.7171225547790527,
"eval_runtime": 184.3676,
"eval_samples_per_second": 97.691,
"eval_steps_per_second": 6.107,
"step": 11000
},
{
"epoch": 1.1913746630727764,
"grad_norm": 0.5612486600875854,
"learning_rate": 0.0005291138694009714,
"loss": 3.7447,
"step": 11050
},
{
"epoch": 1.196765498652291,
"grad_norm": 0.6519018411636353,
"learning_rate": 0.0005287900701565029,
"loss": 3.7382,
"step": 11100
},
{
"epoch": 1.202156334231806,
"grad_norm": 0.6329710483551025,
"learning_rate": 0.0005284662709120345,
"loss": 3.7377,
"step": 11150
},
{
"epoch": 1.2075471698113207,
"grad_norm": 0.6218352317810059,
"learning_rate": 0.0005281424716675661,
"loss": 3.7355,
"step": 11200
},
{
"epoch": 1.2129380053908356,
"grad_norm": 0.588185727596283,
"learning_rate": 0.0005278186724230977,
"loss": 3.7293,
"step": 11250
},
{
"epoch": 1.2183288409703503,
"grad_norm": 0.5499549508094788,
"learning_rate": 0.0005274948731786292,
"loss": 3.7271,
"step": 11300
},
{
"epoch": 1.2237196765498652,
"grad_norm": 0.6718111038208008,
"learning_rate": 0.0005271710739341608,
"loss": 3.7454,
"step": 11350
},
{
"epoch": 1.2291105121293802,
"grad_norm": 0.5798413753509521,
"learning_rate": 0.0005268472746896923,
"loss": 3.7403,
"step": 11400
},
{
"epoch": 1.2345013477088949,
"grad_norm": 0.6953092217445374,
"learning_rate": 0.000526523475445224,
"loss": 3.7288,
"step": 11450
},
{
"epoch": 1.2398921832884098,
"grad_norm": 0.5516878962516785,
"learning_rate": 0.0005261996762007554,
"loss": 3.7159,
"step": 11500
},
{
"epoch": 1.2452830188679245,
"grad_norm": 0.4981798231601715,
"learning_rate": 0.0005258758769562871,
"loss": 3.7374,
"step": 11550
},
{
"epoch": 1.2506738544474394,
"grad_norm": 0.6841148734092712,
"learning_rate": 0.0005255520777118186,
"loss": 3.7389,
"step": 11600
},
{
"epoch": 1.256064690026954,
"grad_norm": 0.5644107460975647,
"learning_rate": 0.0005252282784673502,
"loss": 3.7286,
"step": 11650
},
{
"epoch": 1.261455525606469,
"grad_norm": 0.5982643365859985,
"learning_rate": 0.0005249044792228817,
"loss": 3.7306,
"step": 11700
},
{
"epoch": 1.266846361185984,
"grad_norm": 0.5152930021286011,
"learning_rate": 0.0005245806799784133,
"loss": 3.7265,
"step": 11750
},
{
"epoch": 1.2722371967654986,
"grad_norm": 0.5411483645439148,
"learning_rate": 0.0005242568807339449,
"loss": 3.7235,
"step": 11800
},
{
"epoch": 1.2776280323450135,
"grad_norm": 0.5929835438728333,
"learning_rate": 0.0005239330814894765,
"loss": 3.7178,
"step": 11850
},
{
"epoch": 1.2830188679245282,
"grad_norm": 0.5304673314094543,
"learning_rate": 0.000523609282245008,
"loss": 3.7394,
"step": 11900
},
{
"epoch": 1.2884097035040432,
"grad_norm": 0.5611473321914673,
"learning_rate": 0.0005232854830005396,
"loss": 3.7352,
"step": 11950
},
{
"epoch": 1.2938005390835579,
"grad_norm": 0.5636021494865417,
"learning_rate": 0.0005229616837560712,
"loss": 3.7299,
"step": 12000
},
{
"epoch": 1.2938005390835579,
"eval_accuracy": 0.3499972456492939,
"eval_loss": 3.697054147720337,
"eval_runtime": 184.616,
"eval_samples_per_second": 97.559,
"eval_steps_per_second": 6.099,
"step": 12000
},
{
"epoch": 1.2991913746630728,
"grad_norm": 0.5578932762145996,
"learning_rate": 0.0005226378845116028,
"loss": 3.712,
"step": 12050
},
{
"epoch": 1.3045822102425877,
"grad_norm": 0.6060883402824402,
"learning_rate": 0.0005223140852671344,
"loss": 3.7244,
"step": 12100
},
{
"epoch": 1.3099730458221024,
"grad_norm": 0.6017573475837708,
"learning_rate": 0.0005219902860226659,
"loss": 3.7343,
"step": 12150
},
{
"epoch": 1.3153638814016173,
"grad_norm": 0.5601928234100342,
"learning_rate": 0.0005216664867781975,
"loss": 3.7345,
"step": 12200
},
{
"epoch": 1.320754716981132,
"grad_norm": 0.6401859521865845,
"learning_rate": 0.000521342687533729,
"loss": 3.7167,
"step": 12250
},
{
"epoch": 1.326145552560647,
"grad_norm": 0.6155033111572266,
"learning_rate": 0.0005210188882892606,
"loss": 3.7253,
"step": 12300
},
{
"epoch": 1.3315363881401616,
"grad_norm": 0.513505220413208,
"learning_rate": 0.0005206950890447922,
"loss": 3.7145,
"step": 12350
},
{
"epoch": 1.3369272237196765,
"grad_norm": 0.5291333794593811,
"learning_rate": 0.0005203712898003238,
"loss": 3.729,
"step": 12400
},
{
"epoch": 1.3423180592991915,
"grad_norm": 0.6405612230300903,
"learning_rate": 0.0005200474905558553,
"loss": 3.7195,
"step": 12450
},
{
"epoch": 1.3477088948787062,
"grad_norm": 0.7008240818977356,
"learning_rate": 0.0005197236913113869,
"loss": 3.7201,
"step": 12500
},
{
"epoch": 1.353099730458221,
"grad_norm": 0.5518884658813477,
"learning_rate": 0.0005193998920669184,
"loss": 3.7288,
"step": 12550
},
{
"epoch": 1.3584905660377358,
"grad_norm": 0.6122757792472839,
"learning_rate": 0.0005190760928224501,
"loss": 3.7118,
"step": 12600
},
{
"epoch": 1.3638814016172507,
"grad_norm": 0.5555350184440613,
"learning_rate": 0.0005187522935779816,
"loss": 3.7205,
"step": 12650
},
{
"epoch": 1.3692722371967654,
"grad_norm": 0.5721654295921326,
"learning_rate": 0.0005184284943335132,
"loss": 3.7342,
"step": 12700
},
{
"epoch": 1.3746630727762803,
"grad_norm": 0.5602584481239319,
"learning_rate": 0.0005181046950890447,
"loss": 3.718,
"step": 12750
},
{
"epoch": 1.3800539083557952,
"grad_norm": 0.5544398427009583,
"learning_rate": 0.0005177808958445764,
"loss": 3.708,
"step": 12800
},
{
"epoch": 1.38544474393531,
"grad_norm": 0.5796840786933899,
"learning_rate": 0.0005174570966001078,
"loss": 3.7055,
"step": 12850
},
{
"epoch": 1.3908355795148248,
"grad_norm": 0.5618937015533447,
"learning_rate": 0.0005171332973556395,
"loss": 3.7108,
"step": 12900
},
{
"epoch": 1.3962264150943398,
"grad_norm": 0.5866388082504272,
"learning_rate": 0.000516809498111171,
"loss": 3.7091,
"step": 12950
},
{
"epoch": 1.4016172506738545,
"grad_norm": 0.5813106894493103,
"learning_rate": 0.0005164856988667026,
"loss": 3.7043,
"step": 13000
},
{
"epoch": 1.4016172506738545,
"eval_accuracy": 0.35193485259768953,
"eval_loss": 3.673987865447998,
"eval_runtime": 184.2133,
"eval_samples_per_second": 97.773,
"eval_steps_per_second": 6.112,
"step": 13000
},
{
"epoch": 1.4070080862533692,
"grad_norm": 0.6186528205871582,
"learning_rate": 0.0005161618996222341,
"loss": 3.7163,
"step": 13050
},
{
"epoch": 1.412398921832884,
"grad_norm": 0.5775291323661804,
"learning_rate": 0.0005158381003777657,
"loss": 3.7291,
"step": 13100
},
{
"epoch": 1.417789757412399,
"grad_norm": 0.5397806167602539,
"learning_rate": 0.0005155143011332973,
"loss": 3.7054,
"step": 13150
},
{
"epoch": 1.4231805929919137,
"grad_norm": 0.5625145435333252,
"learning_rate": 0.0005151905018888289,
"loss": 3.7086,
"step": 13200
},
{
"epoch": 1.4285714285714286,
"grad_norm": 0.5848190188407898,
"learning_rate": 0.0005148667026443604,
"loss": 3.7066,
"step": 13250
},
{
"epoch": 1.4339622641509435,
"grad_norm": 0.5622788071632385,
"learning_rate": 0.000514542903399892,
"loss": 3.704,
"step": 13300
},
{
"epoch": 1.4393530997304582,
"grad_norm": 0.5974804162979126,
"learning_rate": 0.0005142191041554237,
"loss": 3.6906,
"step": 13350
},
{
"epoch": 1.444743935309973,
"grad_norm": 0.5385304093360901,
"learning_rate": 0.0005138953049109552,
"loss": 3.6993,
"step": 13400
},
{
"epoch": 1.4501347708894878,
"grad_norm": 0.603137195110321,
"learning_rate": 0.0005135715056664868,
"loss": 3.709,
"step": 13450
},
{
"epoch": 1.4555256064690028,
"grad_norm": 0.5059567093849182,
"learning_rate": 0.0005132477064220183,
"loss": 3.6931,
"step": 13500
},
{
"epoch": 1.4609164420485174,
"grad_norm": 0.6063768267631531,
"learning_rate": 0.0005129239071775499,
"loss": 3.7039,
"step": 13550
},
{
"epoch": 1.4663072776280324,
"grad_norm": 0.6452860236167908,
"learning_rate": 0.0005126001079330814,
"loss": 3.7095,
"step": 13600
},
{
"epoch": 1.4716981132075473,
"grad_norm": 0.5889171361923218,
"learning_rate": 0.000512276308688613,
"loss": 3.6812,
"step": 13650
},
{
"epoch": 1.477088948787062,
"grad_norm": 0.5586681962013245,
"learning_rate": 0.0005119525094441446,
"loss": 3.6798,
"step": 13700
},
{
"epoch": 1.482479784366577,
"grad_norm": 0.5412774682044983,
"learning_rate": 0.0005116287101996762,
"loss": 3.6986,
"step": 13750
},
{
"epoch": 1.4878706199460916,
"grad_norm": 0.6378118395805359,
"learning_rate": 0.0005113049109552077,
"loss": 3.7131,
"step": 13800
},
{
"epoch": 1.4932614555256065,
"grad_norm": 0.525088369846344,
"learning_rate": 0.0005109811117107393,
"loss": 3.7069,
"step": 13850
},
{
"epoch": 1.4986522911051212,
"grad_norm": 0.5727896094322205,
"learning_rate": 0.0005106573124662708,
"loss": 3.7206,
"step": 13900
},
{
"epoch": 1.5040431266846361,
"grad_norm": 0.5150138139724731,
"learning_rate": 0.0005103335132218025,
"loss": 3.6951,
"step": 13950
},
{
"epoch": 1.509433962264151,
"grad_norm": 0.5480995178222656,
"learning_rate": 0.000510009713977334,
"loss": 3.6953,
"step": 14000
},
{
"epoch": 1.509433962264151,
"eval_accuracy": 0.3542066758290025,
"eval_loss": 3.6530120372772217,
"eval_runtime": 183.8892,
"eval_samples_per_second": 97.945,
"eval_steps_per_second": 6.123,
"step": 14000
},
{
"epoch": 1.5148247978436657,
"grad_norm": 0.5239468812942505,
"learning_rate": 0.0005096859147328656,
"loss": 3.6865,
"step": 14050
},
{
"epoch": 1.5202156334231804,
"grad_norm": Infinity,
"learning_rate": 0.0005093685914732865,
"loss": 3.6845,
"step": 14100
},
{
"epoch": 1.5256064690026954,
"grad_norm": 0.5564777255058289,
"learning_rate": 0.0005090447922288181,
"loss": 3.6808,
"step": 14150
},
{
"epoch": 1.5309973045822103,
"grad_norm": 0.5200632810592651,
"learning_rate": 0.0005087209929843496,
"loss": 3.7141,
"step": 14200
},
{
"epoch": 1.536388140161725,
"grad_norm": 0.5934597849845886,
"learning_rate": 0.0005083971937398812,
"loss": 3.6875,
"step": 14250
},
{
"epoch": 1.54177897574124,
"grad_norm": 0.6403568387031555,
"learning_rate": 0.0005080733944954127,
"loss": 3.691,
"step": 14300
},
{
"epoch": 1.5471698113207548,
"grad_norm": 0.5741222500801086,
"learning_rate": 0.0005077495952509444,
"loss": 3.6722,
"step": 14350
},
{
"epoch": 1.5525606469002695,
"grad_norm": 0.568626344203949,
"learning_rate": 0.0005074257960064759,
"loss": 3.6874,
"step": 14400
},
{
"epoch": 1.5579514824797842,
"grad_norm": 0.569987416267395,
"learning_rate": 0.0005071019967620075,
"loss": 3.6921,
"step": 14450
},
{
"epoch": 1.5633423180592994,
"grad_norm": 0.57674640417099,
"learning_rate": 0.000506778197517539,
"loss": 3.6823,
"step": 14500
},
{
"epoch": 1.568733153638814,
"grad_norm": 0.6189175248146057,
"learning_rate": 0.0005064543982730707,
"loss": 3.6929,
"step": 14550
},
{
"epoch": 1.5741239892183287,
"grad_norm": 0.6155492067337036,
"learning_rate": 0.0005061305990286023,
"loss": 3.6864,
"step": 14600
},
{
"epoch": 1.5795148247978437,
"grad_norm": 0.5935265421867371,
"learning_rate": 0.0005058067997841338,
"loss": 3.6938,
"step": 14650
},
{
"epoch": 1.5849056603773586,
"grad_norm": 0.5613476634025574,
"learning_rate": 0.0005054830005396654,
"loss": 3.6797,
"step": 14700
},
{
"epoch": 1.5902964959568733,
"grad_norm": 0.5881309509277344,
"learning_rate": 0.0005051592012951969,
"loss": 3.6727,
"step": 14750
},
{
"epoch": 1.595687331536388,
"grad_norm": 0.5774725079536438,
"learning_rate": 0.0005048354020507286,
"loss": 3.6929,
"step": 14800
},
{
"epoch": 1.6010781671159031,
"grad_norm": 0.5917358994483948,
"learning_rate": 0.00050451160280626,
"loss": 3.6944,
"step": 14850
},
{
"epoch": 1.6064690026954178,
"grad_norm": 0.5740583539009094,
"learning_rate": 0.0005041878035617917,
"loss": 3.6845,
"step": 14900
},
{
"epoch": 1.6118598382749325,
"grad_norm": 0.5753791332244873,
"learning_rate": 0.0005038640043173232,
"loss": 3.6789,
"step": 14950
},
{
"epoch": 1.6172506738544474,
"grad_norm": 0.689713716506958,
"learning_rate": 0.0005035402050728548,
"loss": 3.6742,
"step": 15000
},
{
"epoch": 1.6172506738544474,
"eval_accuracy": 0.35600477246944434,
"eval_loss": 3.634129285812378,
"eval_runtime": 184.3974,
"eval_samples_per_second": 97.675,
"eval_steps_per_second": 6.106,
"step": 15000
},
{
"epoch": 1.6226415094339623,
"grad_norm": 0.525371253490448,
"learning_rate": 0.0005032164058283863,
"loss": 3.6757,
"step": 15050
},
{
"epoch": 1.628032345013477,
"grad_norm": 0.5436705946922302,
"learning_rate": 0.0005028926065839179,
"loss": 3.6721,
"step": 15100
},
{
"epoch": 1.633423180592992,
"grad_norm": 0.5768628120422363,
"learning_rate": 0.0005025688073394495,
"loss": 3.6721,
"step": 15150
},
{
"epoch": 1.6388140161725069,
"grad_norm": 0.5604543685913086,
"learning_rate": 0.0005022450080949811,
"loss": 3.6793,
"step": 15200
},
{
"epoch": 1.6442048517520216,
"grad_norm": 0.6073822379112244,
"learning_rate": 0.0005019212088505126,
"loss": 3.6818,
"step": 15250
},
{
"epoch": 1.6495956873315363,
"grad_norm": 0.5365363955497742,
"learning_rate": 0.0005015974096060442,
"loss": 3.6831,
"step": 15300
},
{
"epoch": 1.6549865229110512,
"grad_norm": 0.5398999452590942,
"learning_rate": 0.0005012736103615758,
"loss": 3.6679,
"step": 15350
},
{
"epoch": 1.6603773584905661,
"grad_norm": 0.6105483174324036,
"learning_rate": 0.0005009498111171074,
"loss": 3.6714,
"step": 15400
},
{
"epoch": 1.6657681940700808,
"grad_norm": 0.5510650873184204,
"learning_rate": 0.0005006260118726389,
"loss": 3.6703,
"step": 15450
},
{
"epoch": 1.6711590296495957,
"grad_norm": 0.5536068081855774,
"learning_rate": 0.0005003022126281705,
"loss": 3.6735,
"step": 15500
},
{
"epoch": 1.6765498652291106,
"grad_norm": 0.617607593536377,
"learning_rate": 0.000499978413383702,
"loss": 3.6657,
"step": 15550
},
{
"epoch": 1.6819407008086253,
"grad_norm": 0.5845173597335815,
"learning_rate": 0.0004996546141392336,
"loss": 3.6631,
"step": 15600
},
{
"epoch": 1.68733153638814,
"grad_norm": 0.5379621386528015,
"learning_rate": 0.0004993308148947651,
"loss": 3.6702,
"step": 15650
},
{
"epoch": 1.692722371967655,
"grad_norm": 0.5235825181007385,
"learning_rate": 0.0004990070156502968,
"loss": 3.6623,
"step": 15700
},
{
"epoch": 1.6981132075471699,
"grad_norm": 0.5610904097557068,
"learning_rate": 0.0004986832164058284,
"loss": 3.6558,
"step": 15750
},
{
"epoch": 1.7035040431266846,
"grad_norm": 0.6140375137329102,
"learning_rate": 0.0004983594171613599,
"loss": 3.6715,
"step": 15800
},
{
"epoch": 1.7088948787061995,
"grad_norm": 0.592974066734314,
"learning_rate": 0.0004980356179168915,
"loss": 3.6758,
"step": 15850
},
{
"epoch": 1.7142857142857144,
"grad_norm": 0.6089529395103455,
"learning_rate": 0.000497711818672423,
"loss": 3.7,
"step": 15900
},
{
"epoch": 1.719676549865229,
"grad_norm": 0.5230991244316101,
"learning_rate": 0.0004973880194279547,
"loss": 3.665,
"step": 15950
},
{
"epoch": 1.7250673854447438,
"grad_norm": 0.6368222832679749,
"learning_rate": 0.0004970642201834862,
"loss": 3.6729,
"step": 16000
},
{
"epoch": 1.7250673854447438,
"eval_accuracy": 0.35766227227305425,
"eval_loss": 3.614490032196045,
"eval_runtime": 183.7598,
"eval_samples_per_second": 98.014,
"eval_steps_per_second": 6.128,
"step": 16000
},
{
"epoch": 1.7304582210242587,
"grad_norm": 0.5433102250099182,
"learning_rate": 0.0004967404209390178,
"loss": 3.6426,
"step": 16050
},
{
"epoch": 1.7358490566037736,
"grad_norm": 0.49149322509765625,
"learning_rate": 0.0004964166216945493,
"loss": 3.6491,
"step": 16100
},
{
"epoch": 1.7412398921832883,
"grad_norm": 0.5774171352386475,
"learning_rate": 0.0004960992984349703,
"loss": 3.6648,
"step": 16150
},
{
"epoch": 1.7466307277628033,
"grad_norm": 0.6078404188156128,
"learning_rate": 0.0004957754991905018,
"loss": 3.6528,
"step": 16200
},
{
"epoch": 1.7520215633423182,
"grad_norm": 0.5405799150466919,
"learning_rate": 0.0004954516999460334,
"loss": 3.6609,
"step": 16250
},
{
"epoch": 1.7574123989218329,
"grad_norm": 0.6638331413269043,
"learning_rate": 0.0004951279007015649,
"loss": 3.6515,
"step": 16300
},
{
"epoch": 1.7628032345013476,
"grad_norm": 0.5499991178512573,
"learning_rate": 0.0004948041014570966,
"loss": 3.6474,
"step": 16350
},
{
"epoch": 1.7681940700808625,
"grad_norm": 0.5749475955963135,
"learning_rate": 0.0004944803022126281,
"loss": 3.6609,
"step": 16400
},
{
"epoch": 1.7735849056603774,
"grad_norm": 0.565037727355957,
"learning_rate": 0.0004941565029681597,
"loss": 3.6599,
"step": 16450
},
{
"epoch": 1.778975741239892,
"grad_norm": 0.5492637157440186,
"learning_rate": 0.0004938327037236912,
"loss": 3.6454,
"step": 16500
},
{
"epoch": 1.784366576819407,
"grad_norm": 0.5822088718414307,
"learning_rate": 0.0004935089044792229,
"loss": 3.6415,
"step": 16550
},
{
"epoch": 1.789757412398922,
"grad_norm": 0.5856910943984985,
"learning_rate": 0.0004931851052347544,
"loss": 3.6483,
"step": 16600
},
{
"epoch": 1.7951482479784366,
"grad_norm": 0.5660757422447205,
"learning_rate": 0.000492861305990286,
"loss": 3.6583,
"step": 16650
},
{
"epoch": 1.8005390835579513,
"grad_norm": 0.5575072169303894,
"learning_rate": 0.0004925375067458175,
"loss": 3.6494,
"step": 16700
},
{
"epoch": 1.8059299191374663,
"grad_norm": 0.5165206789970398,
"learning_rate": 0.0004922137075013491,
"loss": 3.6427,
"step": 16750
},
{
"epoch": 1.8113207547169812,
"grad_norm": 0.5528300404548645,
"learning_rate": 0.0004918899082568807,
"loss": 3.6484,
"step": 16800
},
{
"epoch": 1.8167115902964959,
"grad_norm": 0.5582894682884216,
"learning_rate": 0.0004915661090124123,
"loss": 3.6386,
"step": 16850
},
{
"epoch": 1.8221024258760108,
"grad_norm": 0.6211629509925842,
"learning_rate": 0.0004912423097679439,
"loss": 3.644,
"step": 16900
},
{
"epoch": 1.8274932614555257,
"grad_norm": 0.5531521439552307,
"learning_rate": 0.0004909185105234754,
"loss": 3.6501,
"step": 16950
},
{
"epoch": 1.8328840970350404,
"grad_norm": 0.5323461890220642,
"learning_rate": 0.000490594711279007,
"loss": 3.6575,
"step": 17000
},
{
"epoch": 1.8328840970350404,
"eval_accuracy": 0.3589206900197216,
"eval_loss": 3.6023108959198,
"eval_runtime": 184.2675,
"eval_samples_per_second": 97.744,
"eval_steps_per_second": 6.111,
"step": 17000
},
{
"epoch": 1.838274932614555,
"grad_norm": 0.5553791522979736,
"learning_rate": 0.0004902709120345385,
"loss": 3.6489,
"step": 17050
},
{
"epoch": 1.8436657681940702,
"grad_norm": 0.5244966745376587,
"learning_rate": 0.00048994711279007,
"loss": 3.6408,
"step": 17100
},
{
"epoch": 1.849056603773585,
"grad_norm": 0.604425847530365,
"learning_rate": 0.0004896233135456017,
"loss": 3.638,
"step": 17150
},
{
"epoch": 1.8544474393530996,
"grad_norm": 0.5427296757698059,
"learning_rate": 0.0004892995143011333,
"loss": 3.6387,
"step": 17200
},
{
"epoch": 1.8598382749326146,
"grad_norm": 0.5460442900657654,
"learning_rate": 0.0004889757150566648,
"loss": 3.6346,
"step": 17250
},
{
"epoch": 1.8652291105121295,
"grad_norm": 0.5109885334968567,
"learning_rate": 0.0004886519158121964,
"loss": 3.6593,
"step": 17300
},
{
"epoch": 1.8706199460916442,
"grad_norm": 0.5372109413146973,
"learning_rate": 0.000488328116567728,
"loss": 3.6606,
"step": 17350
},
{
"epoch": 1.8760107816711589,
"grad_norm": 0.5951082110404968,
"learning_rate": 0.0004880043173232595,
"loss": 3.6414,
"step": 17400
},
{
"epoch": 1.881401617250674,
"grad_norm": 0.5463950037956238,
"learning_rate": 0.0004876805180787911,
"loss": 3.6508,
"step": 17450
},
{
"epoch": 1.8867924528301887,
"grad_norm": 0.5363849401473999,
"learning_rate": 0.0004873567188343227,
"loss": 3.6352,
"step": 17500
},
{
"epoch": 1.8921832884097034,
"grad_norm": 0.6682934165000916,
"learning_rate": 0.0004870329195898542,
"loss": 3.649,
"step": 17550
},
{
"epoch": 1.8975741239892183,
"grad_norm": 0.5239797830581665,
"learning_rate": 0.00048670912034538583,
"loss": 3.6366,
"step": 17600
},
{
"epoch": 1.9029649595687332,
"grad_norm": 0.5734475255012512,
"learning_rate": 0.0004863853211009174,
"loss": 3.6361,
"step": 17650
},
{
"epoch": 1.908355795148248,
"grad_norm": 0.5510110855102539,
"learning_rate": 0.000486061521856449,
"loss": 3.6451,
"step": 17700
},
{
"epoch": 1.9137466307277629,
"grad_norm": 0.5507350564002991,
"learning_rate": 0.00048573772261198054,
"loss": 3.6351,
"step": 17750
},
{
"epoch": 1.9191374663072778,
"grad_norm": 0.6318540573120117,
"learning_rate": 0.00048541392336751214,
"loss": 3.6497,
"step": 17800
},
{
"epoch": 1.9245283018867925,
"grad_norm": 0.5466942191123962,
"learning_rate": 0.0004850901241230437,
"loss": 3.6569,
"step": 17850
},
{
"epoch": 1.9299191374663072,
"grad_norm": 0.5481278896331787,
"learning_rate": 0.0004847663248785753,
"loss": 3.634,
"step": 17900
},
{
"epoch": 1.935309973045822,
"grad_norm": 0.5548292398452759,
"learning_rate": 0.0004844425256341068,
"loss": 3.6441,
"step": 17950
},
{
"epoch": 1.940700808625337,
"grad_norm": 0.5729618668556213,
"learning_rate": 0.00048411872638963834,
"loss": 3.6395,
"step": 18000
},
{
"epoch": 1.940700808625337,
"eval_accuracy": 0.36053570654419603,
"eval_loss": 3.585592746734619,
"eval_runtime": 184.3811,
"eval_samples_per_second": 97.684,
"eval_steps_per_second": 6.107,
"step": 18000
},
{
"epoch": 1.9460916442048517,
"grad_norm": 0.5850827693939209,
"learning_rate": 0.00048379492714516995,
"loss": 3.6226,
"step": 18050
},
{
"epoch": 1.9514824797843666,
"grad_norm": 0.5491641163825989,
"learning_rate": 0.0004834711279007015,
"loss": 3.6519,
"step": 18100
},
{
"epoch": 1.9568733153638815,
"grad_norm": 0.5910980105400085,
"learning_rate": 0.00048315380464112245,
"loss": 3.6527,
"step": 18150
},
{
"epoch": 1.9622641509433962,
"grad_norm": 0.5838024616241455,
"learning_rate": 0.00048283000539665405,
"loss": 3.6389,
"step": 18200
},
{
"epoch": 1.967654986522911,
"grad_norm": 0.6026673913002014,
"learning_rate": 0.0004825062061521856,
"loss": 3.643,
"step": 18250
},
{
"epoch": 1.9730458221024259,
"grad_norm": 0.5433106422424316,
"learning_rate": 0.00048218240690771716,
"loss": 3.6222,
"step": 18300
},
{
"epoch": 1.9784366576819408,
"grad_norm": 0.7172867655754089,
"learning_rate": 0.00048185860766324876,
"loss": 3.6484,
"step": 18350
},
{
"epoch": 1.9838274932614555,
"grad_norm": 0.5216423273086548,
"learning_rate": 0.0004815348084187803,
"loss": 3.6226,
"step": 18400
},
{
"epoch": 1.9892183288409704,
"grad_norm": 0.5735328793525696,
"learning_rate": 0.0004812110091743119,
"loss": 3.6263,
"step": 18450
},
{
"epoch": 1.9946091644204853,
"grad_norm": 0.5877717137336731,
"learning_rate": 0.00048088720992984347,
"loss": 3.618,
"step": 18500
},
{
"epoch": 2.0,
"grad_norm": 1.1028350591659546,
"learning_rate": 0.00048056341068537507,
"loss": 3.6329,
"step": 18550
},
{
"epoch": 2.0053908355795147,
"grad_norm": 0.6162480711936951,
"learning_rate": 0.00048023961144090657,
"loss": 3.5468,
"step": 18600
},
{
"epoch": 2.01078167115903,
"grad_norm": 0.548809826374054,
"learning_rate": 0.00047991581219643817,
"loss": 3.5377,
"step": 18650
},
{
"epoch": 2.0161725067385445,
"grad_norm": 0.5910597443580627,
"learning_rate": 0.0004795920129519697,
"loss": 3.5497,
"step": 18700
},
{
"epoch": 2.0215633423180592,
"grad_norm": 0.5971364974975586,
"learning_rate": 0.0004792682137075013,
"loss": 3.5397,
"step": 18750
},
{
"epoch": 2.026954177897574,
"grad_norm": 0.5920816659927368,
"learning_rate": 0.0004789444144630329,
"loss": 3.5478,
"step": 18800
},
{
"epoch": 2.032345013477089,
"grad_norm": 0.5575721263885498,
"learning_rate": 0.00047862061521856443,
"loss": 3.5451,
"step": 18850
},
{
"epoch": 2.0377358490566038,
"grad_norm": 0.586161196231842,
"learning_rate": 0.00047829681597409603,
"loss": 3.5571,
"step": 18900
},
{
"epoch": 2.0431266846361185,
"grad_norm": 0.5850305557250977,
"learning_rate": 0.0004779730167296276,
"loss": 3.5524,
"step": 18950
},
{
"epoch": 2.0485175202156336,
"grad_norm": 0.5819904208183289,
"learning_rate": 0.0004776492174851592,
"loss": 3.5539,
"step": 19000
},
{
"epoch": 2.0485175202156336,
"eval_accuracy": 0.362115845491682,
"eval_loss": 3.576646566390991,
"eval_runtime": 184.0348,
"eval_samples_per_second": 97.867,
"eval_steps_per_second": 6.118,
"step": 19000
},
{
"epoch": 2.0539083557951483,
"grad_norm": 0.5057418942451477,
"learning_rate": 0.00047732541824069074,
"loss": 3.5636,
"step": 19050
},
{
"epoch": 2.059299191374663,
"grad_norm": 0.5615925192832947,
"learning_rate": 0.0004770016189962223,
"loss": 3.5547,
"step": 19100
},
{
"epoch": 2.0646900269541777,
"grad_norm": 0.54108726978302,
"learning_rate": 0.0004766778197517539,
"loss": 3.5521,
"step": 19150
},
{
"epoch": 2.070080862533693,
"grad_norm": 0.5467185378074646,
"learning_rate": 0.0004763540205072854,
"loss": 3.5706,
"step": 19200
},
{
"epoch": 2.0754716981132075,
"grad_norm": 0.5604600310325623,
"learning_rate": 0.00047603022126281705,
"loss": 3.5594,
"step": 19250
},
{
"epoch": 2.0808625336927222,
"grad_norm": 0.5926553010940552,
"learning_rate": 0.00047570642201834855,
"loss": 3.5507,
"step": 19300
},
{
"epoch": 2.0862533692722374,
"grad_norm": 0.6105896234512329,
"learning_rate": 0.00047538262277388015,
"loss": 3.5677,
"step": 19350
},
{
"epoch": 2.091644204851752,
"grad_norm": 0.5920027494430542,
"learning_rate": 0.0004750588235294117,
"loss": 3.5587,
"step": 19400
},
{
"epoch": 2.0970350404312668,
"grad_norm": 0.6913124322891235,
"learning_rate": 0.0004747350242849433,
"loss": 3.5423,
"step": 19450
},
{
"epoch": 2.1024258760107815,
"grad_norm": 0.5646089911460876,
"learning_rate": 0.00047441122504047486,
"loss": 3.5663,
"step": 19500
},
{
"epoch": 2.1078167115902966,
"grad_norm": 0.590248703956604,
"learning_rate": 0.0004740874257960064,
"loss": 3.5576,
"step": 19550
},
{
"epoch": 2.1132075471698113,
"grad_norm": 0.5555785298347473,
"learning_rate": 0.000473763626551538,
"loss": 3.554,
"step": 19600
},
{
"epoch": 2.118598382749326,
"grad_norm": 0.5776288509368896,
"learning_rate": 0.00047343982730706956,
"loss": 3.5566,
"step": 19650
},
{
"epoch": 2.123989218328841,
"grad_norm": 0.5516104698181152,
"learning_rate": 0.00047311602806260117,
"loss": 3.5478,
"step": 19700
},
{
"epoch": 2.129380053908356,
"grad_norm": 0.5960516929626465,
"learning_rate": 0.0004727922288181327,
"loss": 3.5392,
"step": 19750
},
{
"epoch": 2.1347708894878705,
"grad_norm": 0.545975387096405,
"learning_rate": 0.0004724684295736643,
"loss": 3.5623,
"step": 19800
},
{
"epoch": 2.1401617250673857,
"grad_norm": 0.5498135089874268,
"learning_rate": 0.0004721446303291959,
"loss": 3.5548,
"step": 19850
},
{
"epoch": 2.1455525606469004,
"grad_norm": 0.5613714456558228,
"learning_rate": 0.0004718208310847275,
"loss": 3.5581,
"step": 19900
},
{
"epoch": 2.150943396226415,
"grad_norm": 0.5755128264427185,
"learning_rate": 0.000471497031840259,
"loss": 3.5488,
"step": 19950
},
{
"epoch": 2.1563342318059298,
"grad_norm": 0.573879063129425,
"learning_rate": 0.0004711732325957905,
"loss": 3.5538,
"step": 20000
},
{
"epoch": 2.1563342318059298,
"eval_accuracy": 0.3630740553093179,
"eval_loss": 3.568387031555176,
"eval_runtime": 184.4179,
"eval_samples_per_second": 97.664,
"eval_steps_per_second": 6.106,
"step": 20000
},
{
"epoch": 2.161725067385445,
"grad_norm": 0.5555899143218994,
"learning_rate": 0.00047084943335132213,
"loss": 3.5542,
"step": 20050
},
{
"epoch": 2.1671159029649596,
"grad_norm": 0.5866010189056396,
"learning_rate": 0.0004705256341068537,
"loss": 3.5528,
"step": 20100
},
{
"epoch": 2.1725067385444743,
"grad_norm": 0.5684776306152344,
"learning_rate": 0.00047020831084727463,
"loss": 3.5624,
"step": 20150
},
{
"epoch": 2.177897574123989,
"grad_norm": 0.5335924625396729,
"learning_rate": 0.00046988451160280624,
"loss": 3.5632,
"step": 20200
},
{
"epoch": 2.183288409703504,
"grad_norm": 0.5421379804611206,
"learning_rate": 0.0004695607123583378,
"loss": 3.5687,
"step": 20250
},
{
"epoch": 2.188679245283019,
"grad_norm": 0.6612055897712708,
"learning_rate": 0.00046923691311386934,
"loss": 3.5511,
"step": 20300
},
{
"epoch": 2.1940700808625335,
"grad_norm": 0.5414841175079346,
"learning_rate": 0.00046891311386940094,
"loss": 3.5449,
"step": 20350
},
{
"epoch": 2.1994609164420487,
"grad_norm": 0.5854169726371765,
"learning_rate": 0.0004685893146249325,
"loss": 3.5581,
"step": 20400
},
{
"epoch": 2.2048517520215634,
"grad_norm": 0.7372918128967285,
"learning_rate": 0.0004682655153804641,
"loss": 3.5519,
"step": 20450
},
{
"epoch": 2.210242587601078,
"grad_norm": 0.5372480154037476,
"learning_rate": 0.00046794171613599565,
"loss": 3.5689,
"step": 20500
},
{
"epoch": 2.215633423180593,
"grad_norm": 0.6007562279701233,
"learning_rate": 0.00046761791689152725,
"loss": 3.5656,
"step": 20550
},
{
"epoch": 2.221024258760108,
"grad_norm": 0.5941622257232666,
"learning_rate": 0.00046729411764705875,
"loss": 3.559,
"step": 20600
},
{
"epoch": 2.2264150943396226,
"grad_norm": 0.5837433338165283,
"learning_rate": 0.0004669703184025904,
"loss": 3.5678,
"step": 20650
},
{
"epoch": 2.2318059299191373,
"grad_norm": 0.5905625224113464,
"learning_rate": 0.0004666465191581219,
"loss": 3.5512,
"step": 20700
},
{
"epoch": 2.2371967654986524,
"grad_norm": 0.5736416578292847,
"learning_rate": 0.00046632271991365346,
"loss": 3.5692,
"step": 20750
},
{
"epoch": 2.242587601078167,
"grad_norm": 0.5766247510910034,
"learning_rate": 0.00046599892066918506,
"loss": 3.5637,
"step": 20800
},
{
"epoch": 2.247978436657682,
"grad_norm": 0.5899847745895386,
"learning_rate": 0.0004656751214247166,
"loss": 3.5649,
"step": 20850
},
{
"epoch": 2.2533692722371965,
"grad_norm": 0.542995810508728,
"learning_rate": 0.0004653513221802482,
"loss": 3.5554,
"step": 20900
},
{
"epoch": 2.2587601078167117,
"grad_norm": 0.5832574367523193,
"learning_rate": 0.00046502752293577977,
"loss": 3.5563,
"step": 20950
},
{
"epoch": 2.2641509433962264,
"grad_norm": 0.598743736743927,
"learning_rate": 0.00046470372369131137,
"loss": 3.5554,
"step": 21000
},
{
"epoch": 2.2641509433962264,
"eval_accuracy": 0.36408865597572954,
"eval_loss": 3.558624029159546,
"eval_runtime": 184.2753,
"eval_samples_per_second": 97.74,
"eval_steps_per_second": 6.11,
"step": 21000
},
{
"epoch": 2.269541778975741,
"grad_norm": 0.568676233291626,
"learning_rate": 0.0004643799244468429,
"loss": 3.5517,
"step": 21050
},
{
"epoch": 2.274932614555256,
"grad_norm": 0.5959727168083191,
"learning_rate": 0.0004640561252023745,
"loss": 3.5417,
"step": 21100
},
{
"epoch": 2.280323450134771,
"grad_norm": 0.633328914642334,
"learning_rate": 0.0004637323259579061,
"loss": 3.568,
"step": 21150
},
{
"epoch": 2.2857142857142856,
"grad_norm": 0.5704327821731567,
"learning_rate": 0.0004634085267134376,
"loss": 3.5483,
"step": 21200
},
{
"epoch": 2.2911051212938007,
"grad_norm": 0.6142733097076416,
"learning_rate": 0.00046308472746896923,
"loss": 3.5451,
"step": 21250
},
{
"epoch": 2.2964959568733154,
"grad_norm": 0.5728433728218079,
"learning_rate": 0.00046276092822450073,
"loss": 3.5549,
"step": 21300
},
{
"epoch": 2.30188679245283,
"grad_norm": 0.5286217927932739,
"learning_rate": 0.00046243712898003233,
"loss": 3.5318,
"step": 21350
},
{
"epoch": 2.3072776280323453,
"grad_norm": 0.5701291561126709,
"learning_rate": 0.0004621133297355639,
"loss": 3.5699,
"step": 21400
},
{
"epoch": 2.31266846361186,
"grad_norm": 0.582324206829071,
"learning_rate": 0.0004617895304910955,
"loss": 3.5287,
"step": 21450
},
{
"epoch": 2.3180592991913747,
"grad_norm": 0.551391065120697,
"learning_rate": 0.00046146573124662704,
"loss": 3.5609,
"step": 21500
},
{
"epoch": 2.3234501347708894,
"grad_norm": 0.597589910030365,
"learning_rate": 0.00046114193200215864,
"loss": 3.5597,
"step": 21550
},
{
"epoch": 2.3288409703504045,
"grad_norm": 0.5848575830459595,
"learning_rate": 0.0004608181327576902,
"loss": 3.5327,
"step": 21600
},
{
"epoch": 2.334231805929919,
"grad_norm": 0.5785087943077087,
"learning_rate": 0.00046049433351322175,
"loss": 3.5537,
"step": 21650
},
{
"epoch": 2.339622641509434,
"grad_norm": 0.5787850022315979,
"learning_rate": 0.00046017053426875335,
"loss": 3.5665,
"step": 21700
},
{
"epoch": 2.3450134770889486,
"grad_norm": 0.6265482306480408,
"learning_rate": 0.0004598467350242849,
"loss": 3.5491,
"step": 21750
},
{
"epoch": 2.3504043126684637,
"grad_norm": 0.5768268704414368,
"learning_rate": 0.0004595229357798165,
"loss": 3.5603,
"step": 21800
},
{
"epoch": 2.3557951482479784,
"grad_norm": 0.6471909880638123,
"learning_rate": 0.00045919913653534806,
"loss": 3.5615,
"step": 21850
},
{
"epoch": 2.361185983827493,
"grad_norm": 0.5320773720741272,
"learning_rate": 0.00045887533729087966,
"loss": 3.5529,
"step": 21900
},
{
"epoch": 2.3665768194070083,
"grad_norm": 0.5688530802726746,
"learning_rate": 0.00045855153804641116,
"loss": 3.5451,
"step": 21950
},
{
"epoch": 2.371967654986523,
"grad_norm": 0.5354400277137756,
"learning_rate": 0.0004582277388019427,
"loss": 3.5401,
"step": 22000
},
{
"epoch": 2.371967654986523,
"eval_accuracy": 0.3651116229145029,
"eval_loss": 3.5474777221679688,
"eval_runtime": 184.5356,
"eval_samples_per_second": 97.602,
"eval_steps_per_second": 6.102,
"step": 22000
},
{
"epoch": 2.3773584905660377,
"grad_norm": 0.5725575089454651,
"learning_rate": 0.0004579039395574743,
"loss": 3.5412,
"step": 22050
},
{
"epoch": 2.382749326145553,
"grad_norm": 0.5488483905792236,
"learning_rate": 0.00045758014031300586,
"loss": 3.5605,
"step": 22100
},
{
"epoch": 2.3881401617250675,
"grad_norm": 0.5680898427963257,
"learning_rate": 0.00045725634106853747,
"loss": 3.5529,
"step": 22150
},
{
"epoch": 2.393530997304582,
"grad_norm": 0.6047888398170471,
"learning_rate": 0.000456932541824069,
"loss": 3.5384,
"step": 22200
},
{
"epoch": 2.398921832884097,
"grad_norm": 0.7227292656898499,
"learning_rate": 0.00045661521856448997,
"loss": 3.532,
"step": 22250
},
{
"epoch": 2.404312668463612,
"grad_norm": 0.5926851034164429,
"learning_rate": 0.0004562914193200216,
"loss": 3.5607,
"step": 22300
},
{
"epoch": 2.4097035040431267,
"grad_norm": 0.5851311683654785,
"learning_rate": 0.0004559676200755531,
"loss": 3.5696,
"step": 22350
},
{
"epoch": 2.4150943396226414,
"grad_norm": 0.6393628120422363,
"learning_rate": 0.0004556438208310847,
"loss": 3.5525,
"step": 22400
},
{
"epoch": 2.420485175202156,
"grad_norm": 0.5410474538803101,
"learning_rate": 0.0004553200215866163,
"loss": 3.5603,
"step": 22450
},
{
"epoch": 2.4258760107816713,
"grad_norm": 0.5866028666496277,
"learning_rate": 0.00045499622234214783,
"loss": 3.5588,
"step": 22500
},
{
"epoch": 2.431266846361186,
"grad_norm": 0.560801088809967,
"learning_rate": 0.00045467242309767944,
"loss": 3.5384,
"step": 22550
},
{
"epoch": 2.4366576819407006,
"grad_norm": 0.5960171222686768,
"learning_rate": 0.00045434862385321093,
"loss": 3.5546,
"step": 22600
},
{
"epoch": 2.442048517520216,
"grad_norm": 0.5768381953239441,
"learning_rate": 0.0004540248246087426,
"loss": 3.5507,
"step": 22650
},
{
"epoch": 2.4474393530997305,
"grad_norm": 0.606055498123169,
"learning_rate": 0.0004537010253642741,
"loss": 3.5572,
"step": 22700
},
{
"epoch": 2.452830188679245,
"grad_norm": 0.6579552888870239,
"learning_rate": 0.00045337722611980564,
"loss": 3.5469,
"step": 22750
},
{
"epoch": 2.4582210242587603,
"grad_norm": 0.5266789793968201,
"learning_rate": 0.00045305342687533724,
"loss": 3.5444,
"step": 22800
},
{
"epoch": 2.463611859838275,
"grad_norm": 0.5739652514457703,
"learning_rate": 0.0004527296276308688,
"loss": 3.5348,
"step": 22850
},
{
"epoch": 2.4690026954177897,
"grad_norm": 0.5625444650650024,
"learning_rate": 0.0004524058283864004,
"loss": 3.547,
"step": 22900
},
{
"epoch": 2.4743935309973044,
"grad_norm": 0.6308554410934448,
"learning_rate": 0.00045208202914193195,
"loss": 3.5318,
"step": 22950
},
{
"epoch": 2.4797843665768196,
"grad_norm": 0.6012802720069885,
"learning_rate": 0.00045175822989746355,
"loss": 3.5483,
"step": 23000
},
{
"epoch": 2.4797843665768196,
"eval_accuracy": 0.3660999295820634,
"eval_loss": 3.5364835262298584,
"eval_runtime": 184.2295,
"eval_samples_per_second": 97.764,
"eval_steps_per_second": 6.112,
"step": 23000
},
{
"epoch": 2.4851752021563343,
"grad_norm": 0.6175972819328308,
"learning_rate": 0.0004514344306529951,
"loss": 3.5536,
"step": 23050
},
{
"epoch": 2.490566037735849,
"grad_norm": 0.5650850534439087,
"learning_rate": 0.0004511106314085267,
"loss": 3.551,
"step": 23100
},
{
"epoch": 2.4959568733153636,
"grad_norm": 0.6012078523635864,
"learning_rate": 0.00045078683216405826,
"loss": 3.5631,
"step": 23150
},
{
"epoch": 2.501347708894879,
"grad_norm": 0.5973430871963501,
"learning_rate": 0.00045046303291958976,
"loss": 3.5552,
"step": 23200
},
{
"epoch": 2.5067385444743935,
"grad_norm": 0.6140353083610535,
"learning_rate": 0.0004501392336751214,
"loss": 3.5546,
"step": 23250
},
{
"epoch": 2.512129380053908,
"grad_norm": 0.53867107629776,
"learning_rate": 0.0004498154344306529,
"loss": 3.5518,
"step": 23300
},
{
"epoch": 2.5175202156334233,
"grad_norm": 0.5804443359375,
"learning_rate": 0.0004494916351861845,
"loss": 3.5514,
"step": 23350
},
{
"epoch": 2.522911051212938,
"grad_norm": 0.616215705871582,
"learning_rate": 0.00044917431192660547,
"loss": 3.5292,
"step": 23400
},
{
"epoch": 2.5283018867924527,
"grad_norm": 0.6087223887443542,
"learning_rate": 0.000448850512682137,
"loss": 3.546,
"step": 23450
},
{
"epoch": 2.533692722371968,
"grad_norm": 0.5890523791313171,
"learning_rate": 0.0004485267134376686,
"loss": 3.5408,
"step": 23500
},
{
"epoch": 2.5390835579514826,
"grad_norm": 0.5296126008033752,
"learning_rate": 0.0004482029141932002,
"loss": 3.5421,
"step": 23550
},
{
"epoch": 2.5444743935309972,
"grad_norm": 0.5980884432792664,
"learning_rate": 0.0004478791149487317,
"loss": 3.5469,
"step": 23600
},
{
"epoch": 2.5498652291105124,
"grad_norm": 0.5312870144844055,
"learning_rate": 0.00044755531570426333,
"loss": 3.5344,
"step": 23650
},
{
"epoch": 2.555256064690027,
"grad_norm": 0.5587490797042847,
"learning_rate": 0.0004472315164597949,
"loss": 3.5226,
"step": 23700
},
{
"epoch": 2.560646900269542,
"grad_norm": 0.626533567905426,
"learning_rate": 0.0004469077172153265,
"loss": 3.5313,
"step": 23750
},
{
"epoch": 2.5660377358490565,
"grad_norm": 0.5318872332572937,
"learning_rate": 0.00044658391797085804,
"loss": 3.55,
"step": 23800
},
{
"epoch": 2.571428571428571,
"grad_norm": 0.5714778900146484,
"learning_rate": 0.00044626011872638964,
"loss": 3.5389,
"step": 23850
},
{
"epoch": 2.5768194070080863,
"grad_norm": 0.5694332718849182,
"learning_rate": 0.0004459363194819212,
"loss": 3.5375,
"step": 23900
},
{
"epoch": 2.582210242587601,
"grad_norm": 0.6230406165122986,
"learning_rate": 0.0004456125202374527,
"loss": 3.5306,
"step": 23950
},
{
"epoch": 2.5876010781671157,
"grad_norm": 0.5774886608123779,
"learning_rate": 0.0004452887209929843,
"loss": 3.5354,
"step": 24000
},
{
"epoch": 2.5876010781671157,
"eval_accuracy": 0.36738714034400155,
"eval_loss": 3.528362989425659,
"eval_runtime": 184.3557,
"eval_samples_per_second": 97.697,
"eval_steps_per_second": 6.108,
"step": 24000
},
{
"epoch": 2.592991913746631,
"grad_norm": 0.5960202217102051,
"learning_rate": 0.00044496492174851584,
"loss": 3.5615,
"step": 24050
},
{
"epoch": 2.5983827493261455,
"grad_norm": 0.5864916443824768,
"learning_rate": 0.00044464112250404745,
"loss": 3.5354,
"step": 24100
},
{
"epoch": 2.6037735849056602,
"grad_norm": 0.5632309913635254,
"learning_rate": 0.000444317323259579,
"loss": 3.5472,
"step": 24150
},
{
"epoch": 2.6091644204851754,
"grad_norm": 0.5750573873519897,
"learning_rate": 0.0004439935240151106,
"loss": 3.5336,
"step": 24200
},
{
"epoch": 2.61455525606469,
"grad_norm": 0.5779033303260803,
"learning_rate": 0.00044366972477064215,
"loss": 3.5506,
"step": 24250
},
{
"epoch": 2.6199460916442048,
"grad_norm": 0.5961576104164124,
"learning_rate": 0.00044334592552617376,
"loss": 3.5435,
"step": 24300
},
{
"epoch": 2.62533692722372,
"grad_norm": 0.576537549495697,
"learning_rate": 0.0004430221262817053,
"loss": 3.5384,
"step": 24350
},
{
"epoch": 2.6307277628032346,
"grad_norm": 0.6238811016082764,
"learning_rate": 0.00044269832703723686,
"loss": 3.5395,
"step": 24400
},
{
"epoch": 2.6361185983827493,
"grad_norm": 0.5383691191673279,
"learning_rate": 0.00044237452779276846,
"loss": 3.5393,
"step": 24450
},
{
"epoch": 2.641509433962264,
"grad_norm": 0.5712140798568726,
"learning_rate": 0.0004420507285483,
"loss": 3.5464,
"step": 24500
},
{
"epoch": 2.6469002695417787,
"grad_norm": 0.6225411891937256,
"learning_rate": 0.0004417269293038316,
"loss": 3.5312,
"step": 24550
},
{
"epoch": 2.652291105121294,
"grad_norm": 0.6105319261550903,
"learning_rate": 0.0004414031300593631,
"loss": 3.5402,
"step": 24600
},
{
"epoch": 2.6576819407008085,
"grad_norm": 0.571915328502655,
"learning_rate": 0.0004410793308148948,
"loss": 3.536,
"step": 24650
},
{
"epoch": 2.6630727762803232,
"grad_norm": 0.597759485244751,
"learning_rate": 0.00044075553157042627,
"loss": 3.542,
"step": 24700
},
{
"epoch": 2.6684636118598384,
"grad_norm": 0.5761103630065918,
"learning_rate": 0.0004404317323259579,
"loss": 3.5578,
"step": 24750
},
{
"epoch": 2.673854447439353,
"grad_norm": 0.5313440561294556,
"learning_rate": 0.0004401079330814894,
"loss": 3.5508,
"step": 24800
},
{
"epoch": 2.6792452830188678,
"grad_norm": 0.589394748210907,
"learning_rate": 0.000439784133837021,
"loss": 3.5511,
"step": 24850
},
{
"epoch": 2.684636118598383,
"grad_norm": 0.647866427898407,
"learning_rate": 0.0004394603345925526,
"loss": 3.5236,
"step": 24900
},
{
"epoch": 2.6900269541778976,
"grad_norm": 0.5529329180717468,
"learning_rate": 0.00043913653534808413,
"loss": 3.5424,
"step": 24950
},
{
"epoch": 2.6954177897574123,
"grad_norm": 0.5735082030296326,
"learning_rate": 0.00043881273610361574,
"loss": 3.522,
"step": 25000
},
{
"epoch": 2.6954177897574123,
"eval_accuracy": 0.3684259706043958,
"eval_loss": 3.5166385173797607,
"eval_runtime": 184.365,
"eval_samples_per_second": 97.692,
"eval_steps_per_second": 6.107,
"step": 25000
},
{
"epoch": 2.7008086253369274,
"grad_norm": 0.5914623737335205,
"learning_rate": 0.0004384889368591473,
"loss": 3.5339,
"step": 25050
},
{
"epoch": 2.706199460916442,
"grad_norm": 0.5720333456993103,
"learning_rate": 0.0004381651376146789,
"loss": 3.5364,
"step": 25100
},
{
"epoch": 2.711590296495957,
"grad_norm": 0.5796301364898682,
"learning_rate": 0.00043784133837021044,
"loss": 3.5378,
"step": 25150
},
{
"epoch": 2.7169811320754715,
"grad_norm": 0.6207363605499268,
"learning_rate": 0.00043751753912574205,
"loss": 3.5378,
"step": 25200
},
{
"epoch": 2.7223719676549867,
"grad_norm": 0.6100524067878723,
"learning_rate": 0.0004371937398812736,
"loss": 3.53,
"step": 25250
},
{
"epoch": 2.7277628032345014,
"grad_norm": 0.5828701257705688,
"learning_rate": 0.0004368699406368051,
"loss": 3.5377,
"step": 25300
},
{
"epoch": 2.733153638814016,
"grad_norm": 0.729020357131958,
"learning_rate": 0.0004365461413923367,
"loss": 3.5345,
"step": 25350
},
{
"epoch": 2.7385444743935308,
"grad_norm": 0.5860224366188049,
"learning_rate": 0.00043622234214786825,
"loss": 3.5276,
"step": 25400
},
{
"epoch": 2.743935309973046,
"grad_norm": 0.6120912432670593,
"learning_rate": 0.00043589854290339985,
"loss": 3.5425,
"step": 25450
},
{
"epoch": 2.7493261455525606,
"grad_norm": 0.6055934429168701,
"learning_rate": 0.0004355747436589314,
"loss": 3.522,
"step": 25500
},
{
"epoch": 2.7547169811320753,
"grad_norm": 0.6246229410171509,
"learning_rate": 0.000435250944414463,
"loss": 3.5148,
"step": 25550
},
{
"epoch": 2.7601078167115904,
"grad_norm": 0.5672488212585449,
"learning_rate": 0.00043492714516999456,
"loss": 3.5203,
"step": 25600
},
{
"epoch": 2.765498652291105,
"grad_norm": 0.5845912098884583,
"learning_rate": 0.0004346033459255261,
"loss": 3.5249,
"step": 25650
},
{
"epoch": 2.77088948787062,
"grad_norm": 0.5401322841644287,
"learning_rate": 0.0004342795466810577,
"loss": 3.5119,
"step": 25700
},
{
"epoch": 2.776280323450135,
"grad_norm": 0.596645712852478,
"learning_rate": 0.00043395574743658927,
"loss": 3.5174,
"step": 25750
},
{
"epoch": 2.7816711590296497,
"grad_norm": 0.6043346524238586,
"learning_rate": 0.00043363194819212087,
"loss": 3.5303,
"step": 25800
},
{
"epoch": 2.7870619946091644,
"grad_norm": 0.562282383441925,
"learning_rate": 0.0004333081489476524,
"loss": 3.517,
"step": 25850
},
{
"epoch": 2.7924528301886795,
"grad_norm": 0.5671759843826294,
"learning_rate": 0.000432984349703184,
"loss": 3.5207,
"step": 25900
},
{
"epoch": 2.797843665768194,
"grad_norm": 0.5863797068595886,
"learning_rate": 0.0004326605504587155,
"loss": 3.5321,
"step": 25950
},
{
"epoch": 2.803234501347709,
"grad_norm": 0.6261507868766785,
"learning_rate": 0.0004323367512142472,
"loss": 3.5353,
"step": 26000
},
{
"epoch": 2.803234501347709,
"eval_accuracy": 0.36920142626472774,
"eval_loss": 3.5082848072052,
"eval_runtime": 184.5035,
"eval_samples_per_second": 97.619,
"eval_steps_per_second": 6.103,
"step": 26000
},
{
"epoch": 2.8086253369272236,
"grad_norm": 0.6280455589294434,
"learning_rate": 0.0004320129519697787,
"loss": 3.537,
"step": 26050
},
{
"epoch": 2.8140161725067383,
"grad_norm": 0.5723023414611816,
"learning_rate": 0.00043168915272531023,
"loss": 3.5175,
"step": 26100
},
{
"epoch": 2.8194070080862534,
"grad_norm": 0.6172739863395691,
"learning_rate": 0.00043136535348084183,
"loss": 3.5238,
"step": 26150
},
{
"epoch": 2.824797843665768,
"grad_norm": 0.636669397354126,
"learning_rate": 0.0004310415542363734,
"loss": 3.5213,
"step": 26200
},
{
"epoch": 2.830188679245283,
"grad_norm": 0.5924001336097717,
"learning_rate": 0.000430717754991905,
"loss": 3.5148,
"step": 26250
},
{
"epoch": 2.835579514824798,
"grad_norm": 0.5929787755012512,
"learning_rate": 0.00043039395574743654,
"loss": 3.5306,
"step": 26300
},
{
"epoch": 2.8409703504043127,
"grad_norm": 0.6071686744689941,
"learning_rate": 0.00043007015650296814,
"loss": 3.5364,
"step": 26350
},
{
"epoch": 2.8463611859838274,
"grad_norm": 0.6135595440864563,
"learning_rate": 0.0004297463572584997,
"loss": 3.5232,
"step": 26400
},
{
"epoch": 2.8517520215633425,
"grad_norm": 0.5787630081176758,
"learning_rate": 0.0004294225580140313,
"loss": 3.5253,
"step": 26450
},
{
"epoch": 2.857142857142857,
"grad_norm": 0.6400113701820374,
"learning_rate": 0.00042909875876956285,
"loss": 3.5219,
"step": 26500
},
{
"epoch": 2.862533692722372,
"grad_norm": 0.6119537949562073,
"learning_rate": 0.0004287749595250944,
"loss": 3.5138,
"step": 26550
},
{
"epoch": 2.867924528301887,
"grad_norm": 0.5760701894760132,
"learning_rate": 0.000428451160280626,
"loss": 3.525,
"step": 26600
},
{
"epoch": 2.8733153638814017,
"grad_norm": 0.5865446925163269,
"learning_rate": 0.0004281273610361575,
"loss": 3.521,
"step": 26650
},
{
"epoch": 2.8787061994609164,
"grad_norm": 0.5874311923980713,
"learning_rate": 0.0004278035617916891,
"loss": 3.5172,
"step": 26700
},
{
"epoch": 2.884097035040431,
"grad_norm": 0.5811386704444885,
"learning_rate": 0.00042747976254722066,
"loss": 3.5365,
"step": 26750
},
{
"epoch": 2.889487870619946,
"grad_norm": 0.6299395561218262,
"learning_rate": 0.00042715596330275226,
"loss": 3.5228,
"step": 26800
},
{
"epoch": 2.894878706199461,
"grad_norm": 0.5832048654556274,
"learning_rate": 0.0004268321640582838,
"loss": 3.5181,
"step": 26850
},
{
"epoch": 2.9002695417789757,
"grad_norm": 0.6102468967437744,
"learning_rate": 0.0004265083648138154,
"loss": 3.5412,
"step": 26900
},
{
"epoch": 2.9056603773584904,
"grad_norm": 0.5789380073547363,
"learning_rate": 0.00042618456556934697,
"loss": 3.5335,
"step": 26950
},
{
"epoch": 2.9110512129380055,
"grad_norm": 0.5606870055198669,
"learning_rate": 0.0004258607663248785,
"loss": 3.5173,
"step": 27000
},
{
"epoch": 2.9110512129380055,
"eval_accuracy": 0.3700289266583229,
"eval_loss": 3.499816417694092,
"eval_runtime": 184.2079,
"eval_samples_per_second": 97.775,
"eval_steps_per_second": 6.113,
"step": 27000
},
{
"epoch": 2.91644204851752,
"grad_norm": 0.6432551145553589,
"learning_rate": 0.0004255369670804101,
"loss": 3.5366,
"step": 27050
},
{
"epoch": 2.921832884097035,
"grad_norm": 0.5721651911735535,
"learning_rate": 0.0004252131678359417,
"loss": 3.5403,
"step": 27100
},
{
"epoch": 2.92722371967655,
"grad_norm": 0.5896191596984863,
"learning_rate": 0.0004248893685914733,
"loss": 3.5244,
"step": 27150
},
{
"epoch": 2.9326145552560647,
"grad_norm": 0.585056483745575,
"learning_rate": 0.00042456556934700483,
"loss": 3.5154,
"step": 27200
},
{
"epoch": 2.9380053908355794,
"grad_norm": 0.5735598206520081,
"learning_rate": 0.00042424177010253643,
"loss": 3.5273,
"step": 27250
},
{
"epoch": 2.9433962264150946,
"grad_norm": 0.6206878423690796,
"learning_rate": 0.00042391797085806793,
"loss": 3.5241,
"step": 27300
},
{
"epoch": 2.9487870619946093,
"grad_norm": 0.619541347026825,
"learning_rate": 0.0004235941716135995,
"loss": 3.5178,
"step": 27350
},
{
"epoch": 2.954177897574124,
"grad_norm": 0.5657435059547424,
"learning_rate": 0.0004232703723691311,
"loss": 3.5188,
"step": 27400
},
{
"epoch": 2.9595687331536387,
"grad_norm": 0.5487257838249207,
"learning_rate": 0.0004229595250944414,
"loss": 3.5161,
"step": 27450
},
{
"epoch": 2.964959568733154,
"grad_norm": 0.5736271142959595,
"learning_rate": 0.000422635725849973,
"loss": 3.5239,
"step": 27500
},
{
"epoch": 2.9703504043126685,
"grad_norm": 0.5725128650665283,
"learning_rate": 0.00042231192660550454,
"loss": 3.5256,
"step": 27550
},
{
"epoch": 2.975741239892183,
"grad_norm": 0.570732057094574,
"learning_rate": 0.0004219881273610361,
"loss": 3.5265,
"step": 27600
},
{
"epoch": 2.981132075471698,
"grad_norm": 0.5639909505844116,
"learning_rate": 0.0004216643281165677,
"loss": 3.5289,
"step": 27650
},
{
"epoch": 2.986522911051213,
"grad_norm": 0.589285135269165,
"learning_rate": 0.00042134052887209925,
"loss": 3.5202,
"step": 27700
},
{
"epoch": 2.9919137466307277,
"grad_norm": 0.5707534551620483,
"learning_rate": 0.00042101672962763085,
"loss": 3.5315,
"step": 27750
},
{
"epoch": 2.9973045822102424,
"grad_norm": 0.6652541160583496,
"learning_rate": 0.0004206929303831624,
"loss": 3.5268,
"step": 27800
},
{
"epoch": 3.0026954177897576,
"grad_norm": 0.5767907500267029,
"learning_rate": 0.000420369131138694,
"loss": 3.4721,
"step": 27850
},
{
"epoch": 3.0080862533692723,
"grad_norm": 0.6381232142448425,
"learning_rate": 0.00042004533189422556,
"loss": 3.4245,
"step": 27900
},
{
"epoch": 3.013477088948787,
"grad_norm": 0.6559780240058899,
"learning_rate": 0.00041972153264975716,
"loss": 3.4217,
"step": 27950
},
{
"epoch": 3.018867924528302,
"grad_norm": 0.5813524127006531,
"learning_rate": 0.00041939773340528866,
"loss": 3.4417,
"step": 28000
},
{
"epoch": 3.018867924528302,
"eval_accuracy": 0.37112805927145415,
"eval_loss": 3.496328115463257,
"eval_runtime": 184.6521,
"eval_samples_per_second": 97.54,
"eval_steps_per_second": 6.098,
"step": 28000
},
{
"epoch": 3.024258760107817,
"grad_norm": 0.6119047999382019,
"learning_rate": 0.0004190739341608202,
"loss": 3.4372,
"step": 28050
},
{
"epoch": 3.0296495956873315,
"grad_norm": 0.5905088186264038,
"learning_rate": 0.0004187501349163518,
"loss": 3.4426,
"step": 28100
},
{
"epoch": 3.035040431266846,
"grad_norm": 0.598530650138855,
"learning_rate": 0.00041842633567188336,
"loss": 3.4333,
"step": 28150
},
{
"epoch": 3.0404312668463613,
"grad_norm": 0.6369975209236145,
"learning_rate": 0.00041810253642741497,
"loss": 3.4381,
"step": 28200
},
{
"epoch": 3.045822102425876,
"grad_norm": 0.6053482890129089,
"learning_rate": 0.0004177787371829465,
"loss": 3.4204,
"step": 28250
},
{
"epoch": 3.0512129380053907,
"grad_norm": 0.5688837766647339,
"learning_rate": 0.0004174549379384781,
"loss": 3.4392,
"step": 28300
},
{
"epoch": 3.056603773584906,
"grad_norm": 0.5971906781196594,
"learning_rate": 0.0004171311386940097,
"loss": 3.4149,
"step": 28350
},
{
"epoch": 3.0619946091644206,
"grad_norm": 0.5894745588302612,
"learning_rate": 0.0004168073394495413,
"loss": 3.4448,
"step": 28400
},
{
"epoch": 3.0673854447439353,
"grad_norm": 0.6146343350410461,
"learning_rate": 0.00041648354020507283,
"loss": 3.4145,
"step": 28450
},
{
"epoch": 3.07277628032345,
"grad_norm": 0.6362399458885193,
"learning_rate": 0.0004161597409606044,
"loss": 3.4404,
"step": 28500
},
{
"epoch": 3.078167115902965,
"grad_norm": 0.6379002332687378,
"learning_rate": 0.000415835941716136,
"loss": 3.4499,
"step": 28550
},
{
"epoch": 3.08355795148248,
"grad_norm": 0.6157952547073364,
"learning_rate": 0.0004155121424716675,
"loss": 3.4352,
"step": 28600
},
{
"epoch": 3.0889487870619945,
"grad_norm": 0.6435121893882751,
"learning_rate": 0.00041518834322719914,
"loss": 3.4469,
"step": 28650
},
{
"epoch": 3.0943396226415096,
"grad_norm": 0.5627097487449646,
"learning_rate": 0.00041486454398273064,
"loss": 3.4435,
"step": 28700
},
{
"epoch": 3.0997304582210243,
"grad_norm": 0.6406444907188416,
"learning_rate": 0.00041454074473826224,
"loss": 3.4681,
"step": 28750
},
{
"epoch": 3.105121293800539,
"grad_norm": 0.6160566806793213,
"learning_rate": 0.0004142169454937938,
"loss": 3.4298,
"step": 28800
},
{
"epoch": 3.1105121293800537,
"grad_norm": 0.6142480373382568,
"learning_rate": 0.0004138931462493254,
"loss": 3.4402,
"step": 28850
},
{
"epoch": 3.115902964959569,
"grad_norm": 0.6434828639030457,
"learning_rate": 0.00041356934700485695,
"loss": 3.4347,
"step": 28900
},
{
"epoch": 3.1212938005390836,
"grad_norm": 0.6118015646934509,
"learning_rate": 0.0004132455477603885,
"loss": 3.4639,
"step": 28950
},
{
"epoch": 3.1266846361185983,
"grad_norm": 0.6289743185043335,
"learning_rate": 0.0004129217485159201,
"loss": 3.4352,
"step": 29000
},
{
"epoch": 3.1266846361185983,
"eval_accuracy": 0.3710968758926514,
"eval_loss": 3.491466760635376,
"eval_runtime": 184.1085,
"eval_samples_per_second": 97.828,
"eval_steps_per_second": 6.116,
"step": 29000
},
{
"epoch": 3.1320754716981134,
"grad_norm": 0.6491495370864868,
"learning_rate": 0.00041259794927145165,
"loss": 3.452,
"step": 29050
},
{
"epoch": 3.137466307277628,
"grad_norm": 0.5878984332084656,
"learning_rate": 0.00041227415002698326,
"loss": 3.4549,
"step": 29100
},
{
"epoch": 3.142857142857143,
"grad_norm": 0.6104332208633423,
"learning_rate": 0.0004119503507825148,
"loss": 3.4469,
"step": 29150
},
{
"epoch": 3.1482479784366575,
"grad_norm": 0.624819815158844,
"learning_rate": 0.0004116265515380464,
"loss": 3.4533,
"step": 29200
},
{
"epoch": 3.1536388140161726,
"grad_norm": 0.5960791707038879,
"learning_rate": 0.00041130275229357796,
"loss": 3.4585,
"step": 29250
},
{
"epoch": 3.1590296495956873,
"grad_norm": 0.605692982673645,
"learning_rate": 0.00041097895304910946,
"loss": 3.4602,
"step": 29300
},
{
"epoch": 3.164420485175202,
"grad_norm": 0.6395569443702698,
"learning_rate": 0.00041065515380464107,
"loss": 3.452,
"step": 29350
},
{
"epoch": 3.169811320754717,
"grad_norm": 0.6180844306945801,
"learning_rate": 0.0004103313545601726,
"loss": 3.4449,
"step": 29400
},
{
"epoch": 3.175202156334232,
"grad_norm": 0.5698346495628357,
"learning_rate": 0.0004100075553157042,
"loss": 3.4287,
"step": 29450
},
{
"epoch": 3.1805929919137466,
"grad_norm": 0.5949390530586243,
"learning_rate": 0.00040968375607123577,
"loss": 3.455,
"step": 29500
},
{
"epoch": 3.1859838274932613,
"grad_norm": 0.6196912527084351,
"learning_rate": 0.0004093599568267674,
"loss": 3.4395,
"step": 29550
},
{
"epoch": 3.1913746630727764,
"grad_norm": 0.5947751402854919,
"learning_rate": 0.0004090361575822989,
"loss": 3.4487,
"step": 29600
},
{
"epoch": 3.196765498652291,
"grad_norm": 0.5829333066940308,
"learning_rate": 0.00040871235833783053,
"loss": 3.4642,
"step": 29650
},
{
"epoch": 3.202156334231806,
"grad_norm": 0.6101020574569702,
"learning_rate": 0.0004083885590933621,
"loss": 3.4481,
"step": 29700
},
{
"epoch": 3.207547169811321,
"grad_norm": 0.5830291509628296,
"learning_rate": 0.00040806475984889363,
"loss": 3.456,
"step": 29750
},
{
"epoch": 3.2129380053908356,
"grad_norm": 0.5854420065879822,
"learning_rate": 0.00040774096060442524,
"loss": 3.4647,
"step": 29800
},
{
"epoch": 3.2183288409703503,
"grad_norm": 0.5989001393318176,
"learning_rate": 0.0004074171613599568,
"loss": 3.4706,
"step": 29850
},
{
"epoch": 3.223719676549865,
"grad_norm": 0.5977945923805237,
"learning_rate": 0.0004070933621154884,
"loss": 3.4374,
"step": 29900
},
{
"epoch": 3.22911051212938,
"grad_norm": 0.5930871367454529,
"learning_rate": 0.00040676956287101994,
"loss": 3.4416,
"step": 29950
},
{
"epoch": 3.234501347708895,
"grad_norm": 0.5958422422409058,
"learning_rate": 0.00040644576362655155,
"loss": 3.4632,
"step": 30000
},
{
"epoch": 3.234501347708895,
"eval_accuracy": 0.3722870596257712,
"eval_loss": 3.486264944076538,
"eval_runtime": 184.5735,
"eval_samples_per_second": 97.582,
"eval_steps_per_second": 6.101,
"step": 30000
},
{
"epoch": 3.2398921832884096,
"grad_norm": 0.5571640133857727,
"learning_rate": 0.00040612196438208304,
"loss": 3.4518,
"step": 30050
},
{
"epoch": 3.2452830188679247,
"grad_norm": 0.5873405337333679,
"learning_rate": 0.00040579816513761465,
"loss": 3.4559,
"step": 30100
},
{
"epoch": 3.2506738544474394,
"grad_norm": 0.6401166319847107,
"learning_rate": 0.0004054743658931462,
"loss": 3.4644,
"step": 30150
},
{
"epoch": 3.256064690026954,
"grad_norm": 0.6204870939254761,
"learning_rate": 0.00040515056664867775,
"loss": 3.4477,
"step": 30200
},
{
"epoch": 3.2614555256064692,
"grad_norm": 0.6777505278587341,
"learning_rate": 0.00040482676740420935,
"loss": 3.4517,
"step": 30250
},
{
"epoch": 3.266846361185984,
"grad_norm": 0.6005361080169678,
"learning_rate": 0.0004045029681597409,
"loss": 3.4378,
"step": 30300
},
{
"epoch": 3.2722371967654986,
"grad_norm": 0.6161072254180908,
"learning_rate": 0.0004041791689152725,
"loss": 3.4526,
"step": 30350
},
{
"epoch": 3.2776280323450133,
"grad_norm": 0.5868663787841797,
"learning_rate": 0.00040385536967080406,
"loss": 3.4641,
"step": 30400
},
{
"epoch": 3.2830188679245285,
"grad_norm": 0.593936562538147,
"learning_rate": 0.00040353157042633567,
"loss": 3.4553,
"step": 30450
},
{
"epoch": 3.288409703504043,
"grad_norm": 0.6273549795150757,
"learning_rate": 0.0004032077711818672,
"loss": 3.474,
"step": 30500
},
{
"epoch": 3.293800539083558,
"grad_norm": 0.5605999827384949,
"learning_rate": 0.0004028839719373988,
"loss": 3.4631,
"step": 30550
},
{
"epoch": 3.2991913746630726,
"grad_norm": 0.5904968976974487,
"learning_rate": 0.00040256017269293037,
"loss": 3.4577,
"step": 30600
},
{
"epoch": 3.3045822102425877,
"grad_norm": 0.5936373472213745,
"learning_rate": 0.00040223637344846187,
"loss": 3.4569,
"step": 30650
},
{
"epoch": 3.3099730458221024,
"grad_norm": 0.6303464770317078,
"learning_rate": 0.00040191257420399347,
"loss": 3.4474,
"step": 30700
},
{
"epoch": 3.315363881401617,
"grad_norm": 0.5822476148605347,
"learning_rate": 0.000401588774959525,
"loss": 3.4532,
"step": 30750
},
{
"epoch": 3.3207547169811322,
"grad_norm": 0.6246159076690674,
"learning_rate": 0.00040126497571505663,
"loss": 3.4576,
"step": 30800
},
{
"epoch": 3.326145552560647,
"grad_norm": 0.5834100842475891,
"learning_rate": 0.0004009411764705882,
"loss": 3.4496,
"step": 30850
},
{
"epoch": 3.3315363881401616,
"grad_norm": 0.6518537998199463,
"learning_rate": 0.0004006173772261198,
"loss": 3.4684,
"step": 30900
},
{
"epoch": 3.3369272237196768,
"grad_norm": 0.602120041847229,
"learning_rate": 0.00040029357798165133,
"loss": 3.4538,
"step": 30950
},
{
"epoch": 3.3423180592991915,
"grad_norm": 0.5848702788352966,
"learning_rate": 0.00039996977873718294,
"loss": 3.4721,
"step": 31000
},
{
"epoch": 3.3423180592991915,
"eval_accuracy": 0.3730831593347879,
"eval_loss": 3.479715347290039,
"eval_runtime": 184.2353,
"eval_samples_per_second": 97.761,
"eval_steps_per_second": 6.112,
"step": 31000
},
{
"epoch": 3.347708894878706,
"grad_norm": 0.6232830882072449,
"learning_rate": 0.0003996459794927145,
"loss": 3.4453,
"step": 31050
},
{
"epoch": 3.353099730458221,
"grad_norm": 0.6237083673477173,
"learning_rate": 0.00039932218024824604,
"loss": 3.4668,
"step": 31100
},
{
"epoch": 3.358490566037736,
"grad_norm": 0.5827064514160156,
"learning_rate": 0.00039899838100377764,
"loss": 3.4463,
"step": 31150
},
{
"epoch": 3.3638814016172507,
"grad_norm": 0.5585026144981384,
"learning_rate": 0.0003986745817593092,
"loss": 3.447,
"step": 31200
},
{
"epoch": 3.3692722371967654,
"grad_norm": 0.6309581398963928,
"learning_rate": 0.0003983507825148408,
"loss": 3.4637,
"step": 31250
},
{
"epoch": 3.37466307277628,
"grad_norm": 0.586483895778656,
"learning_rate": 0.00039802698327037235,
"loss": 3.4493,
"step": 31300
},
{
"epoch": 3.3800539083557952,
"grad_norm": 0.5912317037582397,
"learning_rate": 0.00039770318402590396,
"loss": 3.4386,
"step": 31350
},
{
"epoch": 3.38544474393531,
"grad_norm": 0.5977102518081665,
"learning_rate": 0.00039737938478143545,
"loss": 3.4446,
"step": 31400
},
{
"epoch": 3.3908355795148246,
"grad_norm": 0.6105404496192932,
"learning_rate": 0.000397055585536967,
"loss": 3.4632,
"step": 31450
},
{
"epoch": 3.3962264150943398,
"grad_norm": 0.6212797164916992,
"learning_rate": 0.00039673826227738795,
"loss": 3.4659,
"step": 31500
},
{
"epoch": 3.4016172506738545,
"grad_norm": 0.6436367034912109,
"learning_rate": 0.00039641446303291956,
"loss": 3.4529,
"step": 31550
},
{
"epoch": 3.407008086253369,
"grad_norm": 0.6197162866592407,
"learning_rate": 0.0003960906637884511,
"loss": 3.4372,
"step": 31600
},
{
"epoch": 3.4123989218328843,
"grad_norm": 0.6225416660308838,
"learning_rate": 0.0003957668645439827,
"loss": 3.4592,
"step": 31650
},
{
"epoch": 3.417789757412399,
"grad_norm": 0.6467655301094055,
"learning_rate": 0.00039544306529951426,
"loss": 3.4633,
"step": 31700
},
{
"epoch": 3.4231805929919137,
"grad_norm": 0.5825949907302856,
"learning_rate": 0.00039511926605504587,
"loss": 3.4601,
"step": 31750
},
{
"epoch": 3.4285714285714284,
"grad_norm": 0.6365548372268677,
"learning_rate": 0.0003947954668105774,
"loss": 3.4478,
"step": 31800
},
{
"epoch": 3.4339622641509435,
"grad_norm": 0.6298771500587463,
"learning_rate": 0.00039447166756610897,
"loss": 3.4503,
"step": 31850
},
{
"epoch": 3.439353099730458,
"grad_norm": 0.6066822409629822,
"learning_rate": 0.0003941478683216406,
"loss": 3.4623,
"step": 31900
},
{
"epoch": 3.444743935309973,
"grad_norm": 0.5848886966705322,
"learning_rate": 0.0003938240690771721,
"loss": 3.4459,
"step": 31950
},
{
"epoch": 3.450134770889488,
"grad_norm": 0.6190735101699829,
"learning_rate": 0.00039350026983270373,
"loss": 3.4584,
"step": 32000
},
{
"epoch": 3.450134770889488,
"eval_accuracy": 0.3734823500446183,
"eval_loss": 3.472633123397827,
"eval_runtime": 184.612,
"eval_samples_per_second": 97.561,
"eval_steps_per_second": 6.099,
"step": 32000
},
{
"epoch": 3.4555256064690028,
"grad_norm": 0.589854896068573,
"learning_rate": 0.00039317647058823523,
"loss": 3.4821,
"step": 32050
},
{
"epoch": 3.4609164420485174,
"grad_norm": 0.5828700065612793,
"learning_rate": 0.00039285267134376683,
"loss": 3.4698,
"step": 32100
},
{
"epoch": 3.466307277628032,
"grad_norm": 0.5920112133026123,
"learning_rate": 0.0003925288720992984,
"loss": 3.4439,
"step": 32150
},
{
"epoch": 3.4716981132075473,
"grad_norm": 0.6191278696060181,
"learning_rate": 0.00039220507285482993,
"loss": 3.4587,
"step": 32200
},
{
"epoch": 3.477088948787062,
"grad_norm": 0.6206852793693542,
"learning_rate": 0.00039188127361036154,
"loss": 3.4488,
"step": 32250
},
{
"epoch": 3.4824797843665767,
"grad_norm": 0.6430778503417969,
"learning_rate": 0.0003915574743658931,
"loss": 3.4502,
"step": 32300
},
{
"epoch": 3.487870619946092,
"grad_norm": 0.5901011824607849,
"learning_rate": 0.0003912336751214247,
"loss": 3.4429,
"step": 32350
},
{
"epoch": 3.4932614555256065,
"grad_norm": 0.6198083758354187,
"learning_rate": 0.00039090987587695624,
"loss": 3.4474,
"step": 32400
},
{
"epoch": 3.498652291105121,
"grad_norm": 0.5977095365524292,
"learning_rate": 0.00039058607663248785,
"loss": 3.4573,
"step": 32450
},
{
"epoch": 3.5040431266846364,
"grad_norm": 0.6178451180458069,
"learning_rate": 0.0003902622773880194,
"loss": 3.4647,
"step": 32500
},
{
"epoch": 3.509433962264151,
"grad_norm": 0.5720937252044678,
"learning_rate": 0.000389938478143551,
"loss": 3.464,
"step": 32550
},
{
"epoch": 3.5148247978436657,
"grad_norm": 0.6495591998100281,
"learning_rate": 0.00038961467889908255,
"loss": 3.4516,
"step": 32600
},
{
"epoch": 3.5202156334231804,
"grad_norm": 0.5951046347618103,
"learning_rate": 0.00038929087965461405,
"loss": 3.4575,
"step": 32650
},
{
"epoch": 3.525606469002695,
"grad_norm": 0.6124663949012756,
"learning_rate": 0.00038896708041014566,
"loss": 3.471,
"step": 32700
},
{
"epoch": 3.5309973045822103,
"grad_norm": 0.6389647126197815,
"learning_rate": 0.0003886432811656772,
"loss": 3.4482,
"step": 32750
},
{
"epoch": 3.536388140161725,
"grad_norm": 0.6239250302314758,
"learning_rate": 0.0003883194819212088,
"loss": 3.4738,
"step": 32800
},
{
"epoch": 3.5417789757412397,
"grad_norm": 0.5783104300498962,
"learning_rate": 0.00038799568267674036,
"loss": 3.4738,
"step": 32850
},
{
"epoch": 3.547169811320755,
"grad_norm": 0.6263895034790039,
"learning_rate": 0.00038767188343227197,
"loss": 3.4671,
"step": 32900
},
{
"epoch": 3.5525606469002695,
"grad_norm": 0.6639639735221863,
"learning_rate": 0.0003873480841878035,
"loss": 3.4549,
"step": 32950
},
{
"epoch": 3.557951482479784,
"grad_norm": 0.6445685029029846,
"learning_rate": 0.0003870242849433351,
"loss": 3.4511,
"step": 33000
},
{
"epoch": 3.557951482479784,
"eval_accuracy": 0.3739473844044643,
"eval_loss": 3.4677462577819824,
"eval_runtime": 184.1749,
"eval_samples_per_second": 97.793,
"eval_steps_per_second": 6.114,
"step": 33000
},
{
"epoch": 3.5633423180592994,
"grad_norm": 0.6111474633216858,
"learning_rate": 0.00038670048569886667,
"loss": 3.4599,
"step": 33050
},
{
"epoch": 3.568733153638814,
"grad_norm": 0.615082323551178,
"learning_rate": 0.0003863766864543982,
"loss": 3.4637,
"step": 33100
},
{
"epoch": 3.5741239892183287,
"grad_norm": 0.6300560832023621,
"learning_rate": 0.00038605288720992983,
"loss": 3.4675,
"step": 33150
},
{
"epoch": 3.579514824797844,
"grad_norm": 0.6000446677207947,
"learning_rate": 0.0003857290879654614,
"loss": 3.4622,
"step": 33200
},
{
"epoch": 3.5849056603773586,
"grad_norm": 0.6588138341903687,
"learning_rate": 0.000385405288720993,
"loss": 3.4448,
"step": 33250
},
{
"epoch": 3.5902964959568733,
"grad_norm": 0.6407126188278198,
"learning_rate": 0.00038508148947652453,
"loss": 3.47,
"step": 33300
},
{
"epoch": 3.595687331536388,
"grad_norm": 0.6264423727989197,
"learning_rate": 0.00038475769023205614,
"loss": 3.4596,
"step": 33350
},
{
"epoch": 3.601078167115903,
"grad_norm": 0.6139709949493408,
"learning_rate": 0.00038443389098758763,
"loss": 3.4571,
"step": 33400
},
{
"epoch": 3.606469002695418,
"grad_norm": 0.6293317675590515,
"learning_rate": 0.00038411009174311924,
"loss": 3.4722,
"step": 33450
},
{
"epoch": 3.6118598382749325,
"grad_norm": 0.5989006161689758,
"learning_rate": 0.0003837862924986508,
"loss": 3.4482,
"step": 33500
},
{
"epoch": 3.617250673854447,
"grad_norm": 0.6376069188117981,
"learning_rate": 0.00038346896923907174,
"loss": 3.4487,
"step": 33550
},
{
"epoch": 3.6226415094339623,
"grad_norm": 0.6293345093727112,
"learning_rate": 0.0003831451699946033,
"loss": 3.4431,
"step": 33600
},
{
"epoch": 3.628032345013477,
"grad_norm": 0.6624619364738464,
"learning_rate": 0.0003828213707501349,
"loss": 3.46,
"step": 33650
},
{
"epoch": 3.6334231805929917,
"grad_norm": 0.6465222835540771,
"learning_rate": 0.00038249757150566645,
"loss": 3.4647,
"step": 33700
},
{
"epoch": 3.638814016172507,
"grad_norm": 0.6340253353118896,
"learning_rate": 0.00038217377226119805,
"loss": 3.4437,
"step": 33750
},
{
"epoch": 3.6442048517520216,
"grad_norm": 0.6425220370292664,
"learning_rate": 0.0003818499730167296,
"loss": 3.4491,
"step": 33800
},
{
"epoch": 3.6495956873315363,
"grad_norm": 0.6248994469642639,
"learning_rate": 0.00038152617377226115,
"loss": 3.432,
"step": 33850
},
{
"epoch": 3.6549865229110514,
"grad_norm": 0.6413818597793579,
"learning_rate": 0.00038120237452779276,
"loss": 3.426,
"step": 33900
},
{
"epoch": 3.660377358490566,
"grad_norm": 0.6339235305786133,
"learning_rate": 0.0003808785752833243,
"loss": 3.4527,
"step": 33950
},
{
"epoch": 3.665768194070081,
"grad_norm": 0.6301815509796143,
"learning_rate": 0.0003805547760388559,
"loss": 3.4489,
"step": 34000
},
{
"epoch": 3.665768194070081,
"eval_accuracy": 0.37472131892436683,
"eval_loss": 3.4617087841033936,
"eval_runtime": 184.7088,
"eval_samples_per_second": 97.51,
"eval_steps_per_second": 6.096,
"step": 34000
},
{
"epoch": 3.671159029649596,
"grad_norm": 0.6945334076881409,
"learning_rate": 0.0003802309767943874,
"loss": 3.4529,
"step": 34050
},
{
"epoch": 3.6765498652291106,
"grad_norm": 0.6272717118263245,
"learning_rate": 0.000379907177549919,
"loss": 3.4506,
"step": 34100
},
{
"epoch": 3.6819407008086253,
"grad_norm": 0.5997910499572754,
"learning_rate": 0.0003795898542903399,
"loss": 3.4464,
"step": 34150
},
{
"epoch": 3.68733153638814,
"grad_norm": 0.7196186184883118,
"learning_rate": 0.0003792660550458715,
"loss": 3.4565,
"step": 34200
},
{
"epoch": 3.6927223719676547,
"grad_norm": 0.6358057260513306,
"learning_rate": 0.00037894225580140307,
"loss": 3.4477,
"step": 34250
},
{
"epoch": 3.69811320754717,
"grad_norm": 0.6382709741592407,
"learning_rate": 0.00037861845655693467,
"loss": 3.4512,
"step": 34300
},
{
"epoch": 3.7035040431266846,
"grad_norm": 0.6352797150611877,
"learning_rate": 0.0003782946573124662,
"loss": 3.45,
"step": 34350
},
{
"epoch": 3.7088948787061993,
"grad_norm": 0.6407597661018372,
"learning_rate": 0.00037797085806799783,
"loss": 3.4581,
"step": 34400
},
{
"epoch": 3.7142857142857144,
"grad_norm": 0.5949084758758545,
"learning_rate": 0.0003776470588235294,
"loss": 3.4283,
"step": 34450
},
{
"epoch": 3.719676549865229,
"grad_norm": 0.6408184766769409,
"learning_rate": 0.000377323259579061,
"loss": 3.468,
"step": 34500
},
{
"epoch": 3.725067385444744,
"grad_norm": 0.6238266825675964,
"learning_rate": 0.00037699946033459253,
"loss": 3.4465,
"step": 34550
},
{
"epoch": 3.730458221024259,
"grad_norm": 0.6232044100761414,
"learning_rate": 0.0003766756610901241,
"loss": 3.4518,
"step": 34600
},
{
"epoch": 3.7358490566037736,
"grad_norm": 0.6160922646522522,
"learning_rate": 0.0003763518618456557,
"loss": 3.4559,
"step": 34650
},
{
"epoch": 3.7412398921832883,
"grad_norm": 0.6382622718811035,
"learning_rate": 0.0003760280626011872,
"loss": 3.4616,
"step": 34700
},
{
"epoch": 3.7466307277628035,
"grad_norm": 0.625793993473053,
"learning_rate": 0.0003757042633567188,
"loss": 3.4667,
"step": 34750
},
{
"epoch": 3.752021563342318,
"grad_norm": 0.6298996210098267,
"learning_rate": 0.00037538046411225034,
"loss": 3.4493,
"step": 34800
},
{
"epoch": 3.757412398921833,
"grad_norm": 0.6189321875572205,
"learning_rate": 0.00037505666486778195,
"loss": 3.4583,
"step": 34850
},
{
"epoch": 3.7628032345013476,
"grad_norm": 0.6454741954803467,
"learning_rate": 0.0003747328656233135,
"loss": 3.4521,
"step": 34900
},
{
"epoch": 3.7681940700808623,
"grad_norm": 0.6494364142417908,
"learning_rate": 0.0003744090663788451,
"loss": 3.4445,
"step": 34950
},
{
"epoch": 3.7735849056603774,
"grad_norm": 0.684683084487915,
"learning_rate": 0.00037408526713437665,
"loss": 3.4669,
"step": 35000
},
{
"epoch": 3.7735849056603774,
"eval_accuracy": 0.3754683075280914,
"eval_loss": 3.454247236251831,
"eval_runtime": 184.2801,
"eval_samples_per_second": 97.737,
"eval_steps_per_second": 6.11,
"step": 35000
},
{
"epoch": 3.778975741239892,
"grad_norm": 0.6288757920265198,
"learning_rate": 0.0003737614678899082,
"loss": 3.451,
"step": 35050
},
{
"epoch": 3.784366576819407,
"grad_norm": 0.6487582325935364,
"learning_rate": 0.0003734376686454398,
"loss": 3.453,
"step": 35100
},
{
"epoch": 3.789757412398922,
"grad_norm": 0.6697840094566345,
"learning_rate": 0.00037311386940097136,
"loss": 3.4407,
"step": 35150
},
{
"epoch": 3.7951482479784366,
"grad_norm": 0.5957277417182922,
"learning_rate": 0.00037279007015650296,
"loss": 3.4429,
"step": 35200
},
{
"epoch": 3.8005390835579513,
"grad_norm": 0.6827387809753418,
"learning_rate": 0.0003724662709120345,
"loss": 3.4519,
"step": 35250
},
{
"epoch": 3.8059299191374665,
"grad_norm": 0.6332709193229675,
"learning_rate": 0.0003721424716675661,
"loss": 3.4462,
"step": 35300
},
{
"epoch": 3.811320754716981,
"grad_norm": 0.6271256804466248,
"learning_rate": 0.0003718186724230976,
"loss": 3.4659,
"step": 35350
},
{
"epoch": 3.816711590296496,
"grad_norm": 0.5937188267707825,
"learning_rate": 0.0003714948731786293,
"loss": 3.4466,
"step": 35400
},
{
"epoch": 3.822102425876011,
"grad_norm": 0.6044777035713196,
"learning_rate": 0.00037117107393416077,
"loss": 3.4523,
"step": 35450
},
{
"epoch": 3.8274932614555257,
"grad_norm": 0.6193053126335144,
"learning_rate": 0.0003708472746896923,
"loss": 3.4435,
"step": 35500
},
{
"epoch": 3.8328840970350404,
"grad_norm": 0.6096031665802002,
"learning_rate": 0.0003705234754452239,
"loss": 3.447,
"step": 35550
},
{
"epoch": 3.838274932614555,
"grad_norm": 0.6659539937973022,
"learning_rate": 0.0003701996762007555,
"loss": 3.458,
"step": 35600
},
{
"epoch": 3.8436657681940702,
"grad_norm": 0.6063067317008972,
"learning_rate": 0.0003698758769562871,
"loss": 3.4427,
"step": 35650
},
{
"epoch": 3.849056603773585,
"grad_norm": 0.610436737537384,
"learning_rate": 0.00036955207771181863,
"loss": 3.4425,
"step": 35700
},
{
"epoch": 3.8544474393530996,
"grad_norm": 0.6312926411628723,
"learning_rate": 0.00036922827846735024,
"loss": 3.4527,
"step": 35750
},
{
"epoch": 3.8598382749326143,
"grad_norm": 0.600482165813446,
"learning_rate": 0.0003689044792228818,
"loss": 3.4406,
"step": 35800
},
{
"epoch": 3.8652291105121295,
"grad_norm": 0.6272764205932617,
"learning_rate": 0.0003685806799784134,
"loss": 3.4493,
"step": 35850
},
{
"epoch": 3.870619946091644,
"grad_norm": 0.6267163753509521,
"learning_rate": 0.00036825688073394494,
"loss": 3.4628,
"step": 35900
},
{
"epoch": 3.876010781671159,
"grad_norm": 0.6345289945602417,
"learning_rate": 0.0003679330814894765,
"loss": 3.462,
"step": 35950
},
{
"epoch": 3.881401617250674,
"grad_norm": 0.6223365664482117,
"learning_rate": 0.0003676092822450081,
"loss": 3.434,
"step": 36000
},
{
"epoch": 3.881401617250674,
"eval_accuracy": 0.376085238625102,
"eval_loss": 3.449181079864502,
"eval_runtime": 184.6159,
"eval_samples_per_second": 97.559,
"eval_steps_per_second": 6.099,
"step": 36000
},
{
"epoch": 3.8867924528301887,
"grad_norm": 0.5973546504974365,
"learning_rate": 0.0003672854830005396,
"loss": 3.4681,
"step": 36050
},
{
"epoch": 3.8921832884097034,
"grad_norm": 0.649315595626831,
"learning_rate": 0.0003669616837560712,
"loss": 3.434,
"step": 36100
},
{
"epoch": 3.8975741239892185,
"grad_norm": 0.6562724113464355,
"learning_rate": 0.00036663788451160275,
"loss": 3.4411,
"step": 36150
},
{
"epoch": 3.9029649595687332,
"grad_norm": 0.5861834287643433,
"learning_rate": 0.00036631408526713435,
"loss": 3.448,
"step": 36200
},
{
"epoch": 3.908355795148248,
"grad_norm": 0.5954273343086243,
"learning_rate": 0.0003659902860226659,
"loss": 3.4542,
"step": 36250
},
{
"epoch": 3.913746630727763,
"grad_norm": 0.5699360370635986,
"learning_rate": 0.00036566648677819745,
"loss": 3.4541,
"step": 36300
},
{
"epoch": 3.9191374663072778,
"grad_norm": 0.6212907433509827,
"learning_rate": 0.00036534268753372906,
"loss": 3.4676,
"step": 36350
},
{
"epoch": 3.9245283018867925,
"grad_norm": 0.5900413393974304,
"learning_rate": 0.0003650188882892606,
"loss": 3.4365,
"step": 36400
},
{
"epoch": 3.929919137466307,
"grad_norm": 0.6464952826499939,
"learning_rate": 0.0003646950890447922,
"loss": 3.4413,
"step": 36450
},
{
"epoch": 3.935309973045822,
"grad_norm": 0.6185500621795654,
"learning_rate": 0.00036437128980032376,
"loss": 3.4545,
"step": 36500
},
{
"epoch": 3.940700808625337,
"grad_norm": 0.5989026427268982,
"learning_rate": 0.00036404749055585537,
"loss": 3.4702,
"step": 36550
},
{
"epoch": 3.9460916442048517,
"grad_norm": 0.6319485902786255,
"learning_rate": 0.0003637236913113869,
"loss": 3.4217,
"step": 36600
},
{
"epoch": 3.9514824797843664,
"grad_norm": 0.6524223685264587,
"learning_rate": 0.0003633998920669185,
"loss": 3.4399,
"step": 36650
},
{
"epoch": 3.9568733153638815,
"grad_norm": 0.670523464679718,
"learning_rate": 0.0003630760928224501,
"loss": 3.4456,
"step": 36700
},
{
"epoch": 3.9622641509433962,
"grad_norm": 0.6367616057395935,
"learning_rate": 0.00036275229357798157,
"loss": 3.4265,
"step": 36750
},
{
"epoch": 3.967654986522911,
"grad_norm": 0.5835162401199341,
"learning_rate": 0.0003624284943335132,
"loss": 3.4409,
"step": 36800
},
{
"epoch": 3.973045822102426,
"grad_norm": 0.6265690922737122,
"learning_rate": 0.00036210469508904473,
"loss": 3.4481,
"step": 36850
},
{
"epoch": 3.9784366576819408,
"grad_norm": 0.6020421385765076,
"learning_rate": 0.00036178089584457633,
"loss": 3.4468,
"step": 36900
},
{
"epoch": 3.9838274932614555,
"grad_norm": 0.6064275503158569,
"learning_rate": 0.0003614570966001079,
"loss": 3.4269,
"step": 36950
},
{
"epoch": 3.9892183288409706,
"grad_norm": 0.6680476069450378,
"learning_rate": 0.0003611332973556395,
"loss": 3.4136,
"step": 37000
},
{
"epoch": 3.9892183288409706,
"eval_accuracy": 0.37626679760064,
"eval_loss": 3.442439317703247,
"eval_runtime": 184.1183,
"eval_samples_per_second": 97.823,
"eval_steps_per_second": 6.116,
"step": 37000
},
{
"epoch": 3.9946091644204853,
"grad_norm": 0.634256899356842,
"learning_rate": 0.00036080949811117104,
"loss": 3.4652,
"step": 37050
},
{
"epoch": 4.0,
"grad_norm": 1.317596197128296,
"learning_rate": 0.00036048569886670264,
"loss": 3.4305,
"step": 37100
},
{
"epoch": 4.005390835579515,
"grad_norm": 0.6005503535270691,
"learning_rate": 0.0003601618996222342,
"loss": 3.3608,
"step": 37150
},
{
"epoch": 4.010781671159029,
"grad_norm": 0.6252793669700623,
"learning_rate": 0.00035983810037776574,
"loss": 3.3552,
"step": 37200
},
{
"epoch": 4.0161725067385445,
"grad_norm": 0.6997048854827881,
"learning_rate": 0.00035951430113329735,
"loss": 3.3524,
"step": 37250
},
{
"epoch": 4.02156334231806,
"grad_norm": 0.6069782972335815,
"learning_rate": 0.0003591905018888289,
"loss": 3.3557,
"step": 37300
},
{
"epoch": 4.026954177897574,
"grad_norm": 0.6234855651855469,
"learning_rate": 0.0003588667026443605,
"loss": 3.3375,
"step": 37350
},
{
"epoch": 4.032345013477089,
"grad_norm": 0.603093147277832,
"learning_rate": 0.000358542903399892,
"loss": 3.34,
"step": 37400
},
{
"epoch": 4.037735849056604,
"grad_norm": 0.6459537148475647,
"learning_rate": 0.0003582191041554236,
"loss": 3.3521,
"step": 37450
},
{
"epoch": 4.0431266846361185,
"grad_norm": 0.6772938370704651,
"learning_rate": 0.00035789530491095516,
"loss": 3.3509,
"step": 37500
},
{
"epoch": 4.048517520215634,
"grad_norm": 0.646106481552124,
"learning_rate": 0.00035757150566648676,
"loss": 3.3608,
"step": 37550
},
{
"epoch": 4.053908355795148,
"grad_norm": 0.6739107966423035,
"learning_rate": 0.0003572477064220183,
"loss": 3.3547,
"step": 37600
},
{
"epoch": 4.059299191374663,
"grad_norm": 0.6679389476776123,
"learning_rate": 0.00035692390717754986,
"loss": 3.3628,
"step": 37650
},
{
"epoch": 4.064690026954178,
"grad_norm": 0.595439612865448,
"learning_rate": 0.00035660010793308147,
"loss": 3.3641,
"step": 37700
},
{
"epoch": 4.070080862533692,
"grad_norm": 0.5981416702270508,
"learning_rate": 0.000356276308688613,
"loss": 3.3694,
"step": 37750
},
{
"epoch": 4.0754716981132075,
"grad_norm": 0.6261246800422668,
"learning_rate": 0.0003559525094441446,
"loss": 3.361,
"step": 37800
},
{
"epoch": 4.080862533692723,
"grad_norm": 0.6274250745773315,
"learning_rate": 0.00035562871019967617,
"loss": 3.3669,
"step": 37850
},
{
"epoch": 4.086253369272237,
"grad_norm": 0.6313136219978333,
"learning_rate": 0.0003553049109552078,
"loss": 3.3774,
"step": 37900
},
{
"epoch": 4.091644204851752,
"grad_norm": 0.6337655186653137,
"learning_rate": 0.00035498111171073933,
"loss": 3.3757,
"step": 37950
},
{
"epoch": 4.097035040431267,
"grad_norm": 0.636246383190155,
"learning_rate": 0.0003546573124662708,
"loss": 3.3583,
"step": 38000
},
{
"epoch": 4.097035040431267,
"eval_accuracy": 0.3768646057893953,
"eval_loss": 3.444782257080078,
"eval_runtime": 184.7601,
"eval_samples_per_second": 97.483,
"eval_steps_per_second": 6.094,
"step": 38000
},
{
"epoch": 4.1024258760107815,
"grad_norm": 0.6387351751327515,
"learning_rate": 0.0003543335132218025,
"loss": 3.3744,
"step": 38050
},
{
"epoch": 4.107816711590297,
"grad_norm": 0.6440540552139282,
"learning_rate": 0.000354009713977334,
"loss": 3.3573,
"step": 38100
},
{
"epoch": 4.113207547169812,
"grad_norm": 0.6814355850219727,
"learning_rate": 0.00035369239071775493,
"loss": 3.3749,
"step": 38150
},
{
"epoch": 4.118598382749326,
"grad_norm": 0.6349874138832092,
"learning_rate": 0.00035336859147328654,
"loss": 3.3776,
"step": 38200
},
{
"epoch": 4.123989218328841,
"grad_norm": 0.6483203172683716,
"learning_rate": 0.0003530447922288181,
"loss": 3.3641,
"step": 38250
},
{
"epoch": 4.129380053908355,
"grad_norm": 0.6210718750953674,
"learning_rate": 0.0003527209929843497,
"loss": 3.3685,
"step": 38300
},
{
"epoch": 4.1347708894878705,
"grad_norm": 0.6137518882751465,
"learning_rate": 0.00035239719373988124,
"loss": 3.3521,
"step": 38350
},
{
"epoch": 4.140161725067386,
"grad_norm": 0.6737520694732666,
"learning_rate": 0.0003520733944954128,
"loss": 3.374,
"step": 38400
},
{
"epoch": 4.1455525606469,
"grad_norm": 0.6203984022140503,
"learning_rate": 0.0003517495952509444,
"loss": 3.371,
"step": 38450
},
{
"epoch": 4.150943396226415,
"grad_norm": 0.6363698840141296,
"learning_rate": 0.00035142579600647595,
"loss": 3.3665,
"step": 38500
},
{
"epoch": 4.15633423180593,
"grad_norm": 0.5890799164772034,
"learning_rate": 0.00035110199676200755,
"loss": 3.3631,
"step": 38550
},
{
"epoch": 4.1617250673854445,
"grad_norm": 0.6854996085166931,
"learning_rate": 0.0003507781975175391,
"loss": 3.3619,
"step": 38600
},
{
"epoch": 4.16711590296496,
"grad_norm": 0.6439552307128906,
"learning_rate": 0.0003504543982730707,
"loss": 3.3674,
"step": 38650
},
{
"epoch": 4.172506738544475,
"grad_norm": 0.648841381072998,
"learning_rate": 0.00035013059902860226,
"loss": 3.3843,
"step": 38700
},
{
"epoch": 4.177897574123989,
"grad_norm": 0.6229033470153809,
"learning_rate": 0.00034980679978413375,
"loss": 3.3586,
"step": 38750
},
{
"epoch": 4.183288409703504,
"grad_norm": 0.6032453775405884,
"learning_rate": 0.00034948300053966536,
"loss": 3.3753,
"step": 38800
},
{
"epoch": 4.188679245283019,
"grad_norm": 0.6159026026725769,
"learning_rate": 0.0003491592012951969,
"loss": 3.3848,
"step": 38850
},
{
"epoch": 4.1940700808625335,
"grad_norm": 0.618169367313385,
"learning_rate": 0.0003488354020507285,
"loss": 3.4007,
"step": 38900
},
{
"epoch": 4.199460916442049,
"grad_norm": 0.6955693364143372,
"learning_rate": 0.00034851160280626007,
"loss": 3.3653,
"step": 38950
},
{
"epoch": 4.204851752021563,
"grad_norm": 0.6858065724372864,
"learning_rate": 0.00034818780356179167,
"loss": 3.3806,
"step": 39000
},
{
"epoch": 4.204851752021563,
"eval_accuracy": 0.37729954329931664,
"eval_loss": 3.4421889781951904,
"eval_runtime": 184.2106,
"eval_samples_per_second": 97.774,
"eval_steps_per_second": 6.113,
"step": 39000
},
{
"epoch": 4.210242587601078,
"grad_norm": 0.6968677639961243,
"learning_rate": 0.0003478640043173232,
"loss": 3.386,
"step": 39050
},
{
"epoch": 4.215633423180593,
"grad_norm": 0.6504750847816467,
"learning_rate": 0.0003475402050728548,
"loss": 3.3888,
"step": 39100
},
{
"epoch": 4.2210242587601075,
"grad_norm": 0.6221845746040344,
"learning_rate": 0.0003472164058283864,
"loss": 3.3777,
"step": 39150
},
{
"epoch": 4.226415094339623,
"grad_norm": 0.701174795627594,
"learning_rate": 0.0003468926065839179,
"loss": 3.3677,
"step": 39200
},
{
"epoch": 4.231805929919138,
"grad_norm": 0.6199811697006226,
"learning_rate": 0.00034656880733944953,
"loss": 3.3719,
"step": 39250
},
{
"epoch": 4.237196765498652,
"grad_norm": 0.6160785555839539,
"learning_rate": 0.0003462450080949811,
"loss": 3.3719,
"step": 39300
},
{
"epoch": 4.242587601078167,
"grad_norm": 0.5981932282447815,
"learning_rate": 0.0003459212088505127,
"loss": 3.3857,
"step": 39350
},
{
"epoch": 4.247978436657682,
"grad_norm": 0.6872910261154175,
"learning_rate": 0.0003455974096060442,
"loss": 3.3796,
"step": 39400
},
{
"epoch": 4.2533692722371965,
"grad_norm": 0.6778494119644165,
"learning_rate": 0.0003452736103615758,
"loss": 3.389,
"step": 39450
},
{
"epoch": 4.258760107816712,
"grad_norm": 0.6239806413650513,
"learning_rate": 0.00034494981111710734,
"loss": 3.3758,
"step": 39500
},
{
"epoch": 4.264150943396227,
"grad_norm": 0.6443771123886108,
"learning_rate": 0.00034462601187263894,
"loss": 3.3632,
"step": 39550
},
{
"epoch": 4.269541778975741,
"grad_norm": 0.6361122727394104,
"learning_rate": 0.0003443022126281705,
"loss": 3.3785,
"step": 39600
},
{
"epoch": 4.274932614555256,
"grad_norm": 0.6191430687904358,
"learning_rate": 0.00034397841338370204,
"loss": 3.3878,
"step": 39650
},
{
"epoch": 4.280323450134771,
"grad_norm": 0.6481466293334961,
"learning_rate": 0.00034365461413923365,
"loss": 3.37,
"step": 39700
},
{
"epoch": 4.285714285714286,
"grad_norm": 0.646091878414154,
"learning_rate": 0.0003433308148947652,
"loss": 3.3948,
"step": 39750
},
{
"epoch": 4.291105121293801,
"grad_norm": 0.6309322714805603,
"learning_rate": 0.0003430070156502968,
"loss": 3.3812,
"step": 39800
},
{
"epoch": 4.296495956873315,
"grad_norm": 0.6581062078475952,
"learning_rate": 0.00034268321640582835,
"loss": 3.3838,
"step": 39850
},
{
"epoch": 4.30188679245283,
"grad_norm": 0.6658921241760254,
"learning_rate": 0.00034235941716135996,
"loss": 3.3844,
"step": 39900
},
{
"epoch": 4.307277628032345,
"grad_norm": 0.7357547283172607,
"learning_rate": 0.0003420356179168915,
"loss": 3.3915,
"step": 39950
},
{
"epoch": 4.3126684636118595,
"grad_norm": 0.689964234828949,
"learning_rate": 0.0003417118186724231,
"loss": 3.3978,
"step": 40000
},
{
"epoch": 4.3126684636118595,
"eval_accuracy": 0.3774527438711351,
"eval_loss": 3.436279058456421,
"eval_runtime": 184.7288,
"eval_samples_per_second": 97.5,
"eval_steps_per_second": 6.095,
"step": 40000
},
{
"epoch": 4.318059299191375,
"grad_norm": 0.6188414692878723,
"learning_rate": 0.00034138801942795467,
"loss": 3.3848,
"step": 40050
},
{
"epoch": 4.32345013477089,
"grad_norm": 0.6350507736206055,
"learning_rate": 0.00034107069616837556,
"loss": 3.3903,
"step": 40100
},
{
"epoch": 4.328840970350404,
"grad_norm": 0.644652783870697,
"learning_rate": 0.0003407468969239071,
"loss": 3.3707,
"step": 40150
},
{
"epoch": 4.334231805929919,
"grad_norm": 0.6473593711853027,
"learning_rate": 0.0003404230976794387,
"loss": 3.4032,
"step": 40200
},
{
"epoch": 4.339622641509434,
"grad_norm": 0.6398594379425049,
"learning_rate": 0.00034009929843497027,
"loss": 3.4058,
"step": 40250
},
{
"epoch": 4.345013477088949,
"grad_norm": 0.666358232498169,
"learning_rate": 0.0003397754991905019,
"loss": 3.3784,
"step": 40300
},
{
"epoch": 4.350404312668464,
"grad_norm": 0.6536188125610352,
"learning_rate": 0.0003394516999460334,
"loss": 3.3882,
"step": 40350
},
{
"epoch": 4.355795148247978,
"grad_norm": 0.6254106163978577,
"learning_rate": 0.000339127900701565,
"loss": 3.3866,
"step": 40400
},
{
"epoch": 4.361185983827493,
"grad_norm": 0.6590883731842041,
"learning_rate": 0.0003388041014570966,
"loss": 3.3859,
"step": 40450
},
{
"epoch": 4.366576819407008,
"grad_norm": 0.6693887114524841,
"learning_rate": 0.00033848030221262813,
"loss": 3.3806,
"step": 40500
},
{
"epoch": 4.3719676549865225,
"grad_norm": 0.6402140855789185,
"learning_rate": 0.00033815650296815974,
"loss": 3.3874,
"step": 40550
},
{
"epoch": 4.377358490566038,
"grad_norm": 0.6448017954826355,
"learning_rate": 0.0003378327037236913,
"loss": 3.3941,
"step": 40600
},
{
"epoch": 4.382749326145553,
"grad_norm": 0.6667120456695557,
"learning_rate": 0.0003375089044792229,
"loss": 3.3842,
"step": 40650
},
{
"epoch": 4.388140161725067,
"grad_norm": 0.6299318075180054,
"learning_rate": 0.00033718510523475444,
"loss": 3.3976,
"step": 40700
},
{
"epoch": 4.393530997304582,
"grad_norm": 0.6535727381706238,
"learning_rate": 0.00033686130599028605,
"loss": 3.3995,
"step": 40750
},
{
"epoch": 4.398921832884097,
"grad_norm": 0.6459347009658813,
"learning_rate": 0.00033653750674581754,
"loss": 3.3906,
"step": 40800
},
{
"epoch": 4.404312668463612,
"grad_norm": 0.6215057969093323,
"learning_rate": 0.0003362137075013491,
"loss": 3.3795,
"step": 40850
},
{
"epoch": 4.409703504043127,
"grad_norm": 0.6472002863883972,
"learning_rate": 0.0003358899082568807,
"loss": 3.3798,
"step": 40900
},
{
"epoch": 4.415094339622642,
"grad_norm": 0.9916529655456543,
"learning_rate": 0.00033556610901241225,
"loss": 3.3768,
"step": 40950
},
{
"epoch": 4.420485175202156,
"grad_norm": 0.6788925528526306,
"learning_rate": 0.00033524230976794385,
"loss": 3.4136,
"step": 41000
},
{
"epoch": 4.420485175202156,
"eval_accuracy": 0.3785148258495216,
"eval_loss": 3.432884931564331,
"eval_runtime": 184.3413,
"eval_samples_per_second": 97.705,
"eval_steps_per_second": 6.108,
"step": 41000
},
{
"epoch": 4.425876010781671,
"grad_norm": 0.6621363162994385,
"learning_rate": 0.0003349185105234754,
"loss": 3.3869,
"step": 41050
},
{
"epoch": 4.431266846361186,
"grad_norm": 0.6567580699920654,
"learning_rate": 0.000334594711279007,
"loss": 3.396,
"step": 41100
},
{
"epoch": 4.436657681940701,
"grad_norm": 0.6388005614280701,
"learning_rate": 0.00033427091203453856,
"loss": 3.4012,
"step": 41150
},
{
"epoch": 4.442048517520216,
"grad_norm": 0.6279880404472351,
"learning_rate": 0.00033394711279007016,
"loss": 3.3871,
"step": 41200
},
{
"epoch": 4.44743935309973,
"grad_norm": 0.636476457118988,
"learning_rate": 0.0003336233135456017,
"loss": 3.4036,
"step": 41250
},
{
"epoch": 4.452830188679245,
"grad_norm": 0.6203899383544922,
"learning_rate": 0.00033329951430113326,
"loss": 3.3754,
"step": 41300
},
{
"epoch": 4.45822102425876,
"grad_norm": 0.6366904377937317,
"learning_rate": 0.00033297571505666487,
"loss": 3.3823,
"step": 41350
},
{
"epoch": 4.463611859838275,
"grad_norm": 0.6318303942680359,
"learning_rate": 0.00033265191581219637,
"loss": 3.4047,
"step": 41400
},
{
"epoch": 4.46900269541779,
"grad_norm": 0.6402361392974854,
"learning_rate": 0.00033232811656772797,
"loss": 3.3801,
"step": 41450
},
{
"epoch": 4.474393530997305,
"grad_norm": 1.2731075286865234,
"learning_rate": 0.0003320043173232595,
"loss": 3.3927,
"step": 41500
},
{
"epoch": 4.479784366576819,
"grad_norm": 0.6158382296562195,
"learning_rate": 0.0003316805180787911,
"loss": 3.385,
"step": 41550
},
{
"epoch": 4.485175202156334,
"grad_norm": 0.6707490682601929,
"learning_rate": 0.0003313567188343227,
"loss": 3.3965,
"step": 41600
},
{
"epoch": 4.490566037735849,
"grad_norm": 0.6214455962181091,
"learning_rate": 0.00033103291958985423,
"loss": 3.3795,
"step": 41650
},
{
"epoch": 4.495956873315364,
"grad_norm": 0.6265653371810913,
"learning_rate": 0.00033070912034538583,
"loss": 3.4041,
"step": 41700
},
{
"epoch": 4.501347708894879,
"grad_norm": 0.6319632530212402,
"learning_rate": 0.0003303853211009174,
"loss": 3.388,
"step": 41750
},
{
"epoch": 4.506738544474393,
"grad_norm": 0.6831185817718506,
"learning_rate": 0.000330061521856449,
"loss": 3.3904,
"step": 41800
},
{
"epoch": 4.512129380053908,
"grad_norm": 0.6215836405754089,
"learning_rate": 0.00032973772261198054,
"loss": 3.3795,
"step": 41850
},
{
"epoch": 4.517520215633423,
"grad_norm": 0.6432912945747375,
"learning_rate": 0.00032941392336751214,
"loss": 3.3905,
"step": 41900
},
{
"epoch": 4.5229110512129385,
"grad_norm": 0.6383686065673828,
"learning_rate": 0.0003290901241230437,
"loss": 3.373,
"step": 41950
},
{
"epoch": 4.528301886792453,
"grad_norm": 0.6388788223266602,
"learning_rate": 0.0003287663248785753,
"loss": 3.3853,
"step": 42000
},
{
"epoch": 4.528301886792453,
"eval_accuracy": 0.37882426927401724,
"eval_loss": 3.4265851974487305,
"eval_runtime": 184.4493,
"eval_samples_per_second": 97.647,
"eval_steps_per_second": 6.105,
"step": 42000
},
{
"epoch": 4.533692722371968,
"grad_norm": 0.6462178826332092,
"learning_rate": 0.00032844252563410685,
"loss": 3.3813,
"step": 42050
},
{
"epoch": 4.539083557951482,
"grad_norm": 0.6646031141281128,
"learning_rate": 0.00032811872638963834,
"loss": 3.3901,
"step": 42100
},
{
"epoch": 4.544474393530997,
"grad_norm": 0.6592669486999512,
"learning_rate": 0.00032779492714516995,
"loss": 3.4179,
"step": 42150
},
{
"epoch": 4.549865229110512,
"grad_norm": 0.633387565612793,
"learning_rate": 0.0003274711279007015,
"loss": 3.385,
"step": 42200
},
{
"epoch": 4.555256064690027,
"grad_norm": 0.6689503788948059,
"learning_rate": 0.0003271473286562331,
"loss": 3.3839,
"step": 42250
},
{
"epoch": 4.560646900269542,
"grad_norm": 0.6962498426437378,
"learning_rate": 0.00032682352941176466,
"loss": 3.386,
"step": 42300
},
{
"epoch": 4.566037735849057,
"grad_norm": 0.6432573199272156,
"learning_rate": 0.00032649973016729626,
"loss": 3.3878,
"step": 42350
},
{
"epoch": 4.571428571428571,
"grad_norm": 0.6435944437980652,
"learning_rate": 0.0003261759309228278,
"loss": 3.3785,
"step": 42400
},
{
"epoch": 4.576819407008086,
"grad_norm": 0.6593371629714966,
"learning_rate": 0.0003258521316783594,
"loss": 3.3901,
"step": 42450
},
{
"epoch": 4.5822102425876015,
"grad_norm": 0.614346981048584,
"learning_rate": 0.00032552833243389097,
"loss": 3.3876,
"step": 42500
},
{
"epoch": 4.587601078167116,
"grad_norm": 0.6690718531608582,
"learning_rate": 0.0003252045331894225,
"loss": 3.3916,
"step": 42550
},
{
"epoch": 4.592991913746631,
"grad_norm": 0.6310614943504333,
"learning_rate": 0.0003248807339449541,
"loss": 3.3719,
"step": 42600
},
{
"epoch": 4.598382749326145,
"grad_norm": 0.6791559457778931,
"learning_rate": 0.00032455693470048567,
"loss": 3.3722,
"step": 42650
},
{
"epoch": 4.60377358490566,
"grad_norm": 0.6348100304603577,
"learning_rate": 0.0003242331354560173,
"loss": 3.3838,
"step": 42700
},
{
"epoch": 4.609164420485175,
"grad_norm": 0.6433316469192505,
"learning_rate": 0.0003239093362115488,
"loss": 3.401,
"step": 42750
},
{
"epoch": 4.6145552560646905,
"grad_norm": 0.6668508052825928,
"learning_rate": 0.00032358553696708043,
"loss": 3.3711,
"step": 42800
},
{
"epoch": 4.619946091644205,
"grad_norm": 0.6440048217773438,
"learning_rate": 0.00032326173772261193,
"loss": 3.3784,
"step": 42850
},
{
"epoch": 4.62533692722372,
"grad_norm": 0.6713200211524963,
"learning_rate": 0.00032293793847814353,
"loss": 3.3767,
"step": 42900
},
{
"epoch": 4.630727762803234,
"grad_norm": 0.6979366540908813,
"learning_rate": 0.0003226141392336751,
"loss": 3.3997,
"step": 42950
},
{
"epoch": 4.636118598382749,
"grad_norm": 0.688811719417572,
"learning_rate": 0.00032229681597409604,
"loss": 3.4043,
"step": 43000
},
{
"epoch": 4.636118598382749,
"eval_accuracy": 0.3792476895778303,
"eval_loss": 3.4223008155822754,
"eval_runtime": 184.5791,
"eval_samples_per_second": 97.579,
"eval_steps_per_second": 6.1,
"step": 43000
},
{
"epoch": 4.6415094339622645,
"grad_norm": 0.6326470971107483,
"learning_rate": 0.0003219730167296276,
"loss": 3.3887,
"step": 43050
},
{
"epoch": 4.646900269541779,
"grad_norm": 0.6202555894851685,
"learning_rate": 0.0003216492174851592,
"loss": 3.3944,
"step": 43100
},
{
"epoch": 4.652291105121294,
"grad_norm": 0.6769513487815857,
"learning_rate": 0.00032132541824069074,
"loss": 3.4076,
"step": 43150
},
{
"epoch": 4.657681940700809,
"grad_norm": 0.6604383587837219,
"learning_rate": 0.00032100161899622235,
"loss": 3.3791,
"step": 43200
},
{
"epoch": 4.663072776280323,
"grad_norm": 0.6450338959693909,
"learning_rate": 0.0003206778197517539,
"loss": 3.3801,
"step": 43250
},
{
"epoch": 4.668463611859838,
"grad_norm": 0.6374929547309875,
"learning_rate": 0.00032035402050728545,
"loss": 3.3892,
"step": 43300
},
{
"epoch": 4.6738544474393535,
"grad_norm": 0.6862507462501526,
"learning_rate": 0.00032003022126281705,
"loss": 3.3881,
"step": 43350
},
{
"epoch": 4.679245283018868,
"grad_norm": 0.6818158030509949,
"learning_rate": 0.00031970642201834855,
"loss": 3.4008,
"step": 43400
},
{
"epoch": 4.684636118598383,
"grad_norm": 0.6392462849617004,
"learning_rate": 0.0003193826227738802,
"loss": 3.3894,
"step": 43450
},
{
"epoch": 4.690026954177897,
"grad_norm": 0.687299907207489,
"learning_rate": 0.0003190588235294117,
"loss": 3.4017,
"step": 43500
},
{
"epoch": 4.695417789757412,
"grad_norm": 0.6224896907806396,
"learning_rate": 0.0003187350242849433,
"loss": 3.3855,
"step": 43550
},
{
"epoch": 4.7008086253369274,
"grad_norm": 0.6661508679389954,
"learning_rate": 0.00031841122504047486,
"loss": 3.3765,
"step": 43600
},
{
"epoch": 4.706199460916442,
"grad_norm": 0.6471050977706909,
"learning_rate": 0.00031808742579600646,
"loss": 3.391,
"step": 43650
},
{
"epoch": 4.711590296495957,
"grad_norm": 0.6618845462799072,
"learning_rate": 0.000317763626551538,
"loss": 3.3775,
"step": 43700
},
{
"epoch": 4.716981132075472,
"grad_norm": 0.6750358939170837,
"learning_rate": 0.00031743982730706957,
"loss": 3.3978,
"step": 43750
},
{
"epoch": 4.722371967654986,
"grad_norm": 0.6272256374359131,
"learning_rate": 0.00031711602806260117,
"loss": 3.3907,
"step": 43800
},
{
"epoch": 4.727762803234501,
"grad_norm": 0.7351711392402649,
"learning_rate": 0.0003167922288181327,
"loss": 3.3915,
"step": 43850
},
{
"epoch": 4.7331536388140165,
"grad_norm": 0.6072528958320618,
"learning_rate": 0.0003164684295736643,
"loss": 3.3935,
"step": 43900
},
{
"epoch": 4.738544474393531,
"grad_norm": 0.6813111305236816,
"learning_rate": 0.0003161446303291959,
"loss": 3.3866,
"step": 43950
},
{
"epoch": 4.743935309973046,
"grad_norm": 0.6698514223098755,
"learning_rate": 0.0003158208310847275,
"loss": 3.3898,
"step": 44000
},
{
"epoch": 4.743935309973046,
"eval_accuracy": 0.37985266885718133,
"eval_loss": 3.417273759841919,
"eval_runtime": 184.1826,
"eval_samples_per_second": 97.789,
"eval_steps_per_second": 6.113,
"step": 44000
},
{
"epoch": 4.74932614555256,
"grad_norm": 0.6945793032646179,
"learning_rate": 0.00031549703184025903,
"loss": 3.4,
"step": 44050
},
{
"epoch": 4.754716981132075,
"grad_norm": 0.6552477478981018,
"learning_rate": 0.00031517323259579064,
"loss": 3.3932,
"step": 44100
},
{
"epoch": 4.7601078167115904,
"grad_norm": 0.6777626276016235,
"learning_rate": 0.00031484943335132213,
"loss": 3.4033,
"step": 44150
},
{
"epoch": 4.765498652291106,
"grad_norm": 0.637958824634552,
"learning_rate": 0.0003145256341068537,
"loss": 3.4018,
"step": 44200
},
{
"epoch": 4.77088948787062,
"grad_norm": 0.6243258118629456,
"learning_rate": 0.0003142018348623853,
"loss": 3.3612,
"step": 44250
},
{
"epoch": 4.776280323450135,
"grad_norm": 0.647382915019989,
"learning_rate": 0.00031387803561791684,
"loss": 3.3877,
"step": 44300
},
{
"epoch": 4.781671159029649,
"grad_norm": 0.6556174159049988,
"learning_rate": 0.00031355423637344844,
"loss": 3.3923,
"step": 44350
},
{
"epoch": 4.787061994609164,
"grad_norm": 0.6015167236328125,
"learning_rate": 0.00031323043712898,
"loss": 3.3765,
"step": 44400
},
{
"epoch": 4.7924528301886795,
"grad_norm": 0.6569023132324219,
"learning_rate": 0.0003129066378845116,
"loss": 3.3987,
"step": 44450
},
{
"epoch": 4.797843665768194,
"grad_norm": 0.6894419193267822,
"learning_rate": 0.00031258283864004315,
"loss": 3.3978,
"step": 44500
},
{
"epoch": 4.803234501347709,
"grad_norm": 0.6735237836837769,
"learning_rate": 0.0003122590393955747,
"loss": 3.3693,
"step": 44550
},
{
"epoch": 4.808625336927224,
"grad_norm": 0.6733729839324951,
"learning_rate": 0.0003119352401511063,
"loss": 3.385,
"step": 44600
},
{
"epoch": 4.814016172506738,
"grad_norm": 0.6388543248176575,
"learning_rate": 0.00031161144090663786,
"loss": 3.3856,
"step": 44650
},
{
"epoch": 4.819407008086253,
"grad_norm": 0.6808621287345886,
"learning_rate": 0.00031128764166216946,
"loss": 3.3861,
"step": 44700
},
{
"epoch": 4.824797843665769,
"grad_norm": 0.6478013396263123,
"learning_rate": 0.00031096384241770096,
"loss": 3.4017,
"step": 44750
},
{
"epoch": 4.830188679245283,
"grad_norm": 0.7333654761314392,
"learning_rate": 0.0003106400431732326,
"loss": 3.3713,
"step": 44800
},
{
"epoch": 4.835579514824798,
"grad_norm": 0.6697977781295776,
"learning_rate": 0.0003103162439287641,
"loss": 3.3859,
"step": 44850
},
{
"epoch": 4.840970350404312,
"grad_norm": 0.6435649394989014,
"learning_rate": 0.0003099924446842957,
"loss": 3.3634,
"step": 44900
},
{
"epoch": 4.846361185983827,
"grad_norm": 0.6521356701850891,
"learning_rate": 0.00030966864543982727,
"loss": 3.4,
"step": 44950
},
{
"epoch": 4.8517520215633425,
"grad_norm": 0.6503953337669373,
"learning_rate": 0.0003093448461953588,
"loss": 3.3855,
"step": 45000
},
{
"epoch": 4.8517520215633425,
"eval_accuracy": 0.3807922990310009,
"eval_loss": 3.409809112548828,
"eval_runtime": 184.4072,
"eval_samples_per_second": 97.67,
"eval_steps_per_second": 6.106,
"step": 45000
},
{
"epoch": 4.857142857142857,
"grad_norm": 0.6946899890899658,
"learning_rate": 0.0003090210469508904,
"loss": 3.3966,
"step": 45050
},
{
"epoch": 4.862533692722372,
"grad_norm": 0.6561038494110107,
"learning_rate": 0.00030869724770642197,
"loss": 3.4003,
"step": 45100
},
{
"epoch": 4.867924528301887,
"grad_norm": 0.6713065505027771,
"learning_rate": 0.0003083734484619536,
"loss": 3.3698,
"step": 45150
},
{
"epoch": 4.873315363881401,
"grad_norm": 0.6470645070075989,
"learning_rate": 0.00030804964921748513,
"loss": 3.3789,
"step": 45200
},
{
"epoch": 4.878706199460916,
"grad_norm": 0.7061941623687744,
"learning_rate": 0.0003077323259579061,
"loss": 3.3956,
"step": 45250
},
{
"epoch": 4.884097035040432,
"grad_norm": 0.6762838959693909,
"learning_rate": 0.00030740852671343763,
"loss": 3.4048,
"step": 45300
},
{
"epoch": 4.889487870619946,
"grad_norm": 0.6910016536712646,
"learning_rate": 0.00030708472746896924,
"loss": 3.3856,
"step": 45350
},
{
"epoch": 4.894878706199461,
"grad_norm": 0.6601076722145081,
"learning_rate": 0.00030676092822450073,
"loss": 3.394,
"step": 45400
},
{
"epoch": 4.900269541778976,
"grad_norm": 0.6640689373016357,
"learning_rate": 0.0003064371289800324,
"loss": 3.3918,
"step": 45450
},
{
"epoch": 4.90566037735849,
"grad_norm": 0.7084178924560547,
"learning_rate": 0.0003061133297355639,
"loss": 3.3894,
"step": 45500
},
{
"epoch": 4.9110512129380055,
"grad_norm": 0.6426222324371338,
"learning_rate": 0.0003057895304910955,
"loss": 3.3905,
"step": 45550
},
{
"epoch": 4.916442048517521,
"grad_norm": 0.6365821361541748,
"learning_rate": 0.00030546573124662704,
"loss": 3.3858,
"step": 45600
},
{
"epoch": 4.921832884097035,
"grad_norm": 0.6634809970855713,
"learning_rate": 0.00030514193200215865,
"loss": 3.3892,
"step": 45650
},
{
"epoch": 4.92722371967655,
"grad_norm": 0.7642471194267273,
"learning_rate": 0.0003048181327576902,
"loss": 3.3634,
"step": 45700
},
{
"epoch": 4.932614555256064,
"grad_norm": 0.7017784118652344,
"learning_rate": 0.00030449433351322175,
"loss": 3.3835,
"step": 45750
},
{
"epoch": 4.938005390835579,
"grad_norm": 0.700731098651886,
"learning_rate": 0.00030417053426875335,
"loss": 3.4001,
"step": 45800
},
{
"epoch": 4.943396226415095,
"grad_norm": 0.7498189806938171,
"learning_rate": 0.0003038467350242849,
"loss": 3.3792,
"step": 45850
},
{
"epoch": 4.948787061994609,
"grad_norm": 0.6925360560417175,
"learning_rate": 0.0003035229357798165,
"loss": 3.3875,
"step": 45900
},
{
"epoch": 4.954177897574124,
"grad_norm": 0.6302282214164734,
"learning_rate": 0.00030319913653534806,
"loss": 3.3783,
"step": 45950
},
{
"epoch": 4.959568733153639,
"grad_norm": 0.6330150365829468,
"learning_rate": 0.00030287533729087966,
"loss": 3.3849,
"step": 46000
},
{
"epoch": 4.959568733153639,
"eval_accuracy": 0.38071156993535477,
"eval_loss": 3.409865379333496,
"eval_runtime": 184.3955,
"eval_samples_per_second": 97.676,
"eval_steps_per_second": 6.106,
"step": 46000
},
{
"epoch": 4.964959568733153,
"grad_norm": 0.678231418132782,
"learning_rate": 0.0003025515380464112,
"loss": 3.3946,
"step": 46050
},
{
"epoch": 4.9703504043126685,
"grad_norm": 0.6548928022384644,
"learning_rate": 0.0003022277388019428,
"loss": 3.3667,
"step": 46100
},
{
"epoch": 4.975741239892184,
"grad_norm": 0.6708270311355591,
"learning_rate": 0.0003019039395574743,
"loss": 3.3732,
"step": 46150
},
{
"epoch": 4.981132075471698,
"grad_norm": 0.7039728164672852,
"learning_rate": 0.00030158014031300587,
"loss": 3.3858,
"step": 46200
},
{
"epoch": 4.986522911051213,
"grad_norm": 0.644809365272522,
"learning_rate": 0.00030125634106853747,
"loss": 3.4017,
"step": 46250
},
{
"epoch": 4.991913746630727,
"grad_norm": 0.6609100699424744,
"learning_rate": 0.000300932541824069,
"loss": 3.3979,
"step": 46300
},
{
"epoch": 4.997304582210242,
"grad_norm": 0.6889398694038391,
"learning_rate": 0.0003006087425796006,
"loss": 3.3766,
"step": 46350
},
{
"epoch": 5.002695417789758,
"grad_norm": 0.6859802007675171,
"learning_rate": 0.0003002849433351322,
"loss": 3.3253,
"step": 46400
},
{
"epoch": 5.008086253369272,
"grad_norm": 0.7371976971626282,
"learning_rate": 0.00029996114409066373,
"loss": 3.3048,
"step": 46450
},
{
"epoch": 5.013477088948787,
"grad_norm": 0.6555555462837219,
"learning_rate": 0.00029963734484619533,
"loss": 3.3043,
"step": 46500
},
{
"epoch": 5.018867924528302,
"grad_norm": 0.6924049258232117,
"learning_rate": 0.0002993135456017269,
"loss": 3.3052,
"step": 46550
},
{
"epoch": 5.024258760107816,
"grad_norm": 0.7244302034378052,
"learning_rate": 0.0002989897463572585,
"loss": 3.2892,
"step": 46600
},
{
"epoch": 5.0296495956873315,
"grad_norm": 0.6777083277702332,
"learning_rate": 0.00029866594711279004,
"loss": 3.2939,
"step": 46650
},
{
"epoch": 5.035040431266847,
"grad_norm": 0.656471312046051,
"learning_rate": 0.00029834214786832164,
"loss": 3.2957,
"step": 46700
},
{
"epoch": 5.040431266846361,
"grad_norm": 0.6822708249092102,
"learning_rate": 0.0002980183486238532,
"loss": 3.3009,
"step": 46750
},
{
"epoch": 5.045822102425876,
"grad_norm": 0.7524915337562561,
"learning_rate": 0.00029769454937938474,
"loss": 3.3109,
"step": 46800
},
{
"epoch": 5.051212938005391,
"grad_norm": 0.6873501539230347,
"learning_rate": 0.0002973707501349163,
"loss": 3.3101,
"step": 46850
},
{
"epoch": 5.056603773584905,
"grad_norm": 0.6635301113128662,
"learning_rate": 0.0002970469508904479,
"loss": 3.3039,
"step": 46900
},
{
"epoch": 5.061994609164421,
"grad_norm": 0.6969589591026306,
"learning_rate": 0.00029672315164597945,
"loss": 3.2999,
"step": 46950
},
{
"epoch": 5.067385444743936,
"grad_norm": 0.6859714984893799,
"learning_rate": 0.00029639935240151105,
"loss": 3.3142,
"step": 47000
},
{
"epoch": 5.067385444743936,
"eval_accuracy": 0.3811283624130111,
"eval_loss": 3.410403251647949,
"eval_runtime": 184.4721,
"eval_samples_per_second": 97.635,
"eval_steps_per_second": 6.104,
"step": 47000
},
{
"epoch": 5.07277628032345,
"grad_norm": 0.6961098313331604,
"learning_rate": 0.0002960755531570426,
"loss": 3.3057,
"step": 47050
},
{
"epoch": 5.078167115902965,
"grad_norm": 0.6898601651191711,
"learning_rate": 0.0002957517539125742,
"loss": 3.3004,
"step": 47100
},
{
"epoch": 5.083557951482479,
"grad_norm": 0.6373685598373413,
"learning_rate": 0.00029542795466810576,
"loss": 3.3122,
"step": 47150
},
{
"epoch": 5.0889487870619945,
"grad_norm": 0.6678171753883362,
"learning_rate": 0.0002951041554236373,
"loss": 3.3103,
"step": 47200
},
{
"epoch": 5.09433962264151,
"grad_norm": 0.6630333065986633,
"learning_rate": 0.00029478035617916886,
"loss": 3.2957,
"step": 47250
},
{
"epoch": 5.099730458221024,
"grad_norm": 0.7204725742340088,
"learning_rate": 0.00029445655693470047,
"loss": 3.3214,
"step": 47300
},
{
"epoch": 5.105121293800539,
"grad_norm": 0.6941262483596802,
"learning_rate": 0.000294132757690232,
"loss": 3.319,
"step": 47350
},
{
"epoch": 5.110512129380054,
"grad_norm": 0.6715006828308105,
"learning_rate": 0.0002938089584457636,
"loss": 3.3016,
"step": 47400
},
{
"epoch": 5.115902964959568,
"grad_norm": 0.6809511780738831,
"learning_rate": 0.00029348515920129517,
"loss": 3.3046,
"step": 47450
},
{
"epoch": 5.121293800539084,
"grad_norm": 0.6851920485496521,
"learning_rate": 0.0002931613599568267,
"loss": 3.2949,
"step": 47500
},
{
"epoch": 5.126684636118599,
"grad_norm": 0.6803140044212341,
"learning_rate": 0.00029283756071235833,
"loss": 3.3028,
"step": 47550
},
{
"epoch": 5.132075471698113,
"grad_norm": 0.7452014088630676,
"learning_rate": 0.0002925137614678899,
"loss": 3.2951,
"step": 47600
},
{
"epoch": 5.137466307277628,
"grad_norm": 0.8049004673957825,
"learning_rate": 0.00029218996222342143,
"loss": 3.3078,
"step": 47650
},
{
"epoch": 5.142857142857143,
"grad_norm": 0.7158187627792358,
"learning_rate": 0.00029186616297895303,
"loss": 3.3235,
"step": 47700
},
{
"epoch": 5.1482479784366575,
"grad_norm": 0.7052319645881653,
"learning_rate": 0.0002915423637344846,
"loss": 3.3119,
"step": 47750
},
{
"epoch": 5.153638814016173,
"grad_norm": 0.6764992475509644,
"learning_rate": 0.00029121856449001613,
"loss": 3.2974,
"step": 47800
},
{
"epoch": 5.159029649595688,
"grad_norm": 0.6758170127868652,
"learning_rate": 0.00029089476524554774,
"loss": 3.3048,
"step": 47850
},
{
"epoch": 5.164420485175202,
"grad_norm": 0.6934885382652283,
"learning_rate": 0.0002905709660010793,
"loss": 3.3302,
"step": 47900
},
{
"epoch": 5.169811320754717,
"grad_norm": 0.6825749278068542,
"learning_rate": 0.0002902471667566109,
"loss": 3.3125,
"step": 47950
},
{
"epoch": 5.175202156334231,
"grad_norm": 0.6711937785148621,
"learning_rate": 0.00028992336751214245,
"loss": 3.3061,
"step": 48000
},
{
"epoch": 5.175202156334231,
"eval_accuracy": 0.38135120948591866,
"eval_loss": 3.409146785736084,
"eval_runtime": 184.5208,
"eval_samples_per_second": 97.61,
"eval_steps_per_second": 6.102,
"step": 48000
},
{
"epoch": 5.180592991913747,
"grad_norm": 0.6625189185142517,
"learning_rate": 0.00028959956826767405,
"loss": 3.312,
"step": 48050
},
{
"epoch": 5.185983827493262,
"grad_norm": 0.6991748809814453,
"learning_rate": 0.00028927576902320555,
"loss": 3.3016,
"step": 48100
},
{
"epoch": 5.191374663072776,
"grad_norm": 0.7087170481681824,
"learning_rate": 0.00028895196977873715,
"loss": 3.3109,
"step": 48150
},
{
"epoch": 5.196765498652291,
"grad_norm": 0.6987467408180237,
"learning_rate": 0.0002886281705342687,
"loss": 3.3262,
"step": 48200
},
{
"epoch": 5.202156334231806,
"grad_norm": 0.6994493007659912,
"learning_rate": 0.0002883043712898003,
"loss": 3.3165,
"step": 48250
},
{
"epoch": 5.2075471698113205,
"grad_norm": 0.6978319883346558,
"learning_rate": 0.00028798057204533186,
"loss": 3.3193,
"step": 48300
},
{
"epoch": 5.212938005390836,
"grad_norm": 0.693446934223175,
"learning_rate": 0.00028765677280086346,
"loss": 3.3255,
"step": 48350
},
{
"epoch": 5.218328840970351,
"grad_norm": 0.7261744737625122,
"learning_rate": 0.000287332973556395,
"loss": 3.3205,
"step": 48400
},
{
"epoch": 5.223719676549865,
"grad_norm": 0.666606068611145,
"learning_rate": 0.0002870091743119266,
"loss": 3.3101,
"step": 48450
},
{
"epoch": 5.22911051212938,
"grad_norm": 0.717832088470459,
"learning_rate": 0.00028668537506745817,
"loss": 3.3465,
"step": 48500
},
{
"epoch": 5.234501347708895,
"grad_norm": 0.629780650138855,
"learning_rate": 0.0002863615758229897,
"loss": 3.3189,
"step": 48550
},
{
"epoch": 5.2398921832884096,
"grad_norm": 0.6390810012817383,
"learning_rate": 0.00028603777657852127,
"loss": 3.319,
"step": 48600
},
{
"epoch": 5.245283018867925,
"grad_norm": 0.6497933864593506,
"learning_rate": 0.0002857139773340529,
"loss": 3.3152,
"step": 48650
},
{
"epoch": 5.250673854447439,
"grad_norm": 0.8252655267715454,
"learning_rate": 0.0002853901780895844,
"loss": 3.3201,
"step": 48700
},
{
"epoch": 5.256064690026954,
"grad_norm": 0.733096182346344,
"learning_rate": 0.00028506637884511603,
"loss": 3.3384,
"step": 48750
},
{
"epoch": 5.261455525606469,
"grad_norm": 0.6700831055641174,
"learning_rate": 0.0002847425796006476,
"loss": 3.3394,
"step": 48800
},
{
"epoch": 5.2668463611859835,
"grad_norm": 0.6866247653961182,
"learning_rate": 0.00028441878035617913,
"loss": 3.3146,
"step": 48850
},
{
"epoch": 5.272237196765499,
"grad_norm": 0.6915675401687622,
"learning_rate": 0.00028409498111171073,
"loss": 3.3143,
"step": 48900
},
{
"epoch": 5.277628032345014,
"grad_norm": 0.7154318690299988,
"learning_rate": 0.0002837711818672423,
"loss": 3.3373,
"step": 48950
},
{
"epoch": 5.283018867924528,
"grad_norm": 0.6428692936897278,
"learning_rate": 0.00028344738262277384,
"loss": 3.3248,
"step": 49000
},
{
"epoch": 5.283018867924528,
"eval_accuracy": 0.3817475752206659,
"eval_loss": 3.4046876430511475,
"eval_runtime": 184.4143,
"eval_samples_per_second": 97.666,
"eval_steps_per_second": 6.106,
"step": 49000
},
{
"epoch": 5.288409703504043,
"grad_norm": 0.6393749713897705,
"learning_rate": 0.00028312358337830544,
"loss": 3.3335,
"step": 49050
},
{
"epoch": 5.293800539083558,
"grad_norm": 0.6843445897102356,
"learning_rate": 0.000282799784133837,
"loss": 3.3306,
"step": 49100
},
{
"epoch": 5.2991913746630726,
"grad_norm": 0.7235457301139832,
"learning_rate": 0.00028247598488936854,
"loss": 3.331,
"step": 49150
},
{
"epoch": 5.304582210242588,
"grad_norm": 0.7626351714134216,
"learning_rate": 0.00028215218564490015,
"loss": 3.3383,
"step": 49200
},
{
"epoch": 5.309973045822103,
"grad_norm": 0.6809303760528564,
"learning_rate": 0.0002818283864004317,
"loss": 3.325,
"step": 49250
},
{
"epoch": 5.315363881401617,
"grad_norm": 0.7182308435440063,
"learning_rate": 0.00028151106314085265,
"loss": 3.3251,
"step": 49300
},
{
"epoch": 5.320754716981132,
"grad_norm": 0.7420695424079895,
"learning_rate": 0.0002811872638963842,
"loss": 3.3357,
"step": 49350
},
{
"epoch": 5.3261455525606465,
"grad_norm": 0.827497124671936,
"learning_rate": 0.0002808634646519158,
"loss": 3.3227,
"step": 49400
},
{
"epoch": 5.331536388140162,
"grad_norm": 0.6738569140434265,
"learning_rate": 0.00028053966540744736,
"loss": 3.3153,
"step": 49450
},
{
"epoch": 5.336927223719677,
"grad_norm": 0.6914460062980652,
"learning_rate": 0.0002802158661629789,
"loss": 3.3413,
"step": 49500
},
{
"epoch": 5.342318059299191,
"grad_norm": 0.6819193363189697,
"learning_rate": 0.0002798920669185105,
"loss": 3.315,
"step": 49550
},
{
"epoch": 5.347708894878706,
"grad_norm": 0.7536855340003967,
"learning_rate": 0.00027956826767404206,
"loss": 3.3333,
"step": 49600
},
{
"epoch": 5.353099730458221,
"grad_norm": 0.7118741869926453,
"learning_rate": 0.00027924446842957367,
"loss": 3.3302,
"step": 49650
},
{
"epoch": 5.3584905660377355,
"grad_norm": 0.6901974678039551,
"learning_rate": 0.0002789206691851052,
"loss": 3.3374,
"step": 49700
},
{
"epoch": 5.363881401617251,
"grad_norm": 0.692066490650177,
"learning_rate": 0.00027859686994063677,
"loss": 3.3207,
"step": 49750
},
{
"epoch": 5.369272237196766,
"grad_norm": 0.6952269077301025,
"learning_rate": 0.0002782730706961683,
"loss": 3.3252,
"step": 49800
},
{
"epoch": 5.37466307277628,
"grad_norm": 0.6739320158958435,
"learning_rate": 0.0002779492714516999,
"loss": 3.3399,
"step": 49850
},
{
"epoch": 5.380053908355795,
"grad_norm": 0.6872915029525757,
"learning_rate": 0.00027762547220723147,
"loss": 3.334,
"step": 49900
},
{
"epoch": 5.38544474393531,
"grad_norm": 0.6614148020744324,
"learning_rate": 0.0002773016729627631,
"loss": 3.3192,
"step": 49950
},
{
"epoch": 5.390835579514825,
"grad_norm": 0.663847804069519,
"learning_rate": 0.00027697787371829463,
"loss": 3.3393,
"step": 50000
},
{
"epoch": 5.390835579514825,
"eval_accuracy": 0.3820735338841093,
"eval_loss": 3.398294448852539,
"eval_runtime": 184.1958,
"eval_samples_per_second": 97.782,
"eval_steps_per_second": 6.113,
"step": 50000
},
{
"epoch": 5.39622641509434,
"grad_norm": 0.6840115189552307,
"learning_rate": 0.0002766605504587156,
"loss": 3.35,
"step": 50050
},
{
"epoch": 5.401617250673855,
"grad_norm": 0.6586406230926514,
"learning_rate": 0.00027633675121424713,
"loss": 3.3403,
"step": 50100
},
{
"epoch": 5.407008086253369,
"grad_norm": 0.7220155000686646,
"learning_rate": 0.0002760129519697787,
"loss": 3.3297,
"step": 50150
},
{
"epoch": 5.412398921832884,
"grad_norm": 0.7050442099571228,
"learning_rate": 0.0002756891527253103,
"loss": 3.3147,
"step": 50200
},
{
"epoch": 5.4177897574123985,
"grad_norm": 0.6927298307418823,
"learning_rate": 0.00027536535348084184,
"loss": 3.3428,
"step": 50250
},
{
"epoch": 5.423180592991914,
"grad_norm": 0.6581311225891113,
"learning_rate": 0.00027504155423637344,
"loss": 3.337,
"step": 50300
},
{
"epoch": 5.428571428571429,
"grad_norm": 0.6973708271980286,
"learning_rate": 0.000274717754991905,
"loss": 3.3253,
"step": 50350
},
{
"epoch": 5.433962264150943,
"grad_norm": 0.7422338724136353,
"learning_rate": 0.0002743939557474366,
"loss": 3.3466,
"step": 50400
},
{
"epoch": 5.439353099730458,
"grad_norm": 0.7879976034164429,
"learning_rate": 0.00027407015650296815,
"loss": 3.3438,
"step": 50450
},
{
"epoch": 5.444743935309973,
"grad_norm": 0.65547114610672,
"learning_rate": 0.0002737463572584997,
"loss": 3.3293,
"step": 50500
},
{
"epoch": 5.450134770889488,
"grad_norm": 0.6915010213851929,
"learning_rate": 0.00027342255801403125,
"loss": 3.3237,
"step": 50550
},
{
"epoch": 5.455525606469003,
"grad_norm": 0.7033429741859436,
"learning_rate": 0.00027309875876956285,
"loss": 3.3419,
"step": 50600
},
{
"epoch": 5.460916442048518,
"grad_norm": 0.8794052600860596,
"learning_rate": 0.0002727749595250944,
"loss": 3.3332,
"step": 50650
},
{
"epoch": 5.466307277628032,
"grad_norm": 0.7087518572807312,
"learning_rate": 0.000272451160280626,
"loss": 3.3256,
"step": 50700
},
{
"epoch": 5.471698113207547,
"grad_norm": 0.6747452616691589,
"learning_rate": 0.00027212736103615756,
"loss": 3.3118,
"step": 50750
},
{
"epoch": 5.4770889487870615,
"grad_norm": 0.6884540915489197,
"learning_rate": 0.00027180356179168916,
"loss": 3.3413,
"step": 50800
},
{
"epoch": 5.482479784366577,
"grad_norm": 0.7081778049468994,
"learning_rate": 0.0002714797625472207,
"loss": 3.3242,
"step": 50850
},
{
"epoch": 5.487870619946092,
"grad_norm": 0.7211419939994812,
"learning_rate": 0.00027115596330275226,
"loss": 3.3345,
"step": 50900
},
{
"epoch": 5.493261455525606,
"grad_norm": 0.6839219927787781,
"learning_rate": 0.0002708321640582838,
"loss": 3.3518,
"step": 50950
},
{
"epoch": 5.498652291105121,
"grad_norm": 0.7011337876319885,
"learning_rate": 0.0002705083648138154,
"loss": 3.3144,
"step": 51000
},
{
"epoch": 5.498652291105121,
"eval_accuracy": 0.3824411066035857,
"eval_loss": 3.395504951477051,
"eval_runtime": 184.3092,
"eval_samples_per_second": 97.722,
"eval_steps_per_second": 6.109,
"step": 51000
},
{
"epoch": 5.504043126684636,
"grad_norm": 0.7008047103881836,
"learning_rate": 0.00027018456556934697,
"loss": 3.323,
"step": 51050
},
{
"epoch": 5.509433962264151,
"grad_norm": 0.6674851179122925,
"learning_rate": 0.0002698607663248786,
"loss": 3.3321,
"step": 51100
},
{
"epoch": 5.514824797843666,
"grad_norm": 0.6711278557777405,
"learning_rate": 0.0002695369670804101,
"loss": 3.3493,
"step": 51150
},
{
"epoch": 5.520215633423181,
"grad_norm": 0.707088828086853,
"learning_rate": 0.0002692131678359417,
"loss": 3.3237,
"step": 51200
},
{
"epoch": 5.525606469002695,
"grad_norm": 0.7699363827705383,
"learning_rate": 0.0002688893685914733,
"loss": 3.3281,
"step": 51250
},
{
"epoch": 5.53099730458221,
"grad_norm": 0.6992379426956177,
"learning_rate": 0.00026856556934700483,
"loss": 3.3417,
"step": 51300
},
{
"epoch": 5.536388140161725,
"grad_norm": 0.6499311923980713,
"learning_rate": 0.0002682417701025364,
"loss": 3.3197,
"step": 51350
},
{
"epoch": 5.54177897574124,
"grad_norm": 0.7100504040718079,
"learning_rate": 0.000267917970858068,
"loss": 3.3304,
"step": 51400
},
{
"epoch": 5.547169811320755,
"grad_norm": 0.742253839969635,
"learning_rate": 0.00026759417161359954,
"loss": 3.3475,
"step": 51450
},
{
"epoch": 5.55256064690027,
"grad_norm": 0.7217435836791992,
"learning_rate": 0.0002672703723691311,
"loss": 3.3341,
"step": 51500
},
{
"epoch": 5.557951482479784,
"grad_norm": 0.7127156853675842,
"learning_rate": 0.0002669465731246627,
"loss": 3.3259,
"step": 51550
},
{
"epoch": 5.563342318059299,
"grad_norm": 0.7210736870765686,
"learning_rate": 0.00026662277388019424,
"loss": 3.3517,
"step": 51600
},
{
"epoch": 5.568733153638814,
"grad_norm": 0.6677258610725403,
"learning_rate": 0.00026629897463572585,
"loss": 3.3479,
"step": 51650
},
{
"epoch": 5.574123989218329,
"grad_norm": 0.7746173739433289,
"learning_rate": 0.0002659751753912574,
"loss": 3.3292,
"step": 51700
},
{
"epoch": 5.579514824797844,
"grad_norm": 0.7025367617607117,
"learning_rate": 0.00026565137614678895,
"loss": 3.3265,
"step": 51750
},
{
"epoch": 5.584905660377358,
"grad_norm": 0.6894749999046326,
"learning_rate": 0.0002653275769023205,
"loss": 3.3406,
"step": 51800
},
{
"epoch": 5.590296495956873,
"grad_norm": 0.7871344685554504,
"learning_rate": 0.0002650037776578521,
"loss": 3.3348,
"step": 51850
},
{
"epoch": 5.595687331536388,
"grad_norm": 0.7126604318618774,
"learning_rate": 0.00026467997841338366,
"loss": 3.3424,
"step": 51900
},
{
"epoch": 5.601078167115903,
"grad_norm": 0.6941819787025452,
"learning_rate": 0.00026435617916891526,
"loss": 3.3308,
"step": 51950
},
{
"epoch": 5.606469002695418,
"grad_norm": 0.7720863223075867,
"learning_rate": 0.0002640323799244468,
"loss": 3.327,
"step": 52000
},
{
"epoch": 5.606469002695418,
"eval_accuracy": 0.38275891630044306,
"eval_loss": 3.3928589820861816,
"eval_runtime": 184.309,
"eval_samples_per_second": 97.722,
"eval_steps_per_second": 6.109,
"step": 52000
},
{
"epoch": 5.611859838274933,
"grad_norm": 0.6593359708786011,
"learning_rate": 0.0002637085806799784,
"loss": 3.3499,
"step": 52050
},
{
"epoch": 5.617250673854447,
"grad_norm": 0.7027683854103088,
"learning_rate": 0.00026338478143550997,
"loss": 3.3291,
"step": 52100
},
{
"epoch": 5.622641509433962,
"grad_norm": 0.7556778192520142,
"learning_rate": 0.00026306098219104157,
"loss": 3.3319,
"step": 52150
},
{
"epoch": 5.628032345013477,
"grad_norm": 0.7045661211013794,
"learning_rate": 0.00026273718294657307,
"loss": 3.3271,
"step": 52200
},
{
"epoch": 5.633423180592992,
"grad_norm": 0.7103812098503113,
"learning_rate": 0.00026241338370210467,
"loss": 3.3283,
"step": 52250
},
{
"epoch": 5.638814016172507,
"grad_norm": 0.696662962436676,
"learning_rate": 0.0002620895844576362,
"loss": 3.3301,
"step": 52300
},
{
"epoch": 5.644204851752022,
"grad_norm": 0.7251359224319458,
"learning_rate": 0.00026176578521316783,
"loss": 3.3389,
"step": 52350
},
{
"epoch": 5.649595687331536,
"grad_norm": 0.7035132646560669,
"learning_rate": 0.0002614419859686994,
"loss": 3.3266,
"step": 52400
},
{
"epoch": 5.654986522911051,
"grad_norm": 0.6961835622787476,
"learning_rate": 0.000261118186724231,
"loss": 3.3562,
"step": 52450
},
{
"epoch": 5.660377358490566,
"grad_norm": 0.764259934425354,
"learning_rate": 0.00026079438747976253,
"loss": 3.3305,
"step": 52500
},
{
"epoch": 5.665768194070081,
"grad_norm": 0.7517282366752625,
"learning_rate": 0.0002604705882352941,
"loss": 3.3467,
"step": 52550
},
{
"epoch": 5.671159029649596,
"grad_norm": 0.7037937641143799,
"learning_rate": 0.00026014678899082563,
"loss": 3.3602,
"step": 52600
},
{
"epoch": 5.67654986522911,
"grad_norm": 0.7822666168212891,
"learning_rate": 0.00025982298974635724,
"loss": 3.3315,
"step": 52650
},
{
"epoch": 5.681940700808625,
"grad_norm": 0.7283600568771362,
"learning_rate": 0.0002594991905018888,
"loss": 3.3207,
"step": 52700
},
{
"epoch": 5.6873315363881405,
"grad_norm": 0.7028849124908447,
"learning_rate": 0.0002591753912574204,
"loss": 3.3381,
"step": 52750
},
{
"epoch": 5.692722371967655,
"grad_norm": 0.6893435716629028,
"learning_rate": 0.00025885159201295195,
"loss": 3.3206,
"step": 52800
},
{
"epoch": 5.69811320754717,
"grad_norm": 0.7093868255615234,
"learning_rate": 0.0002585277927684835,
"loss": 3.3183,
"step": 52850
},
{
"epoch": 5.703504043126685,
"grad_norm": 0.7332795858383179,
"learning_rate": 0.0002582039935240151,
"loss": 3.3213,
"step": 52900
},
{
"epoch": 5.708894878706199,
"grad_norm": 0.7287384867668152,
"learning_rate": 0.00025788019427954665,
"loss": 3.3255,
"step": 52950
},
{
"epoch": 5.714285714285714,
"grad_norm": 0.7492279410362244,
"learning_rate": 0.00025755639503507826,
"loss": 3.3392,
"step": 53000
},
{
"epoch": 5.714285714285714,
"eval_accuracy": 0.3834841743266047,
"eval_loss": 3.386990785598755,
"eval_runtime": 184.5467,
"eval_samples_per_second": 97.596,
"eval_steps_per_second": 6.101,
"step": 53000
},
{
"epoch": 5.719676549865229,
"grad_norm": 0.7081884741783142,
"learning_rate": 0.00025723907177549915,
"loss": 3.3299,
"step": 53050
},
{
"epoch": 5.725067385444744,
"grad_norm": 0.7364562749862671,
"learning_rate": 0.00025691527253103076,
"loss": 3.3386,
"step": 53100
},
{
"epoch": 5.730458221024259,
"grad_norm": 0.6906730532646179,
"learning_rate": 0.0002565914732865623,
"loss": 3.3338,
"step": 53150
},
{
"epoch": 5.735849056603773,
"grad_norm": 0.7850713133811951,
"learning_rate": 0.00025626767404209386,
"loss": 3.3444,
"step": 53200
},
{
"epoch": 5.741239892183288,
"grad_norm": 0.7125964164733887,
"learning_rate": 0.00025594387479762546,
"loss": 3.3144,
"step": 53250
},
{
"epoch": 5.7466307277628035,
"grad_norm": 0.7189776301383972,
"learning_rate": 0.000255620075553157,
"loss": 3.3564,
"step": 53300
},
{
"epoch": 5.752021563342318,
"grad_norm": 0.6696745157241821,
"learning_rate": 0.00025529627630868857,
"loss": 3.3393,
"step": 53350
},
{
"epoch": 5.757412398921833,
"grad_norm": 0.7350406646728516,
"learning_rate": 0.00025497247706422017,
"loss": 3.3453,
"step": 53400
},
{
"epoch": 5.762803234501348,
"grad_norm": 0.6493203043937683,
"learning_rate": 0.0002546486778197517,
"loss": 3.342,
"step": 53450
},
{
"epoch": 5.768194070080862,
"grad_norm": 0.705399751663208,
"learning_rate": 0.00025432487857528327,
"loss": 3.3233,
"step": 53500
},
{
"epoch": 5.773584905660377,
"grad_norm": 0.7020214796066284,
"learning_rate": 0.0002540010793308149,
"loss": 3.3254,
"step": 53550
},
{
"epoch": 5.7789757412398925,
"grad_norm": 0.7862454652786255,
"learning_rate": 0.0002536772800863464,
"loss": 3.3441,
"step": 53600
},
{
"epoch": 5.784366576819407,
"grad_norm": 0.655527651309967,
"learning_rate": 0.00025335348084187803,
"loss": 3.3258,
"step": 53650
},
{
"epoch": 5.789757412398922,
"grad_norm": 0.7039090394973755,
"learning_rate": 0.0002530296815974096,
"loss": 3.3453,
"step": 53700
},
{
"epoch": 5.795148247978437,
"grad_norm": 0.7265486717224121,
"learning_rate": 0.0002527058823529412,
"loss": 3.3329,
"step": 53750
},
{
"epoch": 5.800539083557951,
"grad_norm": 0.7473453879356384,
"learning_rate": 0.0002523820831084727,
"loss": 3.3257,
"step": 53800
},
{
"epoch": 5.8059299191374665,
"grad_norm": 0.7098454236984253,
"learning_rate": 0.0002520582838640043,
"loss": 3.3291,
"step": 53850
},
{
"epoch": 5.811320754716981,
"grad_norm": 0.7181734442710876,
"learning_rate": 0.00025173448461953584,
"loss": 3.3299,
"step": 53900
},
{
"epoch": 5.816711590296496,
"grad_norm": 0.734505295753479,
"learning_rate": 0.00025141068537506744,
"loss": 3.3265,
"step": 53950
},
{
"epoch": 5.822102425876011,
"grad_norm": 0.7421190738677979,
"learning_rate": 0.000251086886130599,
"loss": 3.3218,
"step": 54000
},
{
"epoch": 5.822102425876011,
"eval_accuracy": 0.38402102824529605,
"eval_loss": 3.3826797008514404,
"eval_runtime": 184.0512,
"eval_samples_per_second": 97.859,
"eval_steps_per_second": 6.118,
"step": 54000
},
{
"epoch": 5.827493261455525,
"grad_norm": 0.6887789964675903,
"learning_rate": 0.0002507630868861306,
"loss": 3.334,
"step": 54050
},
{
"epoch": 5.83288409703504,
"grad_norm": 0.7157623767852783,
"learning_rate": 0.00025043928764166215,
"loss": 3.3217,
"step": 54100
},
{
"epoch": 5.8382749326145555,
"grad_norm": 0.7711674571037292,
"learning_rate": 0.00025011548839719375,
"loss": 3.3362,
"step": 54150
},
{
"epoch": 5.84366576819407,
"grad_norm": 0.7098376154899597,
"learning_rate": 0.0002497916891527253,
"loss": 3.326,
"step": 54200
},
{
"epoch": 5.849056603773585,
"grad_norm": 0.7083906531333923,
"learning_rate": 0.00024946788990825686,
"loss": 3.3265,
"step": 54250
},
{
"epoch": 5.8544474393531,
"grad_norm": 0.7594617605209351,
"learning_rate": 0.0002491440906637884,
"loss": 3.3261,
"step": 54300
},
{
"epoch": 5.859838274932614,
"grad_norm": 0.76218581199646,
"learning_rate": 0.00024882029141932,
"loss": 3.3404,
"step": 54350
},
{
"epoch": 5.8652291105121295,
"grad_norm": 0.7147226333618164,
"learning_rate": 0.00024849649217485156,
"loss": 3.3268,
"step": 54400
},
{
"epoch": 5.870619946091644,
"grad_norm": 0.7391573190689087,
"learning_rate": 0.00024817269293038317,
"loss": 3.3431,
"step": 54450
},
{
"epoch": 5.876010781671159,
"grad_norm": 0.712948203086853,
"learning_rate": 0.0002478488936859147,
"loss": 3.3428,
"step": 54500
},
{
"epoch": 5.881401617250674,
"grad_norm": 0.7360401153564453,
"learning_rate": 0.00024752509444144627,
"loss": 3.3151,
"step": 54550
},
{
"epoch": 5.886792452830189,
"grad_norm": 0.7227504253387451,
"learning_rate": 0.00024720129519697787,
"loss": 3.3424,
"step": 54600
},
{
"epoch": 5.892183288409703,
"grad_norm": 0.6818044781684875,
"learning_rate": 0.0002468774959525094,
"loss": 3.3241,
"step": 54650
},
{
"epoch": 5.8975741239892185,
"grad_norm": 0.7365143895149231,
"learning_rate": 0.00024655369670804097,
"loss": 3.33,
"step": 54700
},
{
"epoch": 5.902964959568733,
"grad_norm": 0.6979695558547974,
"learning_rate": 0.0002462298974635726,
"loss": 3.3392,
"step": 54750
},
{
"epoch": 5.908355795148248,
"grad_norm": 0.7166529893875122,
"learning_rate": 0.00024590609821910413,
"loss": 3.3322,
"step": 54800
},
{
"epoch": 5.913746630727763,
"grad_norm": 0.8114291429519653,
"learning_rate": 0.0002455822989746357,
"loss": 3.34,
"step": 54850
},
{
"epoch": 5.919137466307277,
"grad_norm": 0.7475738525390625,
"learning_rate": 0.0002452584997301673,
"loss": 3.3395,
"step": 54900
},
{
"epoch": 5.9245283018867925,
"grad_norm": 0.7053887844085693,
"learning_rate": 0.00024493470048569883,
"loss": 3.3418,
"step": 54950
},
{
"epoch": 5.929919137466308,
"grad_norm": 0.7225222587585449,
"learning_rate": 0.00024461090124123044,
"loss": 3.3259,
"step": 55000
},
{
"epoch": 5.929919137466308,
"eval_accuracy": 0.3841736855526754,
"eval_loss": 3.378859043121338,
"eval_runtime": 184.2092,
"eval_samples_per_second": 97.775,
"eval_steps_per_second": 6.113,
"step": 55000
},
{
"epoch": 5.935309973045822,
"grad_norm": 0.7196410298347473,
"learning_rate": 0.000244287101996762,
"loss": 3.3539,
"step": 55050
},
{
"epoch": 5.940700808625337,
"grad_norm": 0.7038321495056152,
"learning_rate": 0.00024396330275229354,
"loss": 3.3192,
"step": 55100
},
{
"epoch": 5.946091644204852,
"grad_norm": 0.709574282169342,
"learning_rate": 0.00024363950350782512,
"loss": 3.3253,
"step": 55150
},
{
"epoch": 5.951482479784366,
"grad_norm": 0.7362520098686218,
"learning_rate": 0.0002433157042633567,
"loss": 3.322,
"step": 55200
},
{
"epoch": 5.9568733153638815,
"grad_norm": 0.7210483551025391,
"learning_rate": 0.00024299190501888827,
"loss": 3.3286,
"step": 55250
},
{
"epoch": 5.962264150943396,
"grad_norm": 0.697116494178772,
"learning_rate": 0.00024266810577441985,
"loss": 3.3205,
"step": 55300
},
{
"epoch": 5.967654986522911,
"grad_norm": 0.7499076128005981,
"learning_rate": 0.00024234430652995143,
"loss": 3.3224,
"step": 55350
},
{
"epoch": 5.973045822102426,
"grad_norm": 0.7028916478157043,
"learning_rate": 0.00024202050728548298,
"loss": 3.3249,
"step": 55400
},
{
"epoch": 5.97843665768194,
"grad_norm": 0.7200966477394104,
"learning_rate": 0.00024169670804101456,
"loss": 3.3443,
"step": 55450
},
{
"epoch": 5.9838274932614555,
"grad_norm": 0.810178279876709,
"learning_rate": 0.0002413729087965461,
"loss": 3.3275,
"step": 55500
},
{
"epoch": 5.989218328840971,
"grad_norm": 0.7342184782028198,
"learning_rate": 0.00024104910955207768,
"loss": 3.3314,
"step": 55550
},
{
"epoch": 5.994609164420485,
"grad_norm": 0.7123401165008545,
"learning_rate": 0.00024072531030760926,
"loss": 3.3497,
"step": 55600
},
{
"epoch": 6.0,
"grad_norm": 1.6110509634017944,
"learning_rate": 0.00024040151106314084,
"loss": 3.3463,
"step": 55650
},
{
"epoch": 6.005390835579515,
"grad_norm": 0.7127512097358704,
"learning_rate": 0.0002400777118186724,
"loss": 3.2398,
"step": 55700
},
{
"epoch": 6.010781671159029,
"grad_norm": 0.7257969975471497,
"learning_rate": 0.00023975391257420397,
"loss": 3.2457,
"step": 55750
},
{
"epoch": 6.0161725067385445,
"grad_norm": 0.7758077383041382,
"learning_rate": 0.00023943011332973555,
"loss": 3.237,
"step": 55800
},
{
"epoch": 6.02156334231806,
"grad_norm": 0.7242987751960754,
"learning_rate": 0.00023910631408526712,
"loss": 3.2442,
"step": 55850
},
{
"epoch": 6.026954177897574,
"grad_norm": 0.703961968421936,
"learning_rate": 0.0002387825148407987,
"loss": 3.2444,
"step": 55900
},
{
"epoch": 6.032345013477089,
"grad_norm": 0.7014211416244507,
"learning_rate": 0.00023845871559633025,
"loss": 3.2396,
"step": 55950
},
{
"epoch": 6.037735849056604,
"grad_norm": 0.7079455256462097,
"learning_rate": 0.0002381349163518618,
"loss": 3.236,
"step": 56000
},
{
"epoch": 6.037735849056604,
"eval_accuracy": 0.38413935124012605,
"eval_loss": 3.382866382598877,
"eval_runtime": 183.6611,
"eval_samples_per_second": 98.066,
"eval_steps_per_second": 6.131,
"step": 56000
},
{
"epoch": 6.0431266846361185,
"grad_norm": 0.7044804096221924,
"learning_rate": 0.00023781111710739338,
"loss": 3.2449,
"step": 56050
},
{
"epoch": 6.048517520215634,
"grad_norm": 0.7535393238067627,
"learning_rate": 0.00023748731786292496,
"loss": 3.2425,
"step": 56100
},
{
"epoch": 6.053908355795148,
"grad_norm": 0.7775534987449646,
"learning_rate": 0.00023716351861845654,
"loss": 3.2582,
"step": 56150
},
{
"epoch": 6.059299191374663,
"grad_norm": 0.7492356896400452,
"learning_rate": 0.0002368397193739881,
"loss": 3.24,
"step": 56200
},
{
"epoch": 6.064690026954178,
"grad_norm": 0.7849737405776978,
"learning_rate": 0.0002365159201295197,
"loss": 3.2494,
"step": 56250
},
{
"epoch": 6.070080862533692,
"grad_norm": 0.6883946657180786,
"learning_rate": 0.00023619212088505127,
"loss": 3.2555,
"step": 56300
},
{
"epoch": 6.0754716981132075,
"grad_norm": 0.7407999634742737,
"learning_rate": 0.0002358683216405828,
"loss": 3.254,
"step": 56350
},
{
"epoch": 6.080862533692723,
"grad_norm": 0.7489752173423767,
"learning_rate": 0.00023554452239611437,
"loss": 3.2738,
"step": 56400
},
{
"epoch": 6.086253369272237,
"grad_norm": 0.7341773509979248,
"learning_rate": 0.00023522072315164595,
"loss": 3.261,
"step": 56450
},
{
"epoch": 6.091644204851752,
"grad_norm": 0.7299098968505859,
"learning_rate": 0.00023489692390717752,
"loss": 3.2435,
"step": 56500
},
{
"epoch": 6.097035040431267,
"grad_norm": 0.6886277794837952,
"learning_rate": 0.0002345731246627091,
"loss": 3.2735,
"step": 56550
},
{
"epoch": 6.1024258760107815,
"grad_norm": 0.7371305227279663,
"learning_rate": 0.00023424932541824068,
"loss": 3.2565,
"step": 56600
},
{
"epoch": 6.107816711590297,
"grad_norm": 0.7455871105194092,
"learning_rate": 0.00023392552617377226,
"loss": 3.2409,
"step": 56650
},
{
"epoch": 6.113207547169812,
"grad_norm": 0.7575079798698425,
"learning_rate": 0.00023360172692930384,
"loss": 3.2715,
"step": 56700
},
{
"epoch": 6.118598382749326,
"grad_norm": 0.7615463137626648,
"learning_rate": 0.00023327792768483539,
"loss": 3.2794,
"step": 56750
},
{
"epoch": 6.123989218328841,
"grad_norm": 0.7489755153656006,
"learning_rate": 0.00023295412844036694,
"loss": 3.2687,
"step": 56800
},
{
"epoch": 6.129380053908355,
"grad_norm": 0.7886244654655457,
"learning_rate": 0.00023263032919589851,
"loss": 3.2509,
"step": 56850
},
{
"epoch": 6.1347708894878705,
"grad_norm": 0.7457435131072998,
"learning_rate": 0.0002323065299514301,
"loss": 3.2685,
"step": 56900
},
{
"epoch": 6.140161725067386,
"grad_norm": 0.7080656290054321,
"learning_rate": 0.00023198273070696167,
"loss": 3.2513,
"step": 56950
},
{
"epoch": 6.1455525606469,
"grad_norm": 0.7249478101730347,
"learning_rate": 0.00023165893146249325,
"loss": 3.267,
"step": 57000
},
{
"epoch": 6.1455525606469,
"eval_accuracy": 0.3847680168490206,
"eval_loss": 3.3801534175872803,
"eval_runtime": 183.9885,
"eval_samples_per_second": 97.892,
"eval_steps_per_second": 6.12,
"step": 57000
},
{
"epoch": 6.150943396226415,
"grad_norm": 0.7496433258056641,
"learning_rate": 0.0002313416082029142,
"loss": 3.2593,
"step": 57050
},
{
"epoch": 6.15633423180593,
"grad_norm": 0.7463822364807129,
"learning_rate": 0.00023101780895844572,
"loss": 3.2671,
"step": 57100
},
{
"epoch": 6.1617250673854445,
"grad_norm": 0.7301409840583801,
"learning_rate": 0.0002306940097139773,
"loss": 3.2811,
"step": 57150
},
{
"epoch": 6.16711590296496,
"grad_norm": 0.7670462131500244,
"learning_rate": 0.00023037021046950888,
"loss": 3.2579,
"step": 57200
},
{
"epoch": 6.172506738544475,
"grad_norm": 0.7218698859214783,
"learning_rate": 0.00023004641122504046,
"loss": 3.2544,
"step": 57250
},
{
"epoch": 6.177897574123989,
"grad_norm": 0.8085379004478455,
"learning_rate": 0.00022972261198057203,
"loss": 3.2721,
"step": 57300
},
{
"epoch": 6.183288409703504,
"grad_norm": 0.7022005319595337,
"learning_rate": 0.0002293988127361036,
"loss": 3.2581,
"step": 57350
},
{
"epoch": 6.188679245283019,
"grad_norm": 0.7380808591842651,
"learning_rate": 0.00022907501349163516,
"loss": 3.2662,
"step": 57400
},
{
"epoch": 6.1940700808625335,
"grad_norm": 0.7618688344955444,
"learning_rate": 0.00022875121424716674,
"loss": 3.2622,
"step": 57450
},
{
"epoch": 6.199460916442049,
"grad_norm": 0.7552737593650818,
"learning_rate": 0.00022842741500269832,
"loss": 3.2549,
"step": 57500
},
{
"epoch": 6.204851752021563,
"grad_norm": 0.756237804889679,
"learning_rate": 0.00022810361575822987,
"loss": 3.2783,
"step": 57550
},
{
"epoch": 6.210242587601078,
"grad_norm": 0.794677197933197,
"learning_rate": 0.00022777981651376145,
"loss": 3.2648,
"step": 57600
},
{
"epoch": 6.215633423180593,
"grad_norm": 0.7384204864501953,
"learning_rate": 0.00022745601726929302,
"loss": 3.2799,
"step": 57650
},
{
"epoch": 6.2210242587601075,
"grad_norm": 0.792015016078949,
"learning_rate": 0.00022713221802482457,
"loss": 3.2719,
"step": 57700
},
{
"epoch": 6.226415094339623,
"grad_norm": 0.7712349891662598,
"learning_rate": 0.00022680841878035615,
"loss": 3.2623,
"step": 57750
},
{
"epoch": 6.231805929919138,
"grad_norm": 0.7555547952651978,
"learning_rate": 0.00022648461953588773,
"loss": 3.256,
"step": 57800
},
{
"epoch": 6.237196765498652,
"grad_norm": 0.7071518898010254,
"learning_rate": 0.0002261608202914193,
"loss": 3.2704,
"step": 57850
},
{
"epoch": 6.242587601078167,
"grad_norm": 0.7165206074714661,
"learning_rate": 0.00022583702104695088,
"loss": 3.2801,
"step": 57900
},
{
"epoch": 6.247978436657682,
"grad_norm": 0.7546677589416504,
"learning_rate": 0.00022551322180248246,
"loss": 3.2713,
"step": 57950
},
{
"epoch": 6.2533692722371965,
"grad_norm": 0.7709723114967346,
"learning_rate": 0.00022518942255801399,
"loss": 3.274,
"step": 58000
},
{
"epoch": 6.2533692722371965,
"eval_accuracy": 0.38488699176117747,
"eval_loss": 3.3794407844543457,
"eval_runtime": 183.8937,
"eval_samples_per_second": 97.942,
"eval_steps_per_second": 6.123,
"step": 58000
},
{
"epoch": 6.258760107816712,
"grad_norm": 0.7347730994224548,
"learning_rate": 0.00022486562331354556,
"loss": 3.2601,
"step": 58050
},
{
"epoch": 6.264150943396227,
"grad_norm": 0.7662330865859985,
"learning_rate": 0.00022454182406907714,
"loss": 3.2747,
"step": 58100
},
{
"epoch": 6.269541778975741,
"grad_norm": 0.7109397053718567,
"learning_rate": 0.00022421802482460872,
"loss": 3.2666,
"step": 58150
},
{
"epoch": 6.274932614555256,
"grad_norm": 0.7161251306533813,
"learning_rate": 0.0002238942255801403,
"loss": 3.2888,
"step": 58200
},
{
"epoch": 6.280323450134771,
"grad_norm": 0.7194538116455078,
"learning_rate": 0.00022357042633567187,
"loss": 3.2704,
"step": 58250
},
{
"epoch": 6.285714285714286,
"grad_norm": 0.7429550886154175,
"learning_rate": 0.00022324662709120345,
"loss": 3.2618,
"step": 58300
},
{
"epoch": 6.291105121293801,
"grad_norm": 0.767665684223175,
"learning_rate": 0.00022292282784673503,
"loss": 3.283,
"step": 58350
},
{
"epoch": 6.296495956873315,
"grad_norm": 0.7169710397720337,
"learning_rate": 0.00022259902860226655,
"loss": 3.2651,
"step": 58400
},
{
"epoch": 6.30188679245283,
"grad_norm": 0.7469593286514282,
"learning_rate": 0.00022227522935779813,
"loss": 3.2934,
"step": 58450
},
{
"epoch": 6.307277628032345,
"grad_norm": 0.6983391642570496,
"learning_rate": 0.0002219514301133297,
"loss": 3.2895,
"step": 58500
},
{
"epoch": 6.3126684636118595,
"grad_norm": 0.8202389478683472,
"learning_rate": 0.00022162763086886129,
"loss": 3.2819,
"step": 58550
},
{
"epoch": 6.318059299191375,
"grad_norm": 0.7569589018821716,
"learning_rate": 0.00022130383162439286,
"loss": 3.2833,
"step": 58600
},
{
"epoch": 6.32345013477089,
"grad_norm": 0.7391425371170044,
"learning_rate": 0.00022098003237992444,
"loss": 3.2803,
"step": 58650
},
{
"epoch": 6.328840970350404,
"grad_norm": 0.7779117226600647,
"learning_rate": 0.00022065623313545602,
"loss": 3.2808,
"step": 58700
},
{
"epoch": 6.334231805929919,
"grad_norm": 0.7451295852661133,
"learning_rate": 0.00022033243389098757,
"loss": 3.2689,
"step": 58750
},
{
"epoch": 6.339622641509434,
"grad_norm": 0.7289618849754333,
"learning_rate": 0.00022000863464651915,
"loss": 3.2779,
"step": 58800
},
{
"epoch": 6.345013477088949,
"grad_norm": 0.7240257263183594,
"learning_rate": 0.0002196848354020507,
"loss": 3.291,
"step": 58850
},
{
"epoch": 6.350404312668464,
"grad_norm": 0.7132275104522705,
"learning_rate": 0.00021936103615758227,
"loss": 3.2772,
"step": 58900
},
{
"epoch": 6.355795148247978,
"grad_norm": 0.7555942535400391,
"learning_rate": 0.00021903723691311385,
"loss": 3.2749,
"step": 58950
},
{
"epoch": 6.361185983827493,
"grad_norm": 0.7417033314704895,
"learning_rate": 0.00021871343766864543,
"loss": 3.2922,
"step": 59000
},
{
"epoch": 6.361185983827493,
"eval_accuracy": 0.3853418127495689,
"eval_loss": 3.373150110244751,
"eval_runtime": 183.8598,
"eval_samples_per_second": 97.96,
"eval_steps_per_second": 6.124,
"step": 59000
},
{
"epoch": 6.366576819407008,
"grad_norm": 0.748611569404602,
"learning_rate": 0.00021838963842417698,
"loss": 3.2779,
"step": 59050
},
{
"epoch": 6.3719676549865225,
"grad_norm": 0.7483624815940857,
"learning_rate": 0.00021807231516459793,
"loss": 3.277,
"step": 59100
},
{
"epoch": 6.377358490566038,
"grad_norm": 0.7252408266067505,
"learning_rate": 0.00021774851592012948,
"loss": 3.2704,
"step": 59150
},
{
"epoch": 6.382749326145553,
"grad_norm": 0.7631791234016418,
"learning_rate": 0.00021742471667566106,
"loss": 3.2742,
"step": 59200
},
{
"epoch": 6.388140161725067,
"grad_norm": 0.7321556806564331,
"learning_rate": 0.00021710091743119264,
"loss": 3.2782,
"step": 59250
},
{
"epoch": 6.393530997304582,
"grad_norm": 0.7184486389160156,
"learning_rate": 0.00021677711818672422,
"loss": 3.2805,
"step": 59300
},
{
"epoch": 6.398921832884097,
"grad_norm": 0.7134908437728882,
"learning_rate": 0.0002164533189422558,
"loss": 3.2771,
"step": 59350
},
{
"epoch": 6.404312668463612,
"grad_norm": 0.7659626603126526,
"learning_rate": 0.00021612951969778734,
"loss": 3.3009,
"step": 59400
},
{
"epoch": 6.409703504043127,
"grad_norm": 0.7915070652961731,
"learning_rate": 0.00021580572045331892,
"loss": 3.2857,
"step": 59450
},
{
"epoch": 6.415094339622642,
"grad_norm": 0.7575498819351196,
"learning_rate": 0.0002154819212088505,
"loss": 3.2741,
"step": 59500
},
{
"epoch": 6.420485175202156,
"grad_norm": 0.7700998783111572,
"learning_rate": 0.00021515812196438208,
"loss": 3.2706,
"step": 59550
},
{
"epoch": 6.425876010781671,
"grad_norm": 0.7642835974693298,
"learning_rate": 0.00021483432271991363,
"loss": 3.2933,
"step": 59600
},
{
"epoch": 6.431266846361186,
"grad_norm": 0.764178454875946,
"learning_rate": 0.0002145105234754452,
"loss": 3.2845,
"step": 59650
},
{
"epoch": 6.436657681940701,
"grad_norm": 0.7783209681510925,
"learning_rate": 0.00021418672423097676,
"loss": 3.2934,
"step": 59700
},
{
"epoch": 6.442048517520216,
"grad_norm": 0.8354188799858093,
"learning_rate": 0.00021386292498650833,
"loss": 3.2911,
"step": 59750
},
{
"epoch": 6.44743935309973,
"grad_norm": 0.7723036408424377,
"learning_rate": 0.0002135391257420399,
"loss": 3.2833,
"step": 59800
},
{
"epoch": 6.452830188679245,
"grad_norm": 0.722313642501831,
"learning_rate": 0.0002132153264975715,
"loss": 3.2663,
"step": 59850
},
{
"epoch": 6.45822102425876,
"grad_norm": 0.7518280148506165,
"learning_rate": 0.00021289152725310307,
"loss": 3.2784,
"step": 59900
},
{
"epoch": 6.463611859838275,
"grad_norm": 0.7570785284042358,
"learning_rate": 0.00021256772800863464,
"loss": 3.2696,
"step": 59950
},
{
"epoch": 6.46900269541779,
"grad_norm": 0.7355751991271973,
"learning_rate": 0.0002122439287641662,
"loss": 3.2824,
"step": 60000
},
{
"epoch": 6.46900269541779,
"eval_accuracy": 0.3859073510306433,
"eval_loss": 3.370386838912964,
"eval_runtime": 184.5463,
"eval_samples_per_second": 97.596,
"eval_steps_per_second": 6.101,
"step": 60000
},
{
"epoch": 6.474393530997305,
"grad_norm": 0.7765397429466248,
"learning_rate": 0.00021192012951969775,
"loss": 3.2759,
"step": 60050
},
{
"epoch": 6.479784366576819,
"grad_norm": 0.7365384101867676,
"learning_rate": 0.00021159633027522932,
"loss": 3.2913,
"step": 60100
},
{
"epoch": 6.485175202156334,
"grad_norm": 0.7551870346069336,
"learning_rate": 0.0002112725310307609,
"loss": 3.2678,
"step": 60150
},
{
"epoch": 6.490566037735849,
"grad_norm": 0.7545482516288757,
"learning_rate": 0.00021094873178629248,
"loss": 3.2822,
"step": 60200
},
{
"epoch": 6.495956873315364,
"grad_norm": 0.7891198992729187,
"learning_rate": 0.00021063140852671343,
"loss": 3.2771,
"step": 60250
},
{
"epoch": 6.501347708894879,
"grad_norm": 0.7503842115402222,
"learning_rate": 0.000210307609282245,
"loss": 3.2733,
"step": 60300
},
{
"epoch": 6.506738544474393,
"grad_norm": 0.7859472036361694,
"learning_rate": 0.00020998381003777653,
"loss": 3.2591,
"step": 60350
},
{
"epoch": 6.512129380053908,
"grad_norm": 0.8164756894111633,
"learning_rate": 0.0002096600107933081,
"loss": 3.2895,
"step": 60400
},
{
"epoch": 6.517520215633423,
"grad_norm": 0.8594491481781006,
"learning_rate": 0.0002093362115488397,
"loss": 3.2782,
"step": 60450
},
{
"epoch": 6.5229110512129385,
"grad_norm": 0.7433143258094788,
"learning_rate": 0.00020901241230437127,
"loss": 3.2749,
"step": 60500
},
{
"epoch": 6.528301886792453,
"grad_norm": 0.7539076209068298,
"learning_rate": 0.00020868861305990284,
"loss": 3.2961,
"step": 60550
},
{
"epoch": 6.533692722371968,
"grad_norm": 0.7461854815483093,
"learning_rate": 0.00020836481381543442,
"loss": 3.2685,
"step": 60600
},
{
"epoch": 6.539083557951482,
"grad_norm": 0.7276204824447632,
"learning_rate": 0.000208041014570966,
"loss": 3.2869,
"step": 60650
},
{
"epoch": 6.544474393530997,
"grad_norm": 0.7827343344688416,
"learning_rate": 0.00020771721532649758,
"loss": 3.2841,
"step": 60700
},
{
"epoch": 6.549865229110512,
"grad_norm": 0.7077012658119202,
"learning_rate": 0.00020739341608202915,
"loss": 3.288,
"step": 60750
},
{
"epoch": 6.555256064690027,
"grad_norm": 0.7708344459533691,
"learning_rate": 0.00020706961683756068,
"loss": 3.2833,
"step": 60800
},
{
"epoch": 6.560646900269542,
"grad_norm": 0.8155523538589478,
"learning_rate": 0.00020674581759309225,
"loss": 3.274,
"step": 60850
},
{
"epoch": 6.566037735849057,
"grad_norm": 0.7324163913726807,
"learning_rate": 0.00020642201834862383,
"loss": 3.2688,
"step": 60900
},
{
"epoch": 6.571428571428571,
"grad_norm": 0.8242278099060059,
"learning_rate": 0.0002060982191041554,
"loss": 3.2865,
"step": 60950
},
{
"epoch": 6.576819407008086,
"grad_norm": 0.7621654868125916,
"learning_rate": 0.000205774419859687,
"loss": 3.2733,
"step": 61000
},
{
"epoch": 6.576819407008086,
"eval_accuracy": 0.38621864155423175,
"eval_loss": 3.365771532058716,
"eval_runtime": 184.4738,
"eval_samples_per_second": 97.634,
"eval_steps_per_second": 6.104,
"step": 61000
},
{
"epoch": 6.5822102425876015,
"grad_norm": 0.7416156530380249,
"learning_rate": 0.00020545062061521857,
"loss": 3.2699,
"step": 61050
},
{
"epoch": 6.587601078167116,
"grad_norm": 0.752616822719574,
"learning_rate": 0.00020512682137075012,
"loss": 3.2779,
"step": 61100
},
{
"epoch": 6.592991913746631,
"grad_norm": 0.7422453165054321,
"learning_rate": 0.0002048030221262817,
"loss": 3.2944,
"step": 61150
},
{
"epoch": 6.598382749326145,
"grad_norm": 0.7171696424484253,
"learning_rate": 0.00020447922288181324,
"loss": 3.2921,
"step": 61200
},
{
"epoch": 6.60377358490566,
"grad_norm": 0.7786928415298462,
"learning_rate": 0.00020415542363734482,
"loss": 3.2834,
"step": 61250
},
{
"epoch": 6.609164420485175,
"grad_norm": 0.7486423254013062,
"learning_rate": 0.0002038316243928764,
"loss": 3.2749,
"step": 61300
},
{
"epoch": 6.6145552560646905,
"grad_norm": 0.8092007040977478,
"learning_rate": 0.00020350782514840798,
"loss": 3.2855,
"step": 61350
},
{
"epoch": 6.619946091644205,
"grad_norm": 0.7663987874984741,
"learning_rate": 0.00020318402590393953,
"loss": 3.287,
"step": 61400
},
{
"epoch": 6.62533692722372,
"grad_norm": 0.751883327960968,
"learning_rate": 0.0002028602266594711,
"loss": 3.2835,
"step": 61450
},
{
"epoch": 6.630727762803234,
"grad_norm": 0.7678404450416565,
"learning_rate": 0.00020253642741500268,
"loss": 3.279,
"step": 61500
},
{
"epoch": 6.636118598382749,
"grad_norm": 0.7480851411819458,
"learning_rate": 0.00020221262817053426,
"loss": 3.2776,
"step": 61550
},
{
"epoch": 6.6415094339622645,
"grad_norm": 0.7450754046440125,
"learning_rate": 0.00020188882892606584,
"loss": 3.3056,
"step": 61600
},
{
"epoch": 6.646900269541779,
"grad_norm": 0.7475999593734741,
"learning_rate": 0.0002015650296815974,
"loss": 3.2614,
"step": 61650
},
{
"epoch": 6.652291105121294,
"grad_norm": 0.8071983456611633,
"learning_rate": 0.00020124123043712897,
"loss": 3.2792,
"step": 61700
},
{
"epoch": 6.657681940700809,
"grad_norm": 0.7974144816398621,
"learning_rate": 0.00020091743119266052,
"loss": 3.2971,
"step": 61750
},
{
"epoch": 6.663072776280323,
"grad_norm": 0.7744117379188538,
"learning_rate": 0.0002005936319481921,
"loss": 3.2815,
"step": 61800
},
{
"epoch": 6.668463611859838,
"grad_norm": 0.7913119792938232,
"learning_rate": 0.00020026983270372367,
"loss": 3.2934,
"step": 61850
},
{
"epoch": 6.6738544474393535,
"grad_norm": 0.7899511456489563,
"learning_rate": 0.00019994603345925525,
"loss": 3.2827,
"step": 61900
},
{
"epoch": 6.679245283018868,
"grad_norm": 0.7601125240325928,
"learning_rate": 0.00019962223421478683,
"loss": 3.2779,
"step": 61950
},
{
"epoch": 6.684636118598383,
"grad_norm": 0.7415378093719482,
"learning_rate": 0.0001992984349703184,
"loss": 3.2758,
"step": 62000
},
{
"epoch": 6.684636118598383,
"eval_accuracy": 0.3865097226406867,
"eval_loss": 3.3623757362365723,
"eval_runtime": 184.6279,
"eval_samples_per_second": 97.553,
"eval_steps_per_second": 6.099,
"step": 62000
},
{
"epoch": 6.690026954177897,
"grad_norm": 0.7848173975944519,
"learning_rate": 0.00019897463572584993,
"loss": 3.2743,
"step": 62050
},
{
"epoch": 6.695417789757412,
"grad_norm": 0.7812760472297668,
"learning_rate": 0.0001986508364813815,
"loss": 3.2798,
"step": 62100
},
{
"epoch": 6.7008086253369274,
"grad_norm": 0.8310157060623169,
"learning_rate": 0.00019832703723691308,
"loss": 3.2972,
"step": 62150
},
{
"epoch": 6.706199460916442,
"grad_norm": 0.8402166366577148,
"learning_rate": 0.00019800323799244466,
"loss": 3.3045,
"step": 62200
},
{
"epoch": 6.711590296495957,
"grad_norm": 0.7929381132125854,
"learning_rate": 0.00019767943874797624,
"loss": 3.2798,
"step": 62250
},
{
"epoch": 6.716981132075472,
"grad_norm": 0.7510131001472473,
"learning_rate": 0.00019735563950350782,
"loss": 3.282,
"step": 62300
},
{
"epoch": 6.722371967654986,
"grad_norm": 0.7740292549133301,
"learning_rate": 0.0001970318402590394,
"loss": 3.2858,
"step": 62350
},
{
"epoch": 6.727762803234501,
"grad_norm": 0.7584080696105957,
"learning_rate": 0.00019670804101457097,
"loss": 3.2699,
"step": 62400
},
{
"epoch": 6.7331536388140165,
"grad_norm": 0.7748124003410339,
"learning_rate": 0.00019638424177010252,
"loss": 3.2755,
"step": 62450
},
{
"epoch": 6.738544474393531,
"grad_norm": 0.7892004251480103,
"learning_rate": 0.00019606044252563407,
"loss": 3.2834,
"step": 62500
},
{
"epoch": 6.743935309973046,
"grad_norm": 0.7743954658508301,
"learning_rate": 0.00019573664328116565,
"loss": 3.2829,
"step": 62550
},
{
"epoch": 6.74932614555256,
"grad_norm": 0.7706778645515442,
"learning_rate": 0.00019541284403669723,
"loss": 3.2763,
"step": 62600
},
{
"epoch": 6.754716981132075,
"grad_norm": 0.7655248045921326,
"learning_rate": 0.0001950890447922288,
"loss": 3.2897,
"step": 62650
},
{
"epoch": 6.7601078167115904,
"grad_norm": 0.8167433142662048,
"learning_rate": 0.00019476524554776038,
"loss": 3.2782,
"step": 62700
},
{
"epoch": 6.765498652291106,
"grad_norm": 0.7712922692298889,
"learning_rate": 0.00019444144630329193,
"loss": 3.2964,
"step": 62750
},
{
"epoch": 6.77088948787062,
"grad_norm": 0.7580761909484863,
"learning_rate": 0.0001941176470588235,
"loss": 3.2924,
"step": 62800
},
{
"epoch": 6.776280323450135,
"grad_norm": 0.8280974626541138,
"learning_rate": 0.0001937938478143551,
"loss": 3.2854,
"step": 62850
},
{
"epoch": 6.781671159029649,
"grad_norm": 0.7896142601966858,
"learning_rate": 0.00019347004856988664,
"loss": 3.2795,
"step": 62900
},
{
"epoch": 6.787061994609164,
"grad_norm": 0.7718231081962585,
"learning_rate": 0.00019314624932541822,
"loss": 3.2799,
"step": 62950
},
{
"epoch": 6.7924528301886795,
"grad_norm": 0.7429760694503784,
"learning_rate": 0.0001928224500809498,
"loss": 3.2863,
"step": 63000
},
{
"epoch": 6.7924528301886795,
"eval_accuracy": 0.3867697290012268,
"eval_loss": 3.358421564102173,
"eval_runtime": 184.6208,
"eval_samples_per_second": 97.557,
"eval_steps_per_second": 6.099,
"step": 63000
},
{
"epoch": 6.797843665768194,
"grad_norm": 0.7694290280342102,
"learning_rate": 0.00019249865083648137,
"loss": 3.2719,
"step": 63050
},
{
"epoch": 6.803234501347709,
"grad_norm": 0.8413361310958862,
"learning_rate": 0.00019217485159201292,
"loss": 3.2978,
"step": 63100
},
{
"epoch": 6.808625336927224,
"grad_norm": 0.7708566188812256,
"learning_rate": 0.0001918510523475445,
"loss": 3.2811,
"step": 63150
},
{
"epoch": 6.814016172506738,
"grad_norm": 0.7719883322715759,
"learning_rate": 0.00019152725310307608,
"loss": 3.2824,
"step": 63200
},
{
"epoch": 6.819407008086253,
"grad_norm": 0.7703409194946289,
"learning_rate": 0.00019120345385860766,
"loss": 3.2905,
"step": 63250
},
{
"epoch": 6.824797843665769,
"grad_norm": 0.7871072888374329,
"learning_rate": 0.00019087965461413923,
"loss": 3.2896,
"step": 63300
},
{
"epoch": 6.830188679245283,
"grad_norm": 0.767267107963562,
"learning_rate": 0.00019055585536967079,
"loss": 3.2809,
"step": 63350
},
{
"epoch": 6.835579514824798,
"grad_norm": 0.7878234386444092,
"learning_rate": 0.00019023205612520234,
"loss": 3.2745,
"step": 63400
},
{
"epoch": 6.840970350404312,
"grad_norm": 0.7571015954017639,
"learning_rate": 0.00018990825688073391,
"loss": 3.2794,
"step": 63450
},
{
"epoch": 6.846361185983827,
"grad_norm": 0.747545063495636,
"learning_rate": 0.0001895844576362655,
"loss": 3.2817,
"step": 63500
},
{
"epoch": 6.8517520215633425,
"grad_norm": 0.765789270401001,
"learning_rate": 0.00018926065839179707,
"loss": 3.2988,
"step": 63550
},
{
"epoch": 6.857142857142857,
"grad_norm": 0.8290010094642639,
"learning_rate": 0.00018893685914732865,
"loss": 3.2837,
"step": 63600
},
{
"epoch": 6.862533692722372,
"grad_norm": 0.8015212416648865,
"learning_rate": 0.00018861305990286022,
"loss": 3.2876,
"step": 63650
},
{
"epoch": 6.867924528301887,
"grad_norm": 0.7811967730522156,
"learning_rate": 0.0001882892606583918,
"loss": 3.2619,
"step": 63700
},
{
"epoch": 6.873315363881401,
"grad_norm": 0.7973096370697021,
"learning_rate": 0.00018796546141392333,
"loss": 3.2736,
"step": 63750
},
{
"epoch": 6.878706199460916,
"grad_norm": 0.8027676343917847,
"learning_rate": 0.0001876416621694549,
"loss": 3.2854,
"step": 63800
},
{
"epoch": 6.884097035040432,
"grad_norm": 0.7508664131164551,
"learning_rate": 0.00018731786292498648,
"loss": 3.2597,
"step": 63850
},
{
"epoch": 6.889487870619946,
"grad_norm": 0.7863243222236633,
"learning_rate": 0.00018699406368051806,
"loss": 3.2728,
"step": 63900
},
{
"epoch": 6.894878706199461,
"grad_norm": 0.7452736496925354,
"learning_rate": 0.00018667026443604964,
"loss": 3.291,
"step": 63950
},
{
"epoch": 6.900269541778976,
"grad_norm": 0.7612056136131287,
"learning_rate": 0.00018634646519158121,
"loss": 3.2991,
"step": 64000
},
{
"epoch": 6.900269541778976,
"eval_accuracy": 0.3873860081809105,
"eval_loss": 3.3538753986358643,
"eval_runtime": 184.3742,
"eval_samples_per_second": 97.687,
"eval_steps_per_second": 6.107,
"step": 64000
},
{
"epoch": 6.90566037735849,
"grad_norm": 0.7467326521873474,
"learning_rate": 0.0001860226659471128,
"loss": 3.2812,
"step": 64050
},
{
"epoch": 6.9110512129380055,
"grad_norm": 0.794744610786438,
"learning_rate": 0.00018569886670264434,
"loss": 3.2805,
"step": 64100
},
{
"epoch": 6.916442048517521,
"grad_norm": 0.7626374363899231,
"learning_rate": 0.00018537506745817592,
"loss": 3.2825,
"step": 64150
},
{
"epoch": 6.921832884097035,
"grad_norm": 0.7787663340568542,
"learning_rate": 0.00018505126821370747,
"loss": 3.3054,
"step": 64200
},
{
"epoch": 6.92722371967655,
"grad_norm": 0.7765306234359741,
"learning_rate": 0.00018472746896923905,
"loss": 3.2754,
"step": 64250
},
{
"epoch": 6.932614555256064,
"grad_norm": 0.7765306830406189,
"learning_rate": 0.00018441014570966,
"loss": 3.2821,
"step": 64300
},
{
"epoch": 6.938005390835579,
"grad_norm": 0.77068692445755,
"learning_rate": 0.00018408634646519158,
"loss": 3.2719,
"step": 64350
},
{
"epoch": 6.943396226415095,
"grad_norm": 0.7640039324760437,
"learning_rate": 0.00018376254722072316,
"loss": 3.2822,
"step": 64400
},
{
"epoch": 6.948787061994609,
"grad_norm": 0.7565797567367554,
"learning_rate": 0.00018344522396114405,
"loss": 3.2577,
"step": 64450
},
{
"epoch": 6.954177897574124,
"grad_norm": 0.7973952889442444,
"learning_rate": 0.00018312142471667563,
"loss": 3.2998,
"step": 64500
},
{
"epoch": 6.959568733153639,
"grad_norm": 0.7847186923027039,
"learning_rate": 0.0001827976254722072,
"loss": 3.2932,
"step": 64550
},
{
"epoch": 6.964959568733153,
"grad_norm": 0.7712359428405762,
"learning_rate": 0.00018247382622773879,
"loss": 3.2944,
"step": 64600
},
{
"epoch": 6.9703504043126685,
"grad_norm": 0.8099783658981323,
"learning_rate": 0.00018215002698327036,
"loss": 3.3064,
"step": 64650
},
{
"epoch": 6.975741239892184,
"grad_norm": 0.7673017978668213,
"learning_rate": 0.00018182622773880194,
"loss": 3.2965,
"step": 64700
},
{
"epoch": 6.981132075471698,
"grad_norm": 0.8172734379768372,
"learning_rate": 0.00018150242849433352,
"loss": 3.2861,
"step": 64750
},
{
"epoch": 6.986522911051213,
"grad_norm": 0.7932207584381104,
"learning_rate": 0.00018117862924986507,
"loss": 3.2614,
"step": 64800
},
{
"epoch": 6.991913746630727,
"grad_norm": 0.7419960498809814,
"learning_rate": 0.00018085483000539662,
"loss": 3.2745,
"step": 64850
},
{
"epoch": 6.997304582210242,
"grad_norm": 0.7454132437705994,
"learning_rate": 0.0001805310307609282,
"loss": 3.2769,
"step": 64900
},
{
"epoch": 7.002695417789758,
"grad_norm": 0.821409285068512,
"learning_rate": 0.00018020723151645978,
"loss": 3.2304,
"step": 64950
},
{
"epoch": 7.008086253369272,
"grad_norm": 0.9058651328086853,
"learning_rate": 0.00017988343227199135,
"loss": 3.2086,
"step": 65000
},
{
"epoch": 7.008086253369272,
"eval_accuracy": 0.38716077074447114,
"eval_loss": 3.358865737915039,
"eval_runtime": 184.4355,
"eval_samples_per_second": 97.655,
"eval_steps_per_second": 6.105,
"step": 65000
},
{
"epoch": 7.013477088948787,
"grad_norm": 0.8006051778793335,
"learning_rate": 0.00017955963302752293,
"loss": 3.1949,
"step": 65050
},
{
"epoch": 7.018867924528302,
"grad_norm": 0.8583719730377197,
"learning_rate": 0.00017923583378305448,
"loss": 3.2084,
"step": 65100
},
{
"epoch": 7.024258760107816,
"grad_norm": 0.7838130593299866,
"learning_rate": 0.00017891203453858606,
"loss": 3.2189,
"step": 65150
},
{
"epoch": 7.0296495956873315,
"grad_norm": 0.8301541805267334,
"learning_rate": 0.00017858823529411764,
"loss": 3.1982,
"step": 65200
},
{
"epoch": 7.035040431266847,
"grad_norm": 0.8048170804977417,
"learning_rate": 0.00017826443604964921,
"loss": 3.1905,
"step": 65250
},
{
"epoch": 7.040431266846361,
"grad_norm": 0.7822536826133728,
"learning_rate": 0.00017794063680518077,
"loss": 3.2079,
"step": 65300
},
{
"epoch": 7.045822102425876,
"grad_norm": 0.7671051621437073,
"learning_rate": 0.00017761683756071234,
"loss": 3.1892,
"step": 65350
},
{
"epoch": 7.051212938005391,
"grad_norm": 0.8259955048561096,
"learning_rate": 0.00017729303831624392,
"loss": 3.2073,
"step": 65400
},
{
"epoch": 7.056603773584905,
"grad_norm": 0.8007810115814209,
"learning_rate": 0.00017696923907177547,
"loss": 3.2055,
"step": 65450
},
{
"epoch": 7.061994609164421,
"grad_norm": 0.8390731811523438,
"learning_rate": 0.00017664543982730705,
"loss": 3.2046,
"step": 65500
},
{
"epoch": 7.067385444743936,
"grad_norm": 0.7977765798568726,
"learning_rate": 0.00017632164058283863,
"loss": 3.2174,
"step": 65550
},
{
"epoch": 7.07277628032345,
"grad_norm": 0.7610532641410828,
"learning_rate": 0.0001759978413383702,
"loss": 3.2145,
"step": 65600
},
{
"epoch": 7.078167115902965,
"grad_norm": 0.8389305472373962,
"learning_rate": 0.00017567404209390178,
"loss": 3.2028,
"step": 65650
},
{
"epoch": 7.083557951482479,
"grad_norm": 0.7955011129379272,
"learning_rate": 0.00017535024284943333,
"loss": 3.2133,
"step": 65700
},
{
"epoch": 7.0889487870619945,
"grad_norm": 0.8065895438194275,
"learning_rate": 0.00017502644360496488,
"loss": 3.2067,
"step": 65750
},
{
"epoch": 7.09433962264151,
"grad_norm": 0.8210367560386658,
"learning_rate": 0.00017470264436049646,
"loss": 3.2119,
"step": 65800
},
{
"epoch": 7.099730458221024,
"grad_norm": 0.752528727054596,
"learning_rate": 0.00017437884511602804,
"loss": 3.1906,
"step": 65850
},
{
"epoch": 7.105121293800539,
"grad_norm": 0.7784146070480347,
"learning_rate": 0.00017405504587155962,
"loss": 3.2068,
"step": 65900
},
{
"epoch": 7.110512129380054,
"grad_norm": 0.7695617079734802,
"learning_rate": 0.0001737312466270912,
"loss": 3.1933,
"step": 65950
},
{
"epoch": 7.115902964959568,
"grad_norm": 0.7499436736106873,
"learning_rate": 0.00017340744738262277,
"loss": 3.2196,
"step": 66000
},
{
"epoch": 7.115902964959568,
"eval_accuracy": 0.38776900961045657,
"eval_loss": 3.3572564125061035,
"eval_runtime": 184.6473,
"eval_samples_per_second": 97.543,
"eval_steps_per_second": 6.098,
"step": 66000
},
{
"epoch": 7.121293800539084,
"grad_norm": 0.7791917324066162,
"learning_rate": 0.00017308364813815435,
"loss": 3.2074,
"step": 66050
},
{
"epoch": 7.126684636118599,
"grad_norm": 0.7864148020744324,
"learning_rate": 0.00017275984889368593,
"loss": 3.2104,
"step": 66100
},
{
"epoch": 7.132075471698113,
"grad_norm": 0.7931457757949829,
"learning_rate": 0.00017243604964921745,
"loss": 3.2181,
"step": 66150
},
{
"epoch": 7.137466307277628,
"grad_norm": 0.793620228767395,
"learning_rate": 0.00017211225040474903,
"loss": 3.2055,
"step": 66200
},
{
"epoch": 7.142857142857143,
"grad_norm": 0.8230220079421997,
"learning_rate": 0.0001717884511602806,
"loss": 3.2233,
"step": 66250
},
{
"epoch": 7.1482479784366575,
"grad_norm": 0.8263036608695984,
"learning_rate": 0.00017146465191581218,
"loss": 3.2091,
"step": 66300
},
{
"epoch": 7.153638814016173,
"grad_norm": 0.7950703501701355,
"learning_rate": 0.00017114085267134376,
"loss": 3.2128,
"step": 66350
},
{
"epoch": 7.159029649595688,
"grad_norm": 0.7634537816047668,
"learning_rate": 0.00017081705342687534,
"loss": 3.2123,
"step": 66400
},
{
"epoch": 7.164420485175202,
"grad_norm": 0.7703549861907959,
"learning_rate": 0.0001704932541824069,
"loss": 3.221,
"step": 66450
},
{
"epoch": 7.169811320754717,
"grad_norm": 0.7877131700515747,
"learning_rate": 0.00017016945493793847,
"loss": 3.2098,
"step": 66500
},
{
"epoch": 7.175202156334231,
"grad_norm": 0.8138350248336792,
"learning_rate": 0.00016984565569347002,
"loss": 3.2262,
"step": 66550
},
{
"epoch": 7.180592991913747,
"grad_norm": 0.8078272342681885,
"learning_rate": 0.0001695218564490016,
"loss": 3.2121,
"step": 66600
},
{
"epoch": 7.185983827493262,
"grad_norm": 0.7591400742530823,
"learning_rate": 0.00016919805720453317,
"loss": 3.2272,
"step": 66650
},
{
"epoch": 7.191374663072776,
"grad_norm": 0.8127170205116272,
"learning_rate": 0.00016887425796006475,
"loss": 3.2185,
"step": 66700
},
{
"epoch": 7.196765498652291,
"grad_norm": 0.797642171382904,
"learning_rate": 0.00016855045871559633,
"loss": 3.2263,
"step": 66750
},
{
"epoch": 7.202156334231806,
"grad_norm": 0.7964255809783936,
"learning_rate": 0.00016822665947112788,
"loss": 3.2002,
"step": 66800
},
{
"epoch": 7.2075471698113205,
"grad_norm": 0.7668558955192566,
"learning_rate": 0.00016790286022665946,
"loss": 3.217,
"step": 66850
},
{
"epoch": 7.212938005390836,
"grad_norm": 0.8412344455718994,
"learning_rate": 0.00016757906098219103,
"loss": 3.2372,
"step": 66900
},
{
"epoch": 7.218328840970351,
"grad_norm": 0.8061116933822632,
"learning_rate": 0.0001672552617377226,
"loss": 3.2244,
"step": 66950
},
{
"epoch": 7.223719676549865,
"grad_norm": 0.7709745764732361,
"learning_rate": 0.00016693146249325416,
"loss": 3.2311,
"step": 67000
},
{
"epoch": 7.223719676549865,
"eval_accuracy": 0.3879154737032305,
"eval_loss": 3.3534882068634033,
"eval_runtime": 184.9839,
"eval_samples_per_second": 97.365,
"eval_steps_per_second": 6.087,
"step": 67000
},
{
"epoch": 7.22911051212938,
"grad_norm": 0.8037967681884766,
"learning_rate": 0.00016661413923367511,
"loss": 3.2242,
"step": 67050
},
{
"epoch": 7.234501347708895,
"grad_norm": 0.8536267876625061,
"learning_rate": 0.00016629033998920666,
"loss": 3.2262,
"step": 67100
},
{
"epoch": 7.2398921832884096,
"grad_norm": 0.7414260506629944,
"learning_rate": 0.00016596654074473824,
"loss": 3.2285,
"step": 67150
},
{
"epoch": 7.245283018867925,
"grad_norm": 0.8779622316360474,
"learning_rate": 0.00016564274150026982,
"loss": 3.2163,
"step": 67200
},
{
"epoch": 7.250673854447439,
"grad_norm": 0.8375000953674316,
"learning_rate": 0.0001653189422558014,
"loss": 3.2175,
"step": 67250
},
{
"epoch": 7.256064690026954,
"grad_norm": 0.8532291054725647,
"learning_rate": 0.00016499514301133298,
"loss": 3.2336,
"step": 67300
},
{
"epoch": 7.261455525606469,
"grad_norm": 0.8551412224769592,
"learning_rate": 0.00016467134376686453,
"loss": 3.2204,
"step": 67350
},
{
"epoch": 7.2668463611859835,
"grad_norm": 0.8138480186462402,
"learning_rate": 0.0001643475445223961,
"loss": 3.2294,
"step": 67400
},
{
"epoch": 7.272237196765499,
"grad_norm": 0.8155137300491333,
"learning_rate": 0.00016402374527792765,
"loss": 3.2372,
"step": 67450
},
{
"epoch": 7.277628032345014,
"grad_norm": 0.82570880651474,
"learning_rate": 0.00016369994603345923,
"loss": 3.2323,
"step": 67500
},
{
"epoch": 7.283018867924528,
"grad_norm": 0.9058399796485901,
"learning_rate": 0.0001633761467889908,
"loss": 3.224,
"step": 67550
},
{
"epoch": 7.288409703504043,
"grad_norm": 0.76006680727005,
"learning_rate": 0.0001630523475445224,
"loss": 3.2469,
"step": 67600
},
{
"epoch": 7.293800539083558,
"grad_norm": 0.9011563062667847,
"learning_rate": 0.00016272854830005396,
"loss": 3.2265,
"step": 67650
},
{
"epoch": 7.2991913746630726,
"grad_norm": 0.7784439325332642,
"learning_rate": 0.00016240474905558554,
"loss": 3.2289,
"step": 67700
},
{
"epoch": 7.304582210242588,
"grad_norm": 0.8018257021903992,
"learning_rate": 0.00016208094981111707,
"loss": 3.2315,
"step": 67750
},
{
"epoch": 7.309973045822103,
"grad_norm": 0.816515326499939,
"learning_rate": 0.00016175715056664864,
"loss": 3.2139,
"step": 67800
},
{
"epoch": 7.315363881401617,
"grad_norm": 0.8524076342582703,
"learning_rate": 0.00016143335132218022,
"loss": 3.2512,
"step": 67850
},
{
"epoch": 7.320754716981132,
"grad_norm": 0.7833554148674011,
"learning_rate": 0.0001611095520777118,
"loss": 3.2287,
"step": 67900
},
{
"epoch": 7.3261455525606465,
"grad_norm": 0.8669237494468689,
"learning_rate": 0.00016078575283324338,
"loss": 3.2336,
"step": 67950
},
{
"epoch": 7.331536388140162,
"grad_norm": 0.8306994438171387,
"learning_rate": 0.00016046195358877495,
"loss": 3.2127,
"step": 68000
},
{
"epoch": 7.331536388140162,
"eval_accuracy": 0.3882462130937378,
"eval_loss": 3.351630210876465,
"eval_runtime": 184.6763,
"eval_samples_per_second": 97.527,
"eval_steps_per_second": 6.097,
"step": 68000
},
{
"epoch": 7.336927223719677,
"grad_norm": 0.8202095627784729,
"learning_rate": 0.00016013815434430653,
"loss": 3.2338,
"step": 68050
},
{
"epoch": 7.342318059299191,
"grad_norm": 0.8112189173698425,
"learning_rate": 0.0001598143550998381,
"loss": 3.2378,
"step": 68100
},
{
"epoch": 7.347708894878706,
"grad_norm": 0.8299844264984131,
"learning_rate": 0.00015949055585536966,
"loss": 3.2199,
"step": 68150
},
{
"epoch": 7.353099730458221,
"grad_norm": 0.8318902850151062,
"learning_rate": 0.0001591667566109012,
"loss": 3.2443,
"step": 68200
},
{
"epoch": 7.3584905660377355,
"grad_norm": 0.8126823306083679,
"learning_rate": 0.0001588429573664328,
"loss": 3.2463,
"step": 68250
},
{
"epoch": 7.363881401617251,
"grad_norm": 0.8080394268035889,
"learning_rate": 0.00015851915812196437,
"loss": 3.2326,
"step": 68300
},
{
"epoch": 7.369272237196766,
"grad_norm": 0.832899272441864,
"learning_rate": 0.00015819535887749594,
"loss": 3.2333,
"step": 68350
},
{
"epoch": 7.37466307277628,
"grad_norm": 0.8399190306663513,
"learning_rate": 0.00015787155963302752,
"loss": 3.2229,
"step": 68400
},
{
"epoch": 7.380053908355795,
"grad_norm": 0.8199726939201355,
"learning_rate": 0.0001575477603885591,
"loss": 3.2091,
"step": 68450
},
{
"epoch": 7.38544474393531,
"grad_norm": 0.827198326587677,
"learning_rate": 0.00015722396114409065,
"loss": 3.2151,
"step": 68500
},
{
"epoch": 7.390835579514825,
"grad_norm": 0.7535482048988342,
"learning_rate": 0.00015690016189962223,
"loss": 3.2261,
"step": 68550
},
{
"epoch": 7.39622641509434,
"grad_norm": 0.7915380001068115,
"learning_rate": 0.00015657636265515378,
"loss": 3.2426,
"step": 68600
},
{
"epoch": 7.401617250673855,
"grad_norm": 0.8676249384880066,
"learning_rate": 0.00015625256341068536,
"loss": 3.2136,
"step": 68650
},
{
"epoch": 7.407008086253369,
"grad_norm": 0.8157048225402832,
"learning_rate": 0.00015592876416621693,
"loss": 3.2181,
"step": 68700
},
{
"epoch": 7.412398921832884,
"grad_norm": 0.8107733726501465,
"learning_rate": 0.0001556049649217485,
"loss": 3.2235,
"step": 68750
},
{
"epoch": 7.4177897574123985,
"grad_norm": 0.8157159090042114,
"learning_rate": 0.00015528116567728006,
"loss": 3.2346,
"step": 68800
},
{
"epoch": 7.423180592991914,
"grad_norm": 0.7814071774482727,
"learning_rate": 0.00015495736643281164,
"loss": 3.2176,
"step": 68850
},
{
"epoch": 7.428571428571429,
"grad_norm": 0.8208259344100952,
"learning_rate": 0.00015463356718834322,
"loss": 3.2339,
"step": 68900
},
{
"epoch": 7.433962264150943,
"grad_norm": 0.8454492688179016,
"learning_rate": 0.0001543097679438748,
"loss": 3.2253,
"step": 68950
},
{
"epoch": 7.439353099730458,
"grad_norm": 0.7895976305007935,
"learning_rate": 0.00015398596869940637,
"loss": 3.2478,
"step": 69000
},
{
"epoch": 7.439353099730458,
"eval_accuracy": 0.3885711938811909,
"eval_loss": 3.3474209308624268,
"eval_runtime": 184.5137,
"eval_samples_per_second": 97.613,
"eval_steps_per_second": 6.103,
"step": 69000
},
{
"epoch": 7.444743935309973,
"grad_norm": 0.7947202324867249,
"learning_rate": 0.00015366216945493792,
"loss": 3.2268,
"step": 69050
},
{
"epoch": 7.450134770889488,
"grad_norm": 0.8274387717247009,
"learning_rate": 0.00015333837021046947,
"loss": 3.2182,
"step": 69100
},
{
"epoch": 7.455525606469003,
"grad_norm": 0.8264995813369751,
"learning_rate": 0.00015301457096600105,
"loss": 3.2015,
"step": 69150
},
{
"epoch": 7.460916442048518,
"grad_norm": 0.8203253149986267,
"learning_rate": 0.00015269077172153263,
"loss": 3.2442,
"step": 69200
},
{
"epoch": 7.466307277628032,
"grad_norm": 0.795678436756134,
"learning_rate": 0.0001523669724770642,
"loss": 3.2306,
"step": 69250
},
{
"epoch": 7.471698113207547,
"grad_norm": 0.809211790561676,
"learning_rate": 0.00015204317323259578,
"loss": 3.2203,
"step": 69300
},
{
"epoch": 7.4770889487870615,
"grad_norm": 0.8124308586120605,
"learning_rate": 0.00015171937398812736,
"loss": 3.2398,
"step": 69350
},
{
"epoch": 7.482479784366577,
"grad_norm": 0.8006120324134827,
"learning_rate": 0.00015139557474365894,
"loss": 3.2411,
"step": 69400
},
{
"epoch": 7.487870619946092,
"grad_norm": 0.8229328393936157,
"learning_rate": 0.00015107177549919046,
"loss": 3.2352,
"step": 69450
},
{
"epoch": 7.493261455525606,
"grad_norm": 0.8513359427452087,
"learning_rate": 0.00015074797625472204,
"loss": 3.2228,
"step": 69500
},
{
"epoch": 7.498652291105121,
"grad_norm": 0.7921926975250244,
"learning_rate": 0.00015042417701025362,
"loss": 3.228,
"step": 69550
},
{
"epoch": 7.504043126684636,
"grad_norm": 0.8342979550361633,
"learning_rate": 0.0001501003777657852,
"loss": 3.2229,
"step": 69600
},
{
"epoch": 7.509433962264151,
"grad_norm": 0.8148936033248901,
"learning_rate": 0.00014977657852131677,
"loss": 3.2278,
"step": 69650
},
{
"epoch": 7.514824797843666,
"grad_norm": 0.7805774211883545,
"learning_rate": 0.00014945277927684835,
"loss": 3.2389,
"step": 69700
},
{
"epoch": 7.520215633423181,
"grad_norm": 0.8420718908309937,
"learning_rate": 0.00014912898003237993,
"loss": 3.2292,
"step": 69750
},
{
"epoch": 7.525606469002695,
"grad_norm": 0.8308120965957642,
"learning_rate": 0.00014880518078791148,
"loss": 3.2435,
"step": 69800
},
{
"epoch": 7.53099730458221,
"grad_norm": 0.8225760459899902,
"learning_rate": 0.00014848138154344306,
"loss": 3.2188,
"step": 69850
},
{
"epoch": 7.536388140161725,
"grad_norm": 0.836409330368042,
"learning_rate": 0.00014815758229897463,
"loss": 3.2279,
"step": 69900
},
{
"epoch": 7.54177897574124,
"grad_norm": 0.8201701641082764,
"learning_rate": 0.0001478337830545062,
"loss": 3.2341,
"step": 69950
},
{
"epoch": 7.547169811320755,
"grad_norm": 0.8581951260566711,
"learning_rate": 0.00014750998381003776,
"loss": 3.2322,
"step": 70000
},
{
"epoch": 7.547169811320755,
"eval_accuracy": 0.38909055468494413,
"eval_loss": 3.344033718109131,
"eval_runtime": 184.6782,
"eval_samples_per_second": 97.526,
"eval_steps_per_second": 6.097,
"step": 70000
},
{
"epoch": 7.55256064690027,
"grad_norm": 0.8392597436904907,
"learning_rate": 0.00014718618456556934,
"loss": 3.2139,
"step": 70050
},
{
"epoch": 7.557951482479784,
"grad_norm": 0.7937889099121094,
"learning_rate": 0.00014686238532110092,
"loss": 3.2333,
"step": 70100
},
{
"epoch": 7.563342318059299,
"grad_norm": 0.8366170525550842,
"learning_rate": 0.00014653858607663247,
"loss": 3.2329,
"step": 70150
},
{
"epoch": 7.568733153638814,
"grad_norm": 0.7545600533485413,
"learning_rate": 0.00014621478683216405,
"loss": 3.2324,
"step": 70200
},
{
"epoch": 7.574123989218329,
"grad_norm": 0.8192431330680847,
"learning_rate": 0.00014589098758769562,
"loss": 3.2286,
"step": 70250
},
{
"epoch": 7.579514824797844,
"grad_norm": 0.8045336604118347,
"learning_rate": 0.00014556718834322717,
"loss": 3.2416,
"step": 70300
},
{
"epoch": 7.584905660377358,
"grad_norm": 0.8145738244056702,
"learning_rate": 0.00014524338909875875,
"loss": 3.2155,
"step": 70350
},
{
"epoch": 7.590296495956873,
"grad_norm": 0.8092181086540222,
"learning_rate": 0.00014491958985429033,
"loss": 3.2202,
"step": 70400
},
{
"epoch": 7.595687331536388,
"grad_norm": 0.8141239285469055,
"learning_rate": 0.00014459579060982188,
"loss": 3.2258,
"step": 70450
},
{
"epoch": 7.601078167115903,
"grad_norm": 0.7778471112251282,
"learning_rate": 0.00014427199136535346,
"loss": 3.2405,
"step": 70500
},
{
"epoch": 7.606469002695418,
"grad_norm": 0.8479476571083069,
"learning_rate": 0.00014394819212088504,
"loss": 3.2268,
"step": 70550
},
{
"epoch": 7.611859838274933,
"grad_norm": 0.8255435824394226,
"learning_rate": 0.0001436243928764166,
"loss": 3.2523,
"step": 70600
},
{
"epoch": 7.617250673854447,
"grad_norm": 0.8285021185874939,
"learning_rate": 0.00014330059363194816,
"loss": 3.2265,
"step": 70650
},
{
"epoch": 7.622641509433962,
"grad_norm": 0.7851364016532898,
"learning_rate": 0.00014297679438747974,
"loss": 3.2288,
"step": 70700
},
{
"epoch": 7.628032345013477,
"grad_norm": 0.8818436861038208,
"learning_rate": 0.00014265299514301132,
"loss": 3.2209,
"step": 70750
},
{
"epoch": 7.633423180592992,
"grad_norm": 0.7907940745353699,
"learning_rate": 0.0001423291958985429,
"loss": 3.2244,
"step": 70800
},
{
"epoch": 7.638814016172507,
"grad_norm": 0.8718692660331726,
"learning_rate": 0.00014200539665407445,
"loss": 3.2308,
"step": 70850
},
{
"epoch": 7.644204851752022,
"grad_norm": 0.8215486407279968,
"learning_rate": 0.00014168159740960602,
"loss": 3.2262,
"step": 70900
},
{
"epoch": 7.649595687331536,
"grad_norm": 0.8401033282279968,
"learning_rate": 0.0001413577981651376,
"loss": 3.2339,
"step": 70950
},
{
"epoch": 7.654986522911051,
"grad_norm": 0.8204228281974792,
"learning_rate": 0.00014103399892066918,
"loss": 3.2343,
"step": 71000
},
{
"epoch": 7.654986522911051,
"eval_accuracy": 0.3894650811892406,
"eval_loss": 3.3388493061065674,
"eval_runtime": 185.7456,
"eval_samples_per_second": 96.966,
"eval_steps_per_second": 6.062,
"step": 71000
},
{
"epoch": 7.660377358490566,
"grad_norm": 0.8586466908454895,
"learning_rate": 0.0001407166756610901,
"loss": 3.2354,
"step": 71050
},
{
"epoch": 7.665768194070081,
"grad_norm": 0.8508881330490112,
"learning_rate": 0.00014039287641662168,
"loss": 3.2176,
"step": 71100
},
{
"epoch": 7.671159029649596,
"grad_norm": 0.8303574323654175,
"learning_rate": 0.00014006907717215326,
"loss": 3.2118,
"step": 71150
},
{
"epoch": 7.67654986522911,
"grad_norm": 0.8184866905212402,
"learning_rate": 0.0001397452779276848,
"loss": 3.2176,
"step": 71200
},
{
"epoch": 7.681940700808625,
"grad_norm": 0.8103206157684326,
"learning_rate": 0.0001394214786832164,
"loss": 3.2255,
"step": 71250
},
{
"epoch": 7.6873315363881405,
"grad_norm": 0.7950066924095154,
"learning_rate": 0.00013909767943874797,
"loss": 3.2498,
"step": 71300
},
{
"epoch": 7.692722371967655,
"grad_norm": 0.7989579439163208,
"learning_rate": 0.00013877388019427954,
"loss": 3.2394,
"step": 71350
},
{
"epoch": 7.69811320754717,
"grad_norm": 0.8339649438858032,
"learning_rate": 0.0001384500809498111,
"loss": 3.2333,
"step": 71400
},
{
"epoch": 7.703504043126685,
"grad_norm": 0.8148152232170105,
"learning_rate": 0.00013812628170534267,
"loss": 3.2234,
"step": 71450
},
{
"epoch": 7.708894878706199,
"grad_norm": 0.8153486251831055,
"learning_rate": 0.00013780248246087425,
"loss": 3.226,
"step": 71500
},
{
"epoch": 7.714285714285714,
"grad_norm": 0.8060972094535828,
"learning_rate": 0.00013747868321640583,
"loss": 3.2524,
"step": 71550
},
{
"epoch": 7.719676549865229,
"grad_norm": 0.8012336492538452,
"learning_rate": 0.00013715488397193738,
"loss": 3.2405,
"step": 71600
},
{
"epoch": 7.725067385444744,
"grad_norm": 0.8267596364021301,
"learning_rate": 0.00013683108472746896,
"loss": 3.2472,
"step": 71650
},
{
"epoch": 7.730458221024259,
"grad_norm": 0.8347733616828918,
"learning_rate": 0.00013650728548300053,
"loss": 3.2445,
"step": 71700
},
{
"epoch": 7.735849056603773,
"grad_norm": 0.8193061351776123,
"learning_rate": 0.0001361834862385321,
"loss": 3.2463,
"step": 71750
},
{
"epoch": 7.741239892183288,
"grad_norm": 0.8073017597198486,
"learning_rate": 0.0001358596869940637,
"loss": 3.2403,
"step": 71800
},
{
"epoch": 7.7466307277628035,
"grad_norm": 0.8158572316169739,
"learning_rate": 0.00013553588774959524,
"loss": 3.253,
"step": 71850
},
{
"epoch": 7.752021563342318,
"grad_norm": 0.8269498348236084,
"learning_rate": 0.00013521208850512682,
"loss": 3.2323,
"step": 71900
},
{
"epoch": 7.757412398921833,
"grad_norm": 0.8673365712165833,
"learning_rate": 0.0001348882892606584,
"loss": 3.263,
"step": 71950
},
{
"epoch": 7.762803234501348,
"grad_norm": 0.8098892569541931,
"learning_rate": 0.00013456449001618995,
"loss": 3.2298,
"step": 72000
},
{
"epoch": 7.762803234501348,
"eval_accuracy": 0.38985992578355844,
"eval_loss": 3.3367950916290283,
"eval_runtime": 184.6595,
"eval_samples_per_second": 97.536,
"eval_steps_per_second": 6.098,
"step": 72000
},
{
"epoch": 7.768194070080862,
"grad_norm": 0.7425093054771423,
"learning_rate": 0.00013424069077172152,
"loss": 3.2291,
"step": 72050
},
{
"epoch": 7.773584905660377,
"grad_norm": 0.8305625915527344,
"learning_rate": 0.0001339168915272531,
"loss": 3.2171,
"step": 72100
},
{
"epoch": 7.7789757412398925,
"grad_norm": 0.8112491965293884,
"learning_rate": 0.00013359309228278465,
"loss": 3.2338,
"step": 72150
},
{
"epoch": 7.784366576819407,
"grad_norm": 0.8389681577682495,
"learning_rate": 0.00013326929303831623,
"loss": 3.2333,
"step": 72200
},
{
"epoch": 7.789757412398922,
"grad_norm": 0.8382710218429565,
"learning_rate": 0.0001329454937938478,
"loss": 3.2317,
"step": 72250
},
{
"epoch": 7.795148247978437,
"grad_norm": 0.8545230627059937,
"learning_rate": 0.00013262169454937936,
"loss": 3.2285,
"step": 72300
},
{
"epoch": 7.800539083557951,
"grad_norm": 0.8553089499473572,
"learning_rate": 0.00013229789530491093,
"loss": 3.2316,
"step": 72350
},
{
"epoch": 7.8059299191374665,
"grad_norm": 0.8017550110816956,
"learning_rate": 0.0001319740960604425,
"loss": 3.2338,
"step": 72400
},
{
"epoch": 7.811320754716981,
"grad_norm": 0.8867925405502319,
"learning_rate": 0.0001316502968159741,
"loss": 3.2265,
"step": 72450
},
{
"epoch": 7.816711590296496,
"grad_norm": 0.8136430978775024,
"learning_rate": 0.00013132649757150564,
"loss": 3.2368,
"step": 72500
},
{
"epoch": 7.822102425876011,
"grad_norm": 0.8786584734916687,
"learning_rate": 0.00013100269832703722,
"loss": 3.2299,
"step": 72550
},
{
"epoch": 7.827493261455525,
"grad_norm": 0.8976890444755554,
"learning_rate": 0.0001306788990825688,
"loss": 3.2413,
"step": 72600
},
{
"epoch": 7.83288409703504,
"grad_norm": 0.790888249874115,
"learning_rate": 0.00013035509983810037,
"loss": 3.2466,
"step": 72650
},
{
"epoch": 7.8382749326145555,
"grad_norm": 0.8226049542427063,
"learning_rate": 0.00013003130059363192,
"loss": 3.2425,
"step": 72700
},
{
"epoch": 7.84366576819407,
"grad_norm": 0.8924099802970886,
"learning_rate": 0.0001297075013491635,
"loss": 3.2371,
"step": 72750
},
{
"epoch": 7.849056603773585,
"grad_norm": 0.8906298875808716,
"learning_rate": 0.00012938370210469508,
"loss": 3.2601,
"step": 72800
},
{
"epoch": 7.8544474393531,
"grad_norm": 0.8076866865158081,
"learning_rate": 0.00012905990286022666,
"loss": 3.2299,
"step": 72850
},
{
"epoch": 7.859838274932614,
"grad_norm": 0.8295146226882935,
"learning_rate": 0.0001287361036157582,
"loss": 3.2341,
"step": 72900
},
{
"epoch": 7.8652291105121295,
"grad_norm": 0.817703366279602,
"learning_rate": 0.00012841230437128979,
"loss": 3.2555,
"step": 72950
},
{
"epoch": 7.870619946091644,
"grad_norm": 0.8261833786964417,
"learning_rate": 0.00012808850512682136,
"loss": 3.2291,
"step": 73000
},
{
"epoch": 7.870619946091644,
"eval_accuracy": 0.39006310668377153,
"eval_loss": 3.3327910900115967,
"eval_runtime": 184.5537,
"eval_samples_per_second": 97.592,
"eval_steps_per_second": 6.101,
"step": 73000
},
{
"epoch": 7.876010781671159,
"grad_norm": 0.8849043250083923,
"learning_rate": 0.00012776470588235294,
"loss": 3.2275,
"step": 73050
},
{
"epoch": 7.881401617250674,
"grad_norm": 0.833169162273407,
"learning_rate": 0.0001274409066378845,
"loss": 3.2389,
"step": 73100
},
{
"epoch": 7.886792452830189,
"grad_norm": 0.8109320998191833,
"learning_rate": 0.00012711710739341607,
"loss": 3.2435,
"step": 73150
},
{
"epoch": 7.892183288409703,
"grad_norm": 0.8014684319496155,
"learning_rate": 0.00012679330814894765,
"loss": 3.2212,
"step": 73200
},
{
"epoch": 7.8975741239892185,
"grad_norm": 0.8155044913291931,
"learning_rate": 0.00012646950890447922,
"loss": 3.2374,
"step": 73250
},
{
"epoch": 7.902964959568733,
"grad_norm": 0.849024772644043,
"learning_rate": 0.00012614570966001077,
"loss": 3.2389,
"step": 73300
},
{
"epoch": 7.908355795148248,
"grad_norm": 0.8464096188545227,
"learning_rate": 0.00012582191041554235,
"loss": 3.2406,
"step": 73350
},
{
"epoch": 7.913746630727763,
"grad_norm": 0.8928974270820618,
"learning_rate": 0.00012549811117107393,
"loss": 3.2471,
"step": 73400
},
{
"epoch": 7.919137466307277,
"grad_norm": 0.8113333582878113,
"learning_rate": 0.0001251743119266055,
"loss": 3.226,
"step": 73450
},
{
"epoch": 7.9245283018867925,
"grad_norm": 0.8421093821525574,
"learning_rate": 0.00012485051268213706,
"loss": 3.2475,
"step": 73500
},
{
"epoch": 7.929919137466308,
"grad_norm": 0.7939735651016235,
"learning_rate": 0.00012452671343766864,
"loss": 3.2317,
"step": 73550
},
{
"epoch": 7.935309973045822,
"grad_norm": 0.8566156625747681,
"learning_rate": 0.00012420291419320021,
"loss": 3.2422,
"step": 73600
},
{
"epoch": 7.940700808625337,
"grad_norm": 0.8841614723205566,
"learning_rate": 0.0001238791149487318,
"loss": 3.2211,
"step": 73650
},
{
"epoch": 7.946091644204852,
"grad_norm": 0.8392506241798401,
"learning_rate": 0.00012355531570426334,
"loss": 3.2267,
"step": 73700
},
{
"epoch": 7.951482479784366,
"grad_norm": 0.8039172291755676,
"learning_rate": 0.00012323151645979492,
"loss": 3.2181,
"step": 73750
},
{
"epoch": 7.9568733153638815,
"grad_norm": 0.8260869979858398,
"learning_rate": 0.0001229077172153265,
"loss": 3.1977,
"step": 73800
},
{
"epoch": 7.962264150943396,
"grad_norm": 0.7950641512870789,
"learning_rate": 0.00012258391797085805,
"loss": 3.2227,
"step": 73850
},
{
"epoch": 7.967654986522911,
"grad_norm": 0.8203497529029846,
"learning_rate": 0.00012226011872638963,
"loss": 3.2304,
"step": 73900
},
{
"epoch": 7.973045822102426,
"grad_norm": 0.8410260081291199,
"learning_rate": 0.00012193631948192119,
"loss": 3.253,
"step": 73950
},
{
"epoch": 7.97843665768194,
"grad_norm": 0.8598277568817139,
"learning_rate": 0.00012161252023745277,
"loss": 3.2438,
"step": 74000
},
{
"epoch": 7.97843665768194,
"eval_accuracy": 0.39048044242586694,
"eval_loss": 3.329894781112671,
"eval_runtime": 184.6904,
"eval_samples_per_second": 97.52,
"eval_steps_per_second": 6.097,
"step": 74000
},
{
"epoch": 7.9838274932614555,
"grad_norm": 0.8428695797920227,
"learning_rate": 0.00012128872099298435,
"loss": 3.2372,
"step": 74050
},
{
"epoch": 7.989218328840971,
"grad_norm": 0.8146273493766785,
"learning_rate": 0.00012096492174851591,
"loss": 3.2281,
"step": 74100
},
{
"epoch": 7.994609164420485,
"grad_norm": 0.8260930776596069,
"learning_rate": 0.00012064112250404749,
"loss": 3.2201,
"step": 74150
},
{
"epoch": 8.0,
"grad_norm": 1.8513790369033813,
"learning_rate": 0.00012031732325957905,
"loss": 3.2151,
"step": 74200
},
{
"epoch": 8.005390835579515,
"grad_norm": 0.8102459907531738,
"learning_rate": 0.00011999352401511062,
"loss": 3.1616,
"step": 74250
},
{
"epoch": 8.01078167115903,
"grad_norm": 0.8537736535072327,
"learning_rate": 0.00011966972477064219,
"loss": 3.1582,
"step": 74300
},
{
"epoch": 8.016172506738544,
"grad_norm": 0.8426727652549744,
"learning_rate": 0.00011934592552617377,
"loss": 3.1461,
"step": 74350
},
{
"epoch": 8.021563342318059,
"grad_norm": 0.8298203945159912,
"learning_rate": 0.00011902212628170532,
"loss": 3.1569,
"step": 74400
},
{
"epoch": 8.026954177897574,
"grad_norm": 0.805033802986145,
"learning_rate": 0.0001186983270372369,
"loss": 3.1601,
"step": 74450
},
{
"epoch": 8.032345013477089,
"grad_norm": 0.8290202617645264,
"learning_rate": 0.00011837452779276848,
"loss": 3.1609,
"step": 74500
},
{
"epoch": 8.037735849056604,
"grad_norm": 0.8169941306114197,
"learning_rate": 0.00011805072854830005,
"loss": 3.1748,
"step": 74550
},
{
"epoch": 8.04312668463612,
"grad_norm": 0.8094680309295654,
"learning_rate": 0.0001177269293038316,
"loss": 3.1651,
"step": 74600
},
{
"epoch": 8.048517520215633,
"grad_norm": 0.865330159664154,
"learning_rate": 0.00011740313005936318,
"loss": 3.1628,
"step": 74650
},
{
"epoch": 8.053908355795148,
"grad_norm": 0.8339937925338745,
"learning_rate": 0.00011707933081489476,
"loss": 3.1506,
"step": 74700
},
{
"epoch": 8.059299191374663,
"grad_norm": 0.8235217332839966,
"learning_rate": 0.00011675553157042632,
"loss": 3.1682,
"step": 74750
},
{
"epoch": 8.064690026954178,
"grad_norm": 0.8427267074584961,
"learning_rate": 0.00011643173232595789,
"loss": 3.1618,
"step": 74800
},
{
"epoch": 8.070080862533693,
"grad_norm": 0.8886874914169312,
"learning_rate": 0.00011610793308148947,
"loss": 3.1626,
"step": 74850
},
{
"epoch": 8.075471698113208,
"grad_norm": 0.8250788450241089,
"learning_rate": 0.00011578413383702104,
"loss": 3.1735,
"step": 74900
},
{
"epoch": 8.080862533692722,
"grad_norm": 0.8339901566505432,
"learning_rate": 0.00011546033459255261,
"loss": 3.1546,
"step": 74950
},
{
"epoch": 8.086253369272237,
"grad_norm": 0.8584937453269958,
"learning_rate": 0.00011513653534808419,
"loss": 3.1579,
"step": 75000
},
{
"epoch": 8.086253369272237,
"eval_accuracy": 0.3902442310477583,
"eval_loss": 3.3350722789764404,
"eval_runtime": 184.4097,
"eval_samples_per_second": 97.668,
"eval_steps_per_second": 6.106,
"step": 75000
},
{
"epoch": 8.091644204851752,
"grad_norm": 0.8601799607276917,
"learning_rate": 0.00011481921208850512,
"loss": 3.1659,
"step": 75050
},
{
"epoch": 8.097035040431267,
"grad_norm": 0.8152199983596802,
"learning_rate": 0.00011449541284403669,
"loss": 3.1552,
"step": 75100
},
{
"epoch": 8.102425876010782,
"grad_norm": 0.848799467086792,
"learning_rate": 0.00011417161359956825,
"loss": 3.1689,
"step": 75150
},
{
"epoch": 8.107816711590296,
"grad_norm": 0.8392152190208435,
"learning_rate": 0.00011384781435509983,
"loss": 3.1823,
"step": 75200
},
{
"epoch": 8.11320754716981,
"grad_norm": 0.8672192692756653,
"learning_rate": 0.0001135240151106314,
"loss": 3.1644,
"step": 75250
},
{
"epoch": 8.118598382749326,
"grad_norm": 0.7989856004714966,
"learning_rate": 0.00011320021586616297,
"loss": 3.1768,
"step": 75300
},
{
"epoch": 8.123989218328841,
"grad_norm": 0.8558405041694641,
"learning_rate": 0.00011287641662169454,
"loss": 3.1764,
"step": 75350
},
{
"epoch": 8.129380053908356,
"grad_norm": 0.8172704577445984,
"learning_rate": 0.0001125526173772261,
"loss": 3.1816,
"step": 75400
},
{
"epoch": 8.134770889487871,
"grad_norm": 0.8455656170845032,
"learning_rate": 0.00011222881813275768,
"loss": 3.1688,
"step": 75450
},
{
"epoch": 8.140161725067385,
"grad_norm": 0.8776059746742249,
"learning_rate": 0.00011190501888828925,
"loss": 3.1691,
"step": 75500
},
{
"epoch": 8.1455525606469,
"grad_norm": 0.8391590714454651,
"learning_rate": 0.00011158121964382083,
"loss": 3.1731,
"step": 75550
},
{
"epoch": 8.150943396226415,
"grad_norm": 0.8365171551704407,
"learning_rate": 0.00011125742039935238,
"loss": 3.1767,
"step": 75600
},
{
"epoch": 8.15633423180593,
"grad_norm": 0.7903484106063843,
"learning_rate": 0.00011093362115488396,
"loss": 3.1633,
"step": 75650
},
{
"epoch": 8.161725067385445,
"grad_norm": 0.8488314151763916,
"learning_rate": 0.00011060982191041554,
"loss": 3.1749,
"step": 75700
},
{
"epoch": 8.167115902964959,
"grad_norm": 0.8512059450149536,
"learning_rate": 0.00011028602266594712,
"loss": 3.1723,
"step": 75750
},
{
"epoch": 8.172506738544474,
"grad_norm": 0.8906243443489075,
"learning_rate": 0.00010996222342147867,
"loss": 3.1905,
"step": 75800
},
{
"epoch": 8.177897574123989,
"grad_norm": 0.8632593154907227,
"learning_rate": 0.00010963842417701024,
"loss": 3.1958,
"step": 75850
},
{
"epoch": 8.183288409703504,
"grad_norm": 0.88079833984375,
"learning_rate": 0.00010931462493254182,
"loss": 3.1587,
"step": 75900
},
{
"epoch": 8.18867924528302,
"grad_norm": 0.8311557173728943,
"learning_rate": 0.00010899082568807339,
"loss": 3.1802,
"step": 75950
},
{
"epoch": 8.194070080862534,
"grad_norm": 0.8374923467636108,
"learning_rate": 0.00010866702644360495,
"loss": 3.1844,
"step": 76000
},
{
"epoch": 8.194070080862534,
"eval_accuracy": 0.39045719070787466,
"eval_loss": 3.3339478969573975,
"eval_runtime": 184.5077,
"eval_samples_per_second": 97.617,
"eval_steps_per_second": 6.103,
"step": 76000
},
{
"epoch": 8.199460916442048,
"grad_norm": 0.8399762511253357,
"learning_rate": 0.00010834322719913653,
"loss": 3.1794,
"step": 76050
},
{
"epoch": 8.204851752021563,
"grad_norm": 0.8227581977844238,
"learning_rate": 0.00010801942795466809,
"loss": 3.154,
"step": 76100
},
{
"epoch": 8.210242587601078,
"grad_norm": 0.8603895902633667,
"learning_rate": 0.00010769562871019967,
"loss": 3.1865,
"step": 76150
},
{
"epoch": 8.215633423180593,
"grad_norm": 0.8445917367935181,
"learning_rate": 0.00010737182946573123,
"loss": 3.1657,
"step": 76200
},
{
"epoch": 8.221024258760108,
"grad_norm": 0.8297895789146423,
"learning_rate": 0.0001070480302212628,
"loss": 3.1629,
"step": 76250
},
{
"epoch": 8.226415094339623,
"grad_norm": 0.8218351006507874,
"learning_rate": 0.00010672423097679438,
"loss": 3.1786,
"step": 76300
},
{
"epoch": 8.231805929919137,
"grad_norm": 0.8612483143806458,
"learning_rate": 0.00010640043173232595,
"loss": 3.1705,
"step": 76350
},
{
"epoch": 8.237196765498652,
"grad_norm": 0.8282567262649536,
"learning_rate": 0.00010607663248785753,
"loss": 3.164,
"step": 76400
},
{
"epoch": 8.242587601078167,
"grad_norm": 0.8315432667732239,
"learning_rate": 0.00010575283324338908,
"loss": 3.1871,
"step": 76450
},
{
"epoch": 8.247978436657682,
"grad_norm": 0.8825376033782959,
"learning_rate": 0.00010542903399892066,
"loss": 3.1884,
"step": 76500
},
{
"epoch": 8.253369272237197,
"grad_norm": 0.8460707068443298,
"learning_rate": 0.00010510523475445224,
"loss": 3.1919,
"step": 76550
},
{
"epoch": 8.25876010781671,
"grad_norm": 0.8482176065444946,
"learning_rate": 0.0001047814355099838,
"loss": 3.1785,
"step": 76600
},
{
"epoch": 8.264150943396226,
"grad_norm": 0.8487524390220642,
"learning_rate": 0.00010445763626551537,
"loss": 3.1706,
"step": 76650
},
{
"epoch": 8.269541778975741,
"grad_norm": 0.8230019211769104,
"learning_rate": 0.00010413383702104694,
"loss": 3.1926,
"step": 76700
},
{
"epoch": 8.274932614555256,
"grad_norm": 0.8345533013343811,
"learning_rate": 0.00010381003777657852,
"loss": 3.1799,
"step": 76750
},
{
"epoch": 8.280323450134771,
"grad_norm": 0.8185267448425293,
"learning_rate": 0.00010348623853211008,
"loss": 3.1743,
"step": 76800
},
{
"epoch": 8.285714285714286,
"grad_norm": 0.7892093658447266,
"learning_rate": 0.00010316243928764165,
"loss": 3.1779,
"step": 76850
},
{
"epoch": 8.2911051212938,
"grad_norm": 0.8616093397140503,
"learning_rate": 0.0001028451160280626,
"loss": 3.1912,
"step": 76900
},
{
"epoch": 8.296495956873315,
"grad_norm": 0.8338601589202881,
"learning_rate": 0.00010252131678359416,
"loss": 3.1753,
"step": 76950
},
{
"epoch": 8.30188679245283,
"grad_norm": 0.8935291171073914,
"learning_rate": 0.00010219751753912573,
"loss": 3.1734,
"step": 77000
},
{
"epoch": 8.30188679245283,
"eval_accuracy": 0.39085735929369536,
"eval_loss": 3.3295018672943115,
"eval_runtime": 184.5477,
"eval_samples_per_second": 97.595,
"eval_steps_per_second": 6.101,
"step": 77000
},
{
"epoch": 8.307277628032345,
"grad_norm": 0.8242422342300415,
"learning_rate": 0.0001018737182946573,
"loss": 3.1815,
"step": 77050
},
{
"epoch": 8.31266846361186,
"grad_norm": 0.8670544028282166,
"learning_rate": 0.00010154991905018887,
"loss": 3.1878,
"step": 77100
},
{
"epoch": 8.318059299191376,
"grad_norm": 0.8715893626213074,
"learning_rate": 0.00010122611980572045,
"loss": 3.174,
"step": 77150
},
{
"epoch": 8.323450134770889,
"grad_norm": 0.8662042617797852,
"learning_rate": 0.00010090232056125201,
"loss": 3.1696,
"step": 77200
},
{
"epoch": 8.328840970350404,
"grad_norm": 0.8212565779685974,
"learning_rate": 0.00010057852131678359,
"loss": 3.1729,
"step": 77250
},
{
"epoch": 8.33423180592992,
"grad_norm": 0.8568588495254517,
"learning_rate": 0.00010025472207231515,
"loss": 3.1769,
"step": 77300
},
{
"epoch": 8.339622641509434,
"grad_norm": 0.8369387984275818,
"learning_rate": 9.993092282784673e-05,
"loss": 3.1672,
"step": 77350
},
{
"epoch": 8.34501347708895,
"grad_norm": 0.8870252966880798,
"learning_rate": 9.96071235833783e-05,
"loss": 3.2012,
"step": 77400
},
{
"epoch": 8.350404312668463,
"grad_norm": 0.8651405572891235,
"learning_rate": 9.928332433890986e-05,
"loss": 3.1769,
"step": 77450
},
{
"epoch": 8.355795148247978,
"grad_norm": 0.8473480343818665,
"learning_rate": 9.895952509444144e-05,
"loss": 3.1686,
"step": 77500
},
{
"epoch": 8.361185983827493,
"grad_norm": 0.8660786747932434,
"learning_rate": 9.863572584997302e-05,
"loss": 3.1906,
"step": 77550
},
{
"epoch": 8.366576819407008,
"grad_norm": 0.8497851490974426,
"learning_rate": 9.831192660550457e-05,
"loss": 3.1817,
"step": 77600
},
{
"epoch": 8.371967654986523,
"grad_norm": 0.8756412863731384,
"learning_rate": 9.798812736103614e-05,
"loss": 3.1796,
"step": 77650
},
{
"epoch": 8.377358490566039,
"grad_norm": 0.8684273958206177,
"learning_rate": 9.766432811656772e-05,
"loss": 3.1827,
"step": 77700
},
{
"epoch": 8.382749326145552,
"grad_norm": 0.8414525389671326,
"learning_rate": 9.73405288720993e-05,
"loss": 3.1747,
"step": 77750
},
{
"epoch": 8.388140161725067,
"grad_norm": 0.8159910440444946,
"learning_rate": 9.701672962763086e-05,
"loss": 3.1819,
"step": 77800
},
{
"epoch": 8.393530997304582,
"grad_norm": 0.8277149796485901,
"learning_rate": 9.669293038316243e-05,
"loss": 3.1675,
"step": 77850
},
{
"epoch": 8.398921832884097,
"grad_norm": 0.8163325190544128,
"learning_rate": 9.6369131138694e-05,
"loss": 3.1747,
"step": 77900
},
{
"epoch": 8.404312668463612,
"grad_norm": 0.8605278134346008,
"learning_rate": 9.604533189422557e-05,
"loss": 3.1804,
"step": 77950
},
{
"epoch": 8.409703504043126,
"grad_norm": 0.8365135788917542,
"learning_rate": 9.572153264975715e-05,
"loss": 3.1913,
"step": 78000
},
{
"epoch": 8.409703504043126,
"eval_accuracy": 0.39124416357431496,
"eval_loss": 3.3270325660705566,
"eval_runtime": 184.4922,
"eval_samples_per_second": 97.625,
"eval_steps_per_second": 6.103,
"step": 78000
},
{
"epoch": 8.415094339622641,
"grad_norm": 0.8370071053504944,
"learning_rate": 9.539773340528871e-05,
"loss": 3.1695,
"step": 78050
},
{
"epoch": 8.420485175202156,
"grad_norm": 0.8824688792228699,
"learning_rate": 9.507393416082027e-05,
"loss": 3.1653,
"step": 78100
},
{
"epoch": 8.425876010781671,
"grad_norm": 0.9018416404724121,
"learning_rate": 9.475013491635185e-05,
"loss": 3.1888,
"step": 78150
},
{
"epoch": 8.431266846361186,
"grad_norm": 0.8426613211631775,
"learning_rate": 9.442633567188343e-05,
"loss": 3.1987,
"step": 78200
},
{
"epoch": 8.436657681940702,
"grad_norm": 0.8415345549583435,
"learning_rate": 9.410253642741498e-05,
"loss": 3.1751,
"step": 78250
},
{
"epoch": 8.442048517520215,
"grad_norm": 0.8651146292686462,
"learning_rate": 9.377873718294656e-05,
"loss": 3.1689,
"step": 78300
},
{
"epoch": 8.44743935309973,
"grad_norm": 0.8710966110229492,
"learning_rate": 9.345493793847814e-05,
"loss": 3.1724,
"step": 78350
},
{
"epoch": 8.452830188679245,
"grad_norm": 0.8346008658409119,
"learning_rate": 9.313113869400971e-05,
"loss": 3.1675,
"step": 78400
},
{
"epoch": 8.45822102425876,
"grad_norm": 0.8859044909477234,
"learning_rate": 9.280733944954126e-05,
"loss": 3.1776,
"step": 78450
},
{
"epoch": 8.463611859838275,
"grad_norm": 0.8346058130264282,
"learning_rate": 9.248354020507284e-05,
"loss": 3.1833,
"step": 78500
},
{
"epoch": 8.46900269541779,
"grad_norm": 0.8823174238204956,
"learning_rate": 9.215974096060442e-05,
"loss": 3.177,
"step": 78550
},
{
"epoch": 8.474393530997304,
"grad_norm": 0.8848249316215515,
"learning_rate": 9.1835941716136e-05,
"loss": 3.1878,
"step": 78600
},
{
"epoch": 8.479784366576819,
"grad_norm": 0.8461087942123413,
"learning_rate": 9.151214247166756e-05,
"loss": 3.1718,
"step": 78650
},
{
"epoch": 8.485175202156334,
"grad_norm": 0.817385733127594,
"learning_rate": 9.118834322719913e-05,
"loss": 3.1866,
"step": 78700
},
{
"epoch": 8.49056603773585,
"grad_norm": 0.8841643929481506,
"learning_rate": 9.08645439827307e-05,
"loss": 3.1854,
"step": 78750
},
{
"epoch": 8.495956873315365,
"grad_norm": 0.9169108271598816,
"learning_rate": 9.054074473826227e-05,
"loss": 3.1966,
"step": 78800
},
{
"epoch": 8.501347708894878,
"grad_norm": 0.8336373567581177,
"learning_rate": 9.021694549379385e-05,
"loss": 3.1685,
"step": 78850
},
{
"epoch": 8.506738544474393,
"grad_norm": 0.8359789848327637,
"learning_rate": 8.989314624932541e-05,
"loss": 3.179,
"step": 78900
},
{
"epoch": 8.512129380053908,
"grad_norm": 0.7882795929908752,
"learning_rate": 8.956934700485697e-05,
"loss": 3.1775,
"step": 78950
},
{
"epoch": 8.517520215633423,
"grad_norm": 0.8604826927185059,
"learning_rate": 8.924554776038855e-05,
"loss": 3.1779,
"step": 79000
},
{
"epoch": 8.517520215633423,
"eval_accuracy": 0.39145495017667503,
"eval_loss": 3.323984146118164,
"eval_runtime": 184.7191,
"eval_samples_per_second": 97.505,
"eval_steps_per_second": 6.096,
"step": 79000
},
{
"epoch": 8.522911051212938,
"grad_norm": 0.9060187339782715,
"learning_rate": 8.892174851592013e-05,
"loss": 3.189,
"step": 79050
},
{
"epoch": 8.528301886792454,
"grad_norm": 0.8653152585029602,
"learning_rate": 8.859794927145168e-05,
"loss": 3.1848,
"step": 79100
},
{
"epoch": 8.533692722371967,
"grad_norm": 0.8778457045555115,
"learning_rate": 8.827415002698326e-05,
"loss": 3.1733,
"step": 79150
},
{
"epoch": 8.539083557951482,
"grad_norm": 0.8845580220222473,
"learning_rate": 8.795035078251483e-05,
"loss": 3.1868,
"step": 79200
},
{
"epoch": 8.544474393530997,
"grad_norm": 0.8744798302650452,
"learning_rate": 8.763302752293577e-05,
"loss": 3.182,
"step": 79250
},
{
"epoch": 8.549865229110512,
"grad_norm": 0.8833090662956238,
"learning_rate": 8.730922827846734e-05,
"loss": 3.1769,
"step": 79300
},
{
"epoch": 8.555256064690028,
"grad_norm": 0.9444142580032349,
"learning_rate": 8.698542903399891e-05,
"loss": 3.2023,
"step": 79350
},
{
"epoch": 8.560646900269543,
"grad_norm": 0.8909146785736084,
"learning_rate": 8.666162978953049e-05,
"loss": 3.1752,
"step": 79400
},
{
"epoch": 8.566037735849056,
"grad_norm": 0.8775643706321716,
"learning_rate": 8.633783054506204e-05,
"loss": 3.1936,
"step": 79450
},
{
"epoch": 8.571428571428571,
"grad_norm": 0.9317524433135986,
"learning_rate": 8.601403130059362e-05,
"loss": 3.171,
"step": 79500
},
{
"epoch": 8.576819407008086,
"grad_norm": 0.8345090746879578,
"learning_rate": 8.56902320561252e-05,
"loss": 3.1716,
"step": 79550
},
{
"epoch": 8.582210242587601,
"grad_norm": 0.8182111978530884,
"learning_rate": 8.536643281165678e-05,
"loss": 3.19,
"step": 79600
},
{
"epoch": 8.587601078167117,
"grad_norm": 0.9039532542228699,
"learning_rate": 8.504263356718833e-05,
"loss": 3.1712,
"step": 79650
},
{
"epoch": 8.59299191374663,
"grad_norm": 0.9555698037147522,
"learning_rate": 8.47188343227199e-05,
"loss": 3.1764,
"step": 79700
},
{
"epoch": 8.598382749326145,
"grad_norm": 0.8400817513465881,
"learning_rate": 8.439503507825148e-05,
"loss": 3.1783,
"step": 79750
},
{
"epoch": 8.60377358490566,
"grad_norm": 0.8604862093925476,
"learning_rate": 8.407123583378305e-05,
"loss": 3.1765,
"step": 79800
},
{
"epoch": 8.609164420485175,
"grad_norm": 0.8044969439506531,
"learning_rate": 8.374743658931461e-05,
"loss": 3.175,
"step": 79850
},
{
"epoch": 8.61455525606469,
"grad_norm": 0.8684996366500854,
"learning_rate": 8.342363734484619e-05,
"loss": 3.1876,
"step": 79900
},
{
"epoch": 8.619946091644206,
"grad_norm": 0.8584052324295044,
"learning_rate": 8.309983810037775e-05,
"loss": 3.1926,
"step": 79950
},
{
"epoch": 8.625336927223719,
"grad_norm": 0.8365562558174133,
"learning_rate": 8.277603885590933e-05,
"loss": 3.1735,
"step": 80000
},
{
"epoch": 8.625336927223719,
"eval_accuracy": 0.39206242847244577,
"eval_loss": 3.3198676109313965,
"eval_runtime": 184.6958,
"eval_samples_per_second": 97.517,
"eval_steps_per_second": 6.097,
"step": 80000
},
{
"epoch": 8.630727762803234,
"grad_norm": 0.893174946308136,
"learning_rate": 8.245223961144091e-05,
"loss": 3.1723,
"step": 80050
},
{
"epoch": 8.63611859838275,
"grad_norm": 0.8763517141342163,
"learning_rate": 8.212844036697246e-05,
"loss": 3.1732,
"step": 80100
},
{
"epoch": 8.641509433962264,
"grad_norm": 0.8741143345832825,
"learning_rate": 8.180464112250404e-05,
"loss": 3.1947,
"step": 80150
},
{
"epoch": 8.64690026954178,
"grad_norm": 0.8090900182723999,
"learning_rate": 8.148084187803561e-05,
"loss": 3.1748,
"step": 80200
},
{
"epoch": 8.652291105121293,
"grad_norm": 0.8432245850563049,
"learning_rate": 8.115704263356719e-05,
"loss": 3.1976,
"step": 80250
},
{
"epoch": 8.657681940700808,
"grad_norm": 0.8647574186325073,
"learning_rate": 8.083324338909874e-05,
"loss": 3.182,
"step": 80300
},
{
"epoch": 8.663072776280323,
"grad_norm": 0.8686016798019409,
"learning_rate": 8.050944414463032e-05,
"loss": 3.1837,
"step": 80350
},
{
"epoch": 8.668463611859838,
"grad_norm": 0.8766531944274902,
"learning_rate": 8.01856449001619e-05,
"loss": 3.1785,
"step": 80400
},
{
"epoch": 8.673854447439354,
"grad_norm": 0.9033280611038208,
"learning_rate": 7.986184565569347e-05,
"loss": 3.1763,
"step": 80450
},
{
"epoch": 8.679245283018869,
"grad_norm": 0.8472071290016174,
"learning_rate": 7.953804641122503e-05,
"loss": 3.1753,
"step": 80500
},
{
"epoch": 8.684636118598382,
"grad_norm": 0.9259135723114014,
"learning_rate": 7.92142471667566e-05,
"loss": 3.1819,
"step": 80550
},
{
"epoch": 8.690026954177897,
"grad_norm": 0.8569573760032654,
"learning_rate": 7.889044792228818e-05,
"loss": 3.1753,
"step": 80600
},
{
"epoch": 8.695417789757412,
"grad_norm": 0.8650641441345215,
"learning_rate": 7.856664867781974e-05,
"loss": 3.1862,
"step": 80650
},
{
"epoch": 8.700808625336927,
"grad_norm": 0.9114779829978943,
"learning_rate": 7.824284943335132e-05,
"loss": 3.1889,
"step": 80700
},
{
"epoch": 8.706199460916443,
"grad_norm": 0.8367800712585449,
"learning_rate": 7.791905018888289e-05,
"loss": 3.1816,
"step": 80750
},
{
"epoch": 8.711590296495956,
"grad_norm": 0.860028088092804,
"learning_rate": 7.759525094441445e-05,
"loss": 3.1518,
"step": 80800
},
{
"epoch": 8.716981132075471,
"grad_norm": 0.8490385413169861,
"learning_rate": 7.727145169994603e-05,
"loss": 3.1831,
"step": 80850
},
{
"epoch": 8.722371967654986,
"grad_norm": 0.853510856628418,
"learning_rate": 7.69476524554776e-05,
"loss": 3.173,
"step": 80900
},
{
"epoch": 8.727762803234501,
"grad_norm": 0.8822104334831238,
"learning_rate": 7.662385321100916e-05,
"loss": 3.1716,
"step": 80950
},
{
"epoch": 8.733153638814017,
"grad_norm": 0.8734488487243652,
"learning_rate": 7.630005396654073e-05,
"loss": 3.1867,
"step": 81000
},
{
"epoch": 8.733153638814017,
"eval_accuracy": 0.39256581730169027,
"eval_loss": 3.315117597579956,
"eval_runtime": 184.2616,
"eval_samples_per_second": 97.747,
"eval_steps_per_second": 6.111,
"step": 81000
},
{
"epoch": 8.738544474393532,
"grad_norm": 0.8785738945007324,
"learning_rate": 7.597625472207231e-05,
"loss": 3.1972,
"step": 81050
},
{
"epoch": 8.743935309973045,
"grad_norm": 0.8223440647125244,
"learning_rate": 7.565245547760389e-05,
"loss": 3.197,
"step": 81100
},
{
"epoch": 8.74932614555256,
"grad_norm": 0.8490769267082214,
"learning_rate": 7.532865623313544e-05,
"loss": 3.1731,
"step": 81150
},
{
"epoch": 8.754716981132075,
"grad_norm": 0.8495398163795471,
"learning_rate": 7.500485698866702e-05,
"loss": 3.1929,
"step": 81200
},
{
"epoch": 8.76010781671159,
"grad_norm": 0.8666577935218811,
"learning_rate": 7.46810577441986e-05,
"loss": 3.1763,
"step": 81250
},
{
"epoch": 8.765498652291106,
"grad_norm": 0.8985541462898254,
"learning_rate": 7.435725849973016e-05,
"loss": 3.1635,
"step": 81300
},
{
"epoch": 8.77088948787062,
"grad_norm": 0.9005835652351379,
"learning_rate": 7.40399352401511e-05,
"loss": 3.1889,
"step": 81350
},
{
"epoch": 8.776280323450134,
"grad_norm": 0.8762881755828857,
"learning_rate": 7.371613599568268e-05,
"loss": 3.2022,
"step": 81400
},
{
"epoch": 8.78167115902965,
"grad_norm": 0.8904643058776855,
"learning_rate": 7.339233675121424e-05,
"loss": 3.1797,
"step": 81450
},
{
"epoch": 8.787061994609164,
"grad_norm": 0.8420517444610596,
"learning_rate": 7.306853750674582e-05,
"loss": 3.1833,
"step": 81500
},
{
"epoch": 8.79245283018868,
"grad_norm": 0.9049489498138428,
"learning_rate": 7.274473826227738e-05,
"loss": 3.1837,
"step": 81550
},
{
"epoch": 8.797843665768195,
"grad_norm": 0.8197122812271118,
"learning_rate": 7.242093901780896e-05,
"loss": 3.1612,
"step": 81600
},
{
"epoch": 8.80323450134771,
"grad_norm": 0.8339477181434631,
"learning_rate": 7.209713977334052e-05,
"loss": 3.1911,
"step": 81650
},
{
"epoch": 8.808625336927223,
"grad_norm": 0.9025993347167969,
"learning_rate": 7.177981651376146e-05,
"loss": 3.1948,
"step": 81700
},
{
"epoch": 8.814016172506738,
"grad_norm": 0.8439270257949829,
"learning_rate": 7.145601726929303e-05,
"loss": 3.1832,
"step": 81750
},
{
"epoch": 8.819407008086253,
"grad_norm": 0.8386377692222595,
"learning_rate": 7.11322180248246e-05,
"loss": 3.1694,
"step": 81800
},
{
"epoch": 8.824797843665769,
"grad_norm": 0.8843401670455933,
"learning_rate": 7.080841878035618e-05,
"loss": 3.1895,
"step": 81850
},
{
"epoch": 8.830188679245284,
"grad_norm": 0.8872575163841248,
"learning_rate": 7.048461953588775e-05,
"loss": 3.1868,
"step": 81900
},
{
"epoch": 8.835579514824797,
"grad_norm": 0.9015069007873535,
"learning_rate": 7.016082029141932e-05,
"loss": 3.1772,
"step": 81950
},
{
"epoch": 8.840970350404312,
"grad_norm": 0.8544393181800842,
"learning_rate": 6.983702104695089e-05,
"loss": 3.193,
"step": 82000
},
{
"epoch": 8.840970350404312,
"eval_accuracy": 0.3926950055853017,
"eval_loss": 3.312908411026001,
"eval_runtime": 184.7408,
"eval_samples_per_second": 97.493,
"eval_steps_per_second": 6.095,
"step": 82000
},
{
"epoch": 8.846361185983827,
"grad_norm": 0.8681711554527283,
"learning_rate": 6.951322180248245e-05,
"loss": 3.1837,
"step": 82050
},
{
"epoch": 8.851752021563343,
"grad_norm": 0.8284791707992554,
"learning_rate": 6.918942255801403e-05,
"loss": 3.1735,
"step": 82100
},
{
"epoch": 8.857142857142858,
"grad_norm": 0.8909093141555786,
"learning_rate": 6.886562331354559e-05,
"loss": 3.192,
"step": 82150
},
{
"epoch": 8.862533692722373,
"grad_norm": 0.8594756722450256,
"learning_rate": 6.854182406907717e-05,
"loss": 3.1784,
"step": 82200
},
{
"epoch": 8.867924528301886,
"grad_norm": 0.8960786461830139,
"learning_rate": 6.821802482460873e-05,
"loss": 3.1938,
"step": 82250
},
{
"epoch": 8.873315363881401,
"grad_norm": 0.8839906454086304,
"learning_rate": 6.78942255801403e-05,
"loss": 3.1797,
"step": 82300
},
{
"epoch": 8.878706199460916,
"grad_norm": 0.8556256890296936,
"learning_rate": 6.757042633567188e-05,
"loss": 3.1884,
"step": 82350
},
{
"epoch": 8.884097035040432,
"grad_norm": 0.8867184519767761,
"learning_rate": 6.724662709120344e-05,
"loss": 3.178,
"step": 82400
},
{
"epoch": 8.889487870619947,
"grad_norm": 0.8656855821609497,
"learning_rate": 6.692282784673502e-05,
"loss": 3.1904,
"step": 82450
},
{
"epoch": 8.89487870619946,
"grad_norm": 0.8861967325210571,
"learning_rate": 6.659902860226658e-05,
"loss": 3.1795,
"step": 82500
},
{
"epoch": 8.900269541778975,
"grad_norm": 0.8820407390594482,
"learning_rate": 6.627522935779816e-05,
"loss": 3.1676,
"step": 82550
},
{
"epoch": 8.90566037735849,
"grad_norm": 0.8465486764907837,
"learning_rate": 6.595143011332972e-05,
"loss": 3.186,
"step": 82600
},
{
"epoch": 8.911051212938006,
"grad_norm": 0.8973132371902466,
"learning_rate": 6.56276308688613e-05,
"loss": 3.1775,
"step": 82650
},
{
"epoch": 8.91644204851752,
"grad_norm": 0.9100452065467834,
"learning_rate": 6.530383162439288e-05,
"loss": 3.1962,
"step": 82700
},
{
"epoch": 8.921832884097036,
"grad_norm": 0.8633553385734558,
"learning_rate": 6.498003237992444e-05,
"loss": 3.1714,
"step": 82750
},
{
"epoch": 8.92722371967655,
"grad_norm": 0.896364688873291,
"learning_rate": 6.465623313545602e-05,
"loss": 3.1731,
"step": 82800
},
{
"epoch": 8.932614555256064,
"grad_norm": 0.8561614751815796,
"learning_rate": 6.433243389098759e-05,
"loss": 3.1777,
"step": 82850
},
{
"epoch": 8.93800539083558,
"grad_norm": 0.8520424365997314,
"learning_rate": 6.400863464651915e-05,
"loss": 3.1741,
"step": 82900
},
{
"epoch": 8.943396226415095,
"grad_norm": 0.8984143733978271,
"learning_rate": 6.368483540205073e-05,
"loss": 3.1753,
"step": 82950
},
{
"epoch": 8.94878706199461,
"grad_norm": 0.8783287405967712,
"learning_rate": 6.336103615758229e-05,
"loss": 3.1696,
"step": 83000
},
{
"epoch": 8.94878706199461,
"eval_accuracy": 0.39304660633026933,
"eval_loss": 3.3102734088897705,
"eval_runtime": 184.4602,
"eval_samples_per_second": 97.642,
"eval_steps_per_second": 6.104,
"step": 83000
},
{
"epoch": 8.954177897574123,
"grad_norm": 0.930332362651825,
"learning_rate": 6.303723691311386e-05,
"loss": 3.1676,
"step": 83050
},
{
"epoch": 8.959568733153638,
"grad_norm": 0.854026734828949,
"learning_rate": 6.271343766864543e-05,
"loss": 3.1752,
"step": 83100
},
{
"epoch": 8.964959568733153,
"grad_norm": 0.8942757844924927,
"learning_rate": 6.2389638424177e-05,
"loss": 3.202,
"step": 83150
},
{
"epoch": 8.970350404312669,
"grad_norm": 0.8567874431610107,
"learning_rate": 6.206583917970857e-05,
"loss": 3.1809,
"step": 83200
},
{
"epoch": 8.975741239892184,
"grad_norm": 0.8844702839851379,
"learning_rate": 6.174203993524014e-05,
"loss": 3.1666,
"step": 83250
},
{
"epoch": 8.981132075471699,
"grad_norm": 0.8592445254325867,
"learning_rate": 6.141824069077172e-05,
"loss": 3.1772,
"step": 83300
},
{
"epoch": 8.986522911051212,
"grad_norm": 0.8552125096321106,
"learning_rate": 6.109444144630328e-05,
"loss": 3.1808,
"step": 83350
},
{
"epoch": 8.991913746630727,
"grad_norm": 0.8916793465614319,
"learning_rate": 6.077064220183486e-05,
"loss": 3.1705,
"step": 83400
},
{
"epoch": 8.997304582210242,
"grad_norm": 0.9587026238441467,
"learning_rate": 6.044684295736643e-05,
"loss": 3.195,
"step": 83450
},
{
"epoch": 9.002695417789758,
"grad_norm": 0.8847026228904724,
"learning_rate": 6.0123043712898e-05,
"loss": 3.1235,
"step": 83500
},
{
"epoch": 9.008086253369273,
"grad_norm": 0.8513318300247192,
"learning_rate": 5.979924446842957e-05,
"loss": 3.1195,
"step": 83550
},
{
"epoch": 9.013477088948788,
"grad_norm": 0.8492772579193115,
"learning_rate": 5.9475445223961135e-05,
"loss": 3.1371,
"step": 83600
},
{
"epoch": 9.018867924528301,
"grad_norm": 0.841739296913147,
"learning_rate": 5.915164597949271e-05,
"loss": 3.1293,
"step": 83650
},
{
"epoch": 9.024258760107816,
"grad_norm": 0.8554039597511292,
"learning_rate": 5.882784673502428e-05,
"loss": 3.1153,
"step": 83700
},
{
"epoch": 9.029649595687331,
"grad_norm": 0.8481983542442322,
"learning_rate": 5.8504047490555855e-05,
"loss": 3.1114,
"step": 83750
},
{
"epoch": 9.035040431266847,
"grad_norm": 0.8705419898033142,
"learning_rate": 5.818024824608742e-05,
"loss": 3.1149,
"step": 83800
},
{
"epoch": 9.040431266846362,
"grad_norm": 0.8700477480888367,
"learning_rate": 5.7856449001618996e-05,
"loss": 3.1036,
"step": 83850
},
{
"epoch": 9.045822102425875,
"grad_norm": 0.898181140422821,
"learning_rate": 5.753264975715056e-05,
"loss": 3.1196,
"step": 83900
},
{
"epoch": 9.05121293800539,
"grad_norm": 0.8250638246536255,
"learning_rate": 5.720885051268213e-05,
"loss": 3.1194,
"step": 83950
},
{
"epoch": 9.056603773584905,
"grad_norm": 0.8478760123252869,
"learning_rate": 5.68850512682137e-05,
"loss": 3.1063,
"step": 84000
},
{
"epoch": 9.056603773584905,
"eval_accuracy": 0.39300466631557296,
"eval_loss": 3.312624454498291,
"eval_runtime": 184.4627,
"eval_samples_per_second": 97.64,
"eval_steps_per_second": 6.104,
"step": 84000
},
{
"epoch": 9.06199460916442,
"grad_norm": 0.8254557251930237,
"learning_rate": 5.656125202374527e-05,
"loss": 3.1188,
"step": 84050
},
{
"epoch": 9.067385444743936,
"grad_norm": 0.884692907333374,
"learning_rate": 5.623745277927684e-05,
"loss": 3.1455,
"step": 84100
},
{
"epoch": 9.07277628032345,
"grad_norm": 0.8918116688728333,
"learning_rate": 5.5913653534808415e-05,
"loss": 3.1243,
"step": 84150
},
{
"epoch": 9.078167115902964,
"grad_norm": 0.8919700384140015,
"learning_rate": 5.558985429033998e-05,
"loss": 3.1391,
"step": 84200
},
{
"epoch": 9.08355795148248,
"grad_norm": 0.8372625112533569,
"learning_rate": 5.526605504587156e-05,
"loss": 3.134,
"step": 84250
},
{
"epoch": 9.088948787061994,
"grad_norm": 0.8807218670845032,
"learning_rate": 5.494225580140313e-05,
"loss": 3.1247,
"step": 84300
},
{
"epoch": 9.09433962264151,
"grad_norm": 0.8806069493293762,
"learning_rate": 5.461845655693469e-05,
"loss": 3.1166,
"step": 84350
},
{
"epoch": 9.099730458221025,
"grad_norm": 0.9334479570388794,
"learning_rate": 5.429465731246627e-05,
"loss": 3.1087,
"step": 84400
},
{
"epoch": 9.10512129380054,
"grad_norm": 0.8704330921173096,
"learning_rate": 5.3970858067997833e-05,
"loss": 3.1345,
"step": 84450
},
{
"epoch": 9.110512129380053,
"grad_norm": 0.8687247037887573,
"learning_rate": 5.364705882352941e-05,
"loss": 3.1202,
"step": 84500
},
{
"epoch": 9.115902964959568,
"grad_norm": 0.8664839267730713,
"learning_rate": 5.3323259579060975e-05,
"loss": 3.13,
"step": 84550
},
{
"epoch": 9.121293800539084,
"grad_norm": 0.8471352458000183,
"learning_rate": 5.299946033459255e-05,
"loss": 3.1172,
"step": 84600
},
{
"epoch": 9.126684636118599,
"grad_norm": 0.9279667735099792,
"learning_rate": 5.267566109012412e-05,
"loss": 3.105,
"step": 84650
},
{
"epoch": 9.132075471698114,
"grad_norm": 0.880308210849762,
"learning_rate": 5.235186184565569e-05,
"loss": 3.1192,
"step": 84700
},
{
"epoch": 9.137466307277627,
"grad_norm": 0.9444160461425781,
"learning_rate": 5.202806260118726e-05,
"loss": 3.1225,
"step": 84750
},
{
"epoch": 9.142857142857142,
"grad_norm": 0.8844714760780334,
"learning_rate": 5.170426335671883e-05,
"loss": 3.1219,
"step": 84800
},
{
"epoch": 9.148247978436657,
"grad_norm": 0.9410858154296875,
"learning_rate": 5.1380464112250394e-05,
"loss": 3.132,
"step": 84850
},
{
"epoch": 9.153638814016173,
"grad_norm": 0.8839712738990784,
"learning_rate": 5.105666486778197e-05,
"loss": 3.1395,
"step": 84900
},
{
"epoch": 9.159029649595688,
"grad_norm": 0.8219244480133057,
"learning_rate": 5.0732865623313536e-05,
"loss": 3.1277,
"step": 84950
},
{
"epoch": 9.164420485175203,
"grad_norm": 0.9190399646759033,
"learning_rate": 5.040906637884511e-05,
"loss": 3.1415,
"step": 85000
},
{
"epoch": 9.164420485175203,
"eval_accuracy": 0.39303758814058076,
"eval_loss": 3.312465190887451,
"eval_runtime": 184.4457,
"eval_samples_per_second": 97.649,
"eval_steps_per_second": 6.105,
"step": 85000
},
{
"epoch": 9.169811320754716,
"grad_norm": 0.8677827715873718,
"learning_rate": 5.0085267134376684e-05,
"loss": 3.1329,
"step": 85050
},
{
"epoch": 9.175202156334231,
"grad_norm": 0.9108098745346069,
"learning_rate": 4.9761467889908255e-05,
"loss": 3.1309,
"step": 85100
},
{
"epoch": 9.180592991913747,
"grad_norm": 0.8512383699417114,
"learning_rate": 4.9437668645439826e-05,
"loss": 3.1368,
"step": 85150
},
{
"epoch": 9.185983827493262,
"grad_norm": 0.9244863986968994,
"learning_rate": 4.911386940097139e-05,
"loss": 3.1399,
"step": 85200
},
{
"epoch": 9.191374663072777,
"grad_norm": 0.9306591153144836,
"learning_rate": 4.879007015650297e-05,
"loss": 3.1145,
"step": 85250
},
{
"epoch": 9.19676549865229,
"grad_norm": 0.8886484503746033,
"learning_rate": 4.846627091203453e-05,
"loss": 3.126,
"step": 85300
},
{
"epoch": 9.202156334231805,
"grad_norm": 0.8817251324653625,
"learning_rate": 4.814247166756611e-05,
"loss": 3.1199,
"step": 85350
},
{
"epoch": 9.20754716981132,
"grad_norm": 0.867629885673523,
"learning_rate": 4.7818672423097674e-05,
"loss": 3.1299,
"step": 85400
},
{
"epoch": 9.212938005390836,
"grad_norm": 0.892943263053894,
"learning_rate": 4.7494873178629244e-05,
"loss": 3.133,
"step": 85450
},
{
"epoch": 9.21832884097035,
"grad_norm": 0.8773700594902039,
"learning_rate": 4.7171073934160815e-05,
"loss": 3.1207,
"step": 85500
},
{
"epoch": 9.223719676549866,
"grad_norm": 0.9094504117965698,
"learning_rate": 4.6847274689692386e-05,
"loss": 3.1074,
"step": 85550
},
{
"epoch": 9.22911051212938,
"grad_norm": 0.8651587963104248,
"learning_rate": 4.652347544522396e-05,
"loss": 3.115,
"step": 85600
},
{
"epoch": 9.234501347708894,
"grad_norm": 0.8531384468078613,
"learning_rate": 4.619967620075553e-05,
"loss": 3.1315,
"step": 85650
},
{
"epoch": 9.23989218328841,
"grad_norm": 0.8743939399719238,
"learning_rate": 4.588235294117647e-05,
"loss": 3.1233,
"step": 85700
},
{
"epoch": 9.245283018867925,
"grad_norm": 0.8862136602401733,
"learning_rate": 4.555855369670804e-05,
"loss": 3.1125,
"step": 85750
},
{
"epoch": 9.25067385444744,
"grad_norm": 0.9095916152000427,
"learning_rate": 4.523475445223961e-05,
"loss": 3.1246,
"step": 85800
},
{
"epoch": 9.256064690026955,
"grad_norm": 0.8724414706230164,
"learning_rate": 4.491095520777118e-05,
"loss": 3.1499,
"step": 85850
},
{
"epoch": 9.261455525606468,
"grad_norm": 0.8505734205245972,
"learning_rate": 4.458715596330275e-05,
"loss": 3.1344,
"step": 85900
},
{
"epoch": 9.266846361185983,
"grad_norm": 0.9521875977516174,
"learning_rate": 4.4263356718834314e-05,
"loss": 3.1343,
"step": 85950
},
{
"epoch": 9.272237196765499,
"grad_norm": 0.9096493721008301,
"learning_rate": 4.393955747436589e-05,
"loss": 3.13,
"step": 86000
},
{
"epoch": 9.272237196765499,
"eval_accuracy": 0.3934594873039644,
"eval_loss": 3.3106472492218018,
"eval_runtime": 184.466,
"eval_samples_per_second": 97.639,
"eval_steps_per_second": 6.104,
"step": 86000
},
{
"epoch": 9.277628032345014,
"grad_norm": 0.8777483105659485,
"learning_rate": 4.3615758229897456e-05,
"loss": 3.1408,
"step": 86050
},
{
"epoch": 9.283018867924529,
"grad_norm": 0.8877856135368347,
"learning_rate": 4.3291958985429034e-05,
"loss": 3.1326,
"step": 86100
},
{
"epoch": 9.288409703504042,
"grad_norm": 0.887473464012146,
"learning_rate": 4.29681597409606e-05,
"loss": 3.1452,
"step": 86150
},
{
"epoch": 9.293800539083557,
"grad_norm": 0.9033775925636292,
"learning_rate": 4.264436049649217e-05,
"loss": 3.1387,
"step": 86200
},
{
"epoch": 9.299191374663073,
"grad_norm": 0.8920115828514099,
"learning_rate": 4.232056125202374e-05,
"loss": 3.1406,
"step": 86250
},
{
"epoch": 9.304582210242588,
"grad_norm": 0.8704814314842224,
"learning_rate": 4.199676200755531e-05,
"loss": 3.141,
"step": 86300
},
{
"epoch": 9.309973045822103,
"grad_norm": 0.8702887296676636,
"learning_rate": 4.167296276308688e-05,
"loss": 3.1116,
"step": 86350
},
{
"epoch": 9.315363881401618,
"grad_norm": 0.8901056051254272,
"learning_rate": 4.134916351861845e-05,
"loss": 3.122,
"step": 86400
},
{
"epoch": 9.320754716981131,
"grad_norm": 0.8449644446372986,
"learning_rate": 4.102536427415003e-05,
"loss": 3.1337,
"step": 86450
},
{
"epoch": 9.326145552560646,
"grad_norm": 0.8996835350990295,
"learning_rate": 4.0701565029681594e-05,
"loss": 3.1294,
"step": 86500
},
{
"epoch": 9.331536388140162,
"grad_norm": 0.883017361164093,
"learning_rate": 4.0377765785213165e-05,
"loss": 3.139,
"step": 86550
},
{
"epoch": 9.336927223719677,
"grad_norm": 0.9052524566650391,
"learning_rate": 4.0053966540744736e-05,
"loss": 3.1503,
"step": 86600
},
{
"epoch": 9.342318059299192,
"grad_norm": 0.8887389302253723,
"learning_rate": 3.973016729627631e-05,
"loss": 3.1186,
"step": 86650
},
{
"epoch": 9.347708894878707,
"grad_norm": 0.8383788466453552,
"learning_rate": 3.940636805180787e-05,
"loss": 3.1292,
"step": 86700
},
{
"epoch": 9.35309973045822,
"grad_norm": 0.8928412199020386,
"learning_rate": 3.908256880733945e-05,
"loss": 3.1121,
"step": 86750
},
{
"epoch": 9.358490566037736,
"grad_norm": 0.9530770182609558,
"learning_rate": 3.875876956287101e-05,
"loss": 3.1149,
"step": 86800
},
{
"epoch": 9.36388140161725,
"grad_norm": 0.8945717811584473,
"learning_rate": 3.843497031840259e-05,
"loss": 3.134,
"step": 86850
},
{
"epoch": 9.369272237196766,
"grad_norm": 0.8866100311279297,
"learning_rate": 3.8111171073934154e-05,
"loss": 3.1381,
"step": 86900
},
{
"epoch": 9.374663072776281,
"grad_norm": 0.8891196846961975,
"learning_rate": 3.778737182946573e-05,
"loss": 3.1398,
"step": 86950
},
{
"epoch": 9.380053908355794,
"grad_norm": 0.8994596600532532,
"learning_rate": 3.7463572584997296e-05,
"loss": 3.1165,
"step": 87000
},
{
"epoch": 9.380053908355794,
"eval_accuracy": 0.3936893968145798,
"eval_loss": 3.3076395988464355,
"eval_runtime": 184.6783,
"eval_samples_per_second": 97.526,
"eval_steps_per_second": 6.097,
"step": 87000
},
{
"epoch": 9.38544474393531,
"grad_norm": 0.898129940032959,
"learning_rate": 3.713977334052887e-05,
"loss": 3.1355,
"step": 87050
},
{
"epoch": 9.390835579514825,
"grad_norm": 0.912468433380127,
"learning_rate": 3.681597409606044e-05,
"loss": 3.126,
"step": 87100
},
{
"epoch": 9.39622641509434,
"grad_norm": 0.8876442909240723,
"learning_rate": 3.649217485159201e-05,
"loss": 3.1312,
"step": 87150
},
{
"epoch": 9.401617250673855,
"grad_norm": 0.8988595008850098,
"learning_rate": 3.616837560712358e-05,
"loss": 3.1266,
"step": 87200
},
{
"epoch": 9.40700808625337,
"grad_norm": 0.9078181385993958,
"learning_rate": 3.584457636265515e-05,
"loss": 3.1421,
"step": 87250
},
{
"epoch": 9.412398921832883,
"grad_norm": 0.9297213554382324,
"learning_rate": 3.552077711818672e-05,
"loss": 3.1491,
"step": 87300
},
{
"epoch": 9.417789757412399,
"grad_norm": 0.8755192756652832,
"learning_rate": 3.519697787371829e-05,
"loss": 3.125,
"step": 87350
},
{
"epoch": 9.423180592991914,
"grad_norm": 0.9355217218399048,
"learning_rate": 3.487317862924986e-05,
"loss": 3.1252,
"step": 87400
},
{
"epoch": 9.428571428571429,
"grad_norm": 0.9134169220924377,
"learning_rate": 3.4549379384781434e-05,
"loss": 3.1318,
"step": 87450
},
{
"epoch": 9.433962264150944,
"grad_norm": 0.9238007664680481,
"learning_rate": 3.4225580140313e-05,
"loss": 3.1327,
"step": 87500
},
{
"epoch": 9.439353099730457,
"grad_norm": 0.9005650877952576,
"learning_rate": 3.390178089584457e-05,
"loss": 3.1445,
"step": 87550
},
{
"epoch": 9.444743935309972,
"grad_norm": 0.893918514251709,
"learning_rate": 3.357798165137615e-05,
"loss": 3.1309,
"step": 87600
},
{
"epoch": 9.450134770889488,
"grad_norm": 1.8303030729293823,
"learning_rate": 3.325418240690772e-05,
"loss": 3.145,
"step": 87650
},
{
"epoch": 9.455525606469003,
"grad_norm": 0.9035360217094421,
"learning_rate": 3.293038316243929e-05,
"loss": 3.1048,
"step": 87700
},
{
"epoch": 9.460916442048518,
"grad_norm": 0.8568888902664185,
"learning_rate": 3.260658391797086e-05,
"loss": 3.1245,
"step": 87750
},
{
"epoch": 9.466307277628033,
"grad_norm": 0.8520007133483887,
"learning_rate": 3.2282784673502424e-05,
"loss": 3.1212,
"step": 87800
},
{
"epoch": 9.471698113207546,
"grad_norm": 0.8967655301094055,
"learning_rate": 3.1958985429033995e-05,
"loss": 3.1312,
"step": 87850
},
{
"epoch": 9.477088948787062,
"grad_norm": 0.9263405203819275,
"learning_rate": 3.1635186184565565e-05,
"loss": 3.1434,
"step": 87900
},
{
"epoch": 9.482479784366577,
"grad_norm": 0.908214271068573,
"learning_rate": 3.1311386940097136e-05,
"loss": 3.1432,
"step": 87950
},
{
"epoch": 9.487870619946092,
"grad_norm": 0.881227970123291,
"learning_rate": 3.098758769562871e-05,
"loss": 3.1336,
"step": 88000
},
{
"epoch": 9.487870619946092,
"eval_accuracy": 0.39417192428936393,
"eval_loss": 3.304806709289551,
"eval_runtime": 184.7481,
"eval_samples_per_second": 97.489,
"eval_steps_per_second": 6.095,
"step": 88000
},
{
"epoch": 9.493261455525607,
"grad_norm": 0.8796641230583191,
"learning_rate": 3.066378845116028e-05,
"loss": 3.1323,
"step": 88050
},
{
"epoch": 9.498652291105122,
"grad_norm": 0.8613610863685608,
"learning_rate": 3.033998920669185e-05,
"loss": 3.1316,
"step": 88100
},
{
"epoch": 9.504043126684635,
"grad_norm": 0.8763812780380249,
"learning_rate": 3.0016189962223416e-05,
"loss": 3.13,
"step": 88150
},
{
"epoch": 9.50943396226415,
"grad_norm": 0.8815705180168152,
"learning_rate": 2.9692390717754987e-05,
"loss": 3.1248,
"step": 88200
},
{
"epoch": 9.514824797843666,
"grad_norm": 0.87941575050354,
"learning_rate": 2.9368591473286558e-05,
"loss": 3.1335,
"step": 88250
},
{
"epoch": 9.520215633423181,
"grad_norm": 0.8994600772857666,
"learning_rate": 2.904479222881813e-05,
"loss": 3.1398,
"step": 88300
},
{
"epoch": 9.525606469002696,
"grad_norm": 0.8343129754066467,
"learning_rate": 2.87209929843497e-05,
"loss": 3.1401,
"step": 88350
},
{
"epoch": 9.530997304582211,
"grad_norm": 0.8785998225212097,
"learning_rate": 2.8397193739881274e-05,
"loss": 3.1398,
"step": 88400
},
{
"epoch": 9.536388140161725,
"grad_norm": 0.889752209186554,
"learning_rate": 2.8073394495412842e-05,
"loss": 3.1353,
"step": 88450
},
{
"epoch": 9.54177897574124,
"grad_norm": 0.8975436091423035,
"learning_rate": 2.7749595250944413e-05,
"loss": 3.1257,
"step": 88500
},
{
"epoch": 9.547169811320755,
"grad_norm": 0.8848623037338257,
"learning_rate": 2.7425796006475984e-05,
"loss": 3.1194,
"step": 88550
},
{
"epoch": 9.55256064690027,
"grad_norm": 0.9181307554244995,
"learning_rate": 2.7101996762007554e-05,
"loss": 3.1369,
"step": 88600
},
{
"epoch": 9.557951482479785,
"grad_norm": 0.8545132279396057,
"learning_rate": 2.6778197517539125e-05,
"loss": 3.1043,
"step": 88650
},
{
"epoch": 9.563342318059298,
"grad_norm": 0.8786166310310364,
"learning_rate": 2.6454398273070693e-05,
"loss": 3.1271,
"step": 88700
},
{
"epoch": 9.568733153638814,
"grad_norm": 0.8818772435188293,
"learning_rate": 2.6130599028602264e-05,
"loss": 3.1327,
"step": 88750
},
{
"epoch": 9.574123989218329,
"grad_norm": 0.8714178800582886,
"learning_rate": 2.5806799784133835e-05,
"loss": 3.1114,
"step": 88800
},
{
"epoch": 9.579514824797844,
"grad_norm": 0.9299944043159485,
"learning_rate": 2.5483000539665406e-05,
"loss": 3.1306,
"step": 88850
},
{
"epoch": 9.584905660377359,
"grad_norm": 0.9006791114807129,
"learning_rate": 2.5159201295196976e-05,
"loss": 3.1503,
"step": 88900
},
{
"epoch": 9.590296495956874,
"grad_norm": 0.8869858980178833,
"learning_rate": 2.4835402050728544e-05,
"loss": 3.1357,
"step": 88950
},
{
"epoch": 9.595687331536388,
"grad_norm": 0.9124794006347656,
"learning_rate": 2.4511602806260115e-05,
"loss": 3.1274,
"step": 89000
},
{
"epoch": 9.595687331536388,
"eval_accuracy": 0.39424341788954587,
"eval_loss": 3.303058385848999,
"eval_runtime": 184.7142,
"eval_samples_per_second": 97.507,
"eval_steps_per_second": 6.096,
"step": 89000
},
{
"epoch": 9.601078167115903,
"grad_norm": 0.8469551205635071,
"learning_rate": 2.4187803561791686e-05,
"loss": 3.1256,
"step": 89050
},
{
"epoch": 9.606469002695418,
"grad_norm": 0.8761835694313049,
"learning_rate": 2.3864004317323257e-05,
"loss": 3.1326,
"step": 89100
},
{
"epoch": 9.611859838274933,
"grad_norm": 0.9657837748527527,
"learning_rate": 2.354020507285483e-05,
"loss": 3.1538,
"step": 89150
},
{
"epoch": 9.617250673854448,
"grad_norm": 0.9288140535354614,
"learning_rate": 2.3216405828386402e-05,
"loss": 3.1204,
"step": 89200
},
{
"epoch": 9.622641509433961,
"grad_norm": 0.9108661413192749,
"learning_rate": 2.289260658391797e-05,
"loss": 3.1401,
"step": 89250
},
{
"epoch": 9.628032345013477,
"grad_norm": 0.916323184967041,
"learning_rate": 2.256880733944954e-05,
"loss": 3.1294,
"step": 89300
},
{
"epoch": 9.633423180592992,
"grad_norm": 0.8727979063987732,
"learning_rate": 2.224500809498111e-05,
"loss": 3.1258,
"step": 89350
},
{
"epoch": 9.638814016172507,
"grad_norm": 0.9134992361068726,
"learning_rate": 2.1921208850512682e-05,
"loss": 3.1309,
"step": 89400
},
{
"epoch": 9.644204851752022,
"grad_norm": 0.9138718843460083,
"learning_rate": 2.1603885590933617e-05,
"loss": 3.1315,
"step": 89450
},
{
"epoch": 9.649595687331537,
"grad_norm": 0.8725118637084961,
"learning_rate": 2.1280086346465188e-05,
"loss": 3.1443,
"step": 89500
},
{
"epoch": 9.65498652291105,
"grad_norm": 0.9107866287231445,
"learning_rate": 2.095628710199676e-05,
"loss": 3.1297,
"step": 89550
},
{
"epoch": 9.660377358490566,
"grad_norm": 0.91167813539505,
"learning_rate": 2.063248785752833e-05,
"loss": 3.1215,
"step": 89600
},
{
"epoch": 9.66576819407008,
"grad_norm": 0.9023905992507935,
"learning_rate": 2.03086886130599e-05,
"loss": 3.1207,
"step": 89650
},
{
"epoch": 9.671159029649596,
"grad_norm": 0.9232180714607239,
"learning_rate": 1.9984889368591468e-05,
"loss": 3.1251,
"step": 89700
},
{
"epoch": 9.676549865229111,
"grad_norm": 0.8913125395774841,
"learning_rate": 1.9661090124123042e-05,
"loss": 3.1404,
"step": 89750
},
{
"epoch": 9.681940700808624,
"grad_norm": 0.8647465705871582,
"learning_rate": 1.9337290879654613e-05,
"loss": 3.121,
"step": 89800
},
{
"epoch": 9.68733153638814,
"grad_norm": 0.8971514105796814,
"learning_rate": 1.9013491635186184e-05,
"loss": 3.1246,
"step": 89850
},
{
"epoch": 9.692722371967655,
"grad_norm": 0.8968616127967834,
"learning_rate": 1.868969239071775e-05,
"loss": 3.1442,
"step": 89900
},
{
"epoch": 9.69811320754717,
"grad_norm": 0.9029696583747864,
"learning_rate": 1.8365893146249326e-05,
"loss": 3.1464,
"step": 89950
},
{
"epoch": 9.703504043126685,
"grad_norm": 0.8942872881889343,
"learning_rate": 1.8042093901780893e-05,
"loss": 3.1301,
"step": 90000
},
{
"epoch": 9.703504043126685,
"eval_accuracy": 0.3945585112642079,
"eval_loss": 3.3003323078155518,
"eval_runtime": 184.5735,
"eval_samples_per_second": 97.582,
"eval_steps_per_second": 6.101,
"step": 90000
},
{
"epoch": 9.7088948787062,
"grad_norm": 0.9043888449668884,
"learning_rate": 1.7718294657312464e-05,
"loss": 3.1165,
"step": 90050
},
{
"epoch": 9.714285714285714,
"grad_norm": 0.9104231595993042,
"learning_rate": 1.7394495412844035e-05,
"loss": 3.1339,
"step": 90100
},
{
"epoch": 9.719676549865229,
"grad_norm": 0.910865068435669,
"learning_rate": 1.7070696168375606e-05,
"loss": 3.1495,
"step": 90150
},
{
"epoch": 9.725067385444744,
"grad_norm": 0.8963308334350586,
"learning_rate": 1.6746896923907177e-05,
"loss": 3.1291,
"step": 90200
},
{
"epoch": 9.730458221024259,
"grad_norm": 0.8748645782470703,
"learning_rate": 1.6423097679438745e-05,
"loss": 3.135,
"step": 90250
},
{
"epoch": 9.735849056603774,
"grad_norm": 0.8760064840316772,
"learning_rate": 1.6099298434970315e-05,
"loss": 3.1281,
"step": 90300
},
{
"epoch": 9.74123989218329,
"grad_norm": 0.9016702175140381,
"learning_rate": 1.577549919050189e-05,
"loss": 3.1361,
"step": 90350
},
{
"epoch": 9.746630727762803,
"grad_norm": 0.872803807258606,
"learning_rate": 1.5451699946033457e-05,
"loss": 3.1442,
"step": 90400
},
{
"epoch": 9.752021563342318,
"grad_norm": 0.8798509836196899,
"learning_rate": 1.5127900701565028e-05,
"loss": 3.1232,
"step": 90450
},
{
"epoch": 9.757412398921833,
"grad_norm": 0.9326780438423157,
"learning_rate": 1.4804101457096599e-05,
"loss": 3.1328,
"step": 90500
},
{
"epoch": 9.762803234501348,
"grad_norm": 0.9405331611633301,
"learning_rate": 1.448030221262817e-05,
"loss": 3.1488,
"step": 90550
},
{
"epoch": 9.768194070080863,
"grad_norm": 0.8831565976142883,
"learning_rate": 1.4156502968159739e-05,
"loss": 3.1268,
"step": 90600
},
{
"epoch": 9.773584905660378,
"grad_norm": 0.8875424861907959,
"learning_rate": 1.383270372369131e-05,
"loss": 3.1198,
"step": 90650
},
{
"epoch": 9.778975741239892,
"grad_norm": 0.8726588487625122,
"learning_rate": 1.3508904479222879e-05,
"loss": 3.1061,
"step": 90700
},
{
"epoch": 9.784366576819407,
"grad_norm": 0.8873887658119202,
"learning_rate": 1.3185105234754452e-05,
"loss": 3.1472,
"step": 90750
},
{
"epoch": 9.789757412398922,
"grad_norm": 0.9473711848258972,
"learning_rate": 1.2861305990286023e-05,
"loss": 3.107,
"step": 90800
},
{
"epoch": 9.795148247978437,
"grad_norm": 0.8777031302452087,
"learning_rate": 1.2537506745817592e-05,
"loss": 3.1201,
"step": 90850
},
{
"epoch": 9.800539083557952,
"grad_norm": 0.9176421165466309,
"learning_rate": 1.2213707501349163e-05,
"loss": 3.1401,
"step": 90900
},
{
"epoch": 9.805929919137466,
"grad_norm": 0.9220699071884155,
"learning_rate": 1.1889908256880732e-05,
"loss": 3.115,
"step": 90950
},
{
"epoch": 9.81132075471698,
"grad_norm": 0.934065580368042,
"learning_rate": 1.1566109012412303e-05,
"loss": 3.13,
"step": 91000
},
{
"epoch": 9.81132075471698,
"eval_accuracy": 0.3948285136904268,
"eval_loss": 3.298856019973755,
"eval_runtime": 184.6381,
"eval_samples_per_second": 97.548,
"eval_steps_per_second": 6.098,
"step": 91000
},
{
"epoch": 9.816711590296496,
"grad_norm": 0.8738986849784851,
"learning_rate": 1.1242309767943874e-05,
"loss": 3.1262,
"step": 91050
},
{
"epoch": 9.822102425876011,
"grad_norm": 0.8833279609680176,
"learning_rate": 1.0918510523475443e-05,
"loss": 3.1231,
"step": 91100
},
{
"epoch": 9.827493261455526,
"grad_norm": 0.8915539979934692,
"learning_rate": 1.0594711279007015e-05,
"loss": 3.1354,
"step": 91150
},
{
"epoch": 9.832884097035041,
"grad_norm": 0.8870952129364014,
"learning_rate": 1.0270912034538586e-05,
"loss": 3.1328,
"step": 91200
},
{
"epoch": 9.838274932614555,
"grad_norm": 0.865484356880188,
"learning_rate": 9.947112790070156e-06,
"loss": 3.1286,
"step": 91250
},
{
"epoch": 9.84366576819407,
"grad_norm": 0.9218941926956177,
"learning_rate": 9.623313545601726e-06,
"loss": 3.1261,
"step": 91300
},
{
"epoch": 9.849056603773585,
"grad_norm": 0.8941046595573425,
"learning_rate": 9.299514301133296e-06,
"loss": 3.1305,
"step": 91350
},
{
"epoch": 9.8544474393531,
"grad_norm": 0.9113878011703491,
"learning_rate": 8.975715056664867e-06,
"loss": 3.1105,
"step": 91400
},
{
"epoch": 9.859838274932615,
"grad_norm": 0.8640049695968628,
"learning_rate": 8.651915812196437e-06,
"loss": 3.1234,
"step": 91450
},
{
"epoch": 9.865229110512129,
"grad_norm": 0.9051119089126587,
"learning_rate": 8.328116567728008e-06,
"loss": 3.1206,
"step": 91500
},
{
"epoch": 9.870619946091644,
"grad_norm": 0.8764170408248901,
"learning_rate": 8.004317323259577e-06,
"loss": 3.1236,
"step": 91550
},
{
"epoch": 9.876010781671159,
"grad_norm": 0.8735215663909912,
"learning_rate": 7.680518078791148e-06,
"loss": 3.1299,
"step": 91600
},
{
"epoch": 9.881401617250674,
"grad_norm": 0.8994240164756775,
"learning_rate": 7.356718834322719e-06,
"loss": 3.1488,
"step": 91650
},
{
"epoch": 9.88679245283019,
"grad_norm": 0.9182621836662292,
"learning_rate": 7.03291958985429e-06,
"loss": 3.1429,
"step": 91700
},
{
"epoch": 9.892183288409704,
"grad_norm": 0.9025397300720215,
"learning_rate": 6.70912034538586e-06,
"loss": 3.1299,
"step": 91750
},
{
"epoch": 9.897574123989218,
"grad_norm": 0.8802763223648071,
"learning_rate": 6.38532110091743e-06,
"loss": 3.1456,
"step": 91800
},
{
"epoch": 9.902964959568733,
"grad_norm": 0.9339905977249146,
"learning_rate": 6.061521856449001e-06,
"loss": 3.1086,
"step": 91850
},
{
"epoch": 9.908355795148248,
"grad_norm": 0.8632808327674866,
"learning_rate": 5.737722611980571e-06,
"loss": 3.1335,
"step": 91900
},
{
"epoch": 9.913746630727763,
"grad_norm": 0.9041385054588318,
"learning_rate": 5.413923367512142e-06,
"loss": 3.1094,
"step": 91950
},
{
"epoch": 9.919137466307278,
"grad_norm": 0.909142255783081,
"learning_rate": 5.090124123043712e-06,
"loss": 3.1407,
"step": 92000
},
{
"epoch": 9.919137466307278,
"eval_accuracy": 0.3949397742475489,
"eval_loss": 3.2971105575561523,
"eval_runtime": 184.6111,
"eval_samples_per_second": 97.562,
"eval_steps_per_second": 6.099,
"step": 92000
},
{
"epoch": 9.924528301886792,
"grad_norm": 0.9313786625862122,
"learning_rate": 4.766324878575283e-06,
"loss": 3.1436,
"step": 92050
},
{
"epoch": 9.929919137466307,
"grad_norm": 0.8563489317893982,
"learning_rate": 4.442525634106853e-06,
"loss": 3.1257,
"step": 92100
},
{
"epoch": 9.935309973045822,
"grad_norm": 0.9260908365249634,
"learning_rate": 4.118726389638424e-06,
"loss": 3.1347,
"step": 92150
},
{
"epoch": 9.940700808625337,
"grad_norm": 0.8603903651237488,
"learning_rate": 3.7949271451699944e-06,
"loss": 3.1165,
"step": 92200
},
{
"epoch": 9.946091644204852,
"grad_norm": 0.8976355791091919,
"learning_rate": 3.4711279007015644e-06,
"loss": 3.1271,
"step": 92250
},
{
"epoch": 9.951482479784367,
"grad_norm": 0.8780588507652283,
"learning_rate": 3.1473286562331353e-06,
"loss": 3.118,
"step": 92300
},
{
"epoch": 9.95687331536388,
"grad_norm": 0.9332984089851379,
"learning_rate": 2.8235294117647054e-06,
"loss": 3.1231,
"step": 92350
},
{
"epoch": 9.962264150943396,
"grad_norm": 0.8642359972000122,
"learning_rate": 2.4997301672962763e-06,
"loss": 3.1244,
"step": 92400
},
{
"epoch": 9.967654986522911,
"grad_norm": 0.9261642694473267,
"learning_rate": 2.1759309228278467e-06,
"loss": 3.1247,
"step": 92450
},
{
"epoch": 9.973045822102426,
"grad_norm": 0.8727864623069763,
"learning_rate": 1.852131678359417e-06,
"loss": 3.1342,
"step": 92500
},
{
"epoch": 9.978436657681941,
"grad_norm": 0.9198867082595825,
"learning_rate": 1.5283324338909875e-06,
"loss": 3.1235,
"step": 92550
},
{
"epoch": 9.983827493261456,
"grad_norm": 0.9133760333061218,
"learning_rate": 1.204533189422558e-06,
"loss": 3.1236,
"step": 92600
},
{
"epoch": 9.98921832884097,
"grad_norm": 0.8645879030227661,
"learning_rate": 8.807339449541284e-07,
"loss": 3.1135,
"step": 92650
},
{
"epoch": 9.994609164420485,
"grad_norm": 0.8892059326171875,
"learning_rate": 5.569347004856989e-07,
"loss": 3.1321,
"step": 92700
},
{
"epoch": 10.0,
"grad_norm": 1.9543075561523438,
"learning_rate": 2.3313545601726927e-07,
"loss": 3.1229,
"step": 92750
},
{
"epoch": 10.0,
"step": 92750,
"total_flos": 7.75449427968e+17,
"train_loss": 3.4540262829556942,
"train_runtime": 79697.0382,
"train_samples_per_second": 37.238,
"train_steps_per_second": 1.164
}
],
"logging_steps": 50,
"max_steps": 92750,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 10000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 7.75449427968e+17,
"train_batch_size": 32,
"trial_name": null,
"trial_params": null
}