| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.6160310277957337, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00808015513897867, | |
| "grad_norm": 3.75984263420105, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.8538, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01616031027795734, | |
| "grad_norm": 3.785407066345215, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.7059, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.024240465416936006, | |
| "grad_norm": 2.6398749351501465, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.5499, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03232062055591468, | |
| "grad_norm": 3.0633111000061035, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.4899, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.040400775694893344, | |
| "grad_norm": 2.5752687454223633, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.4542, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.04848093083387201, | |
| "grad_norm": 2.7746145725250244, | |
| "learning_rate": 3e-06, | |
| "loss": 0.4448, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05656108597285068, | |
| "grad_norm": 2.5188117027282715, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.4047, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06464124111182935, | |
| "grad_norm": 2.549173593521118, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.3918, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.07272139625080802, | |
| "grad_norm": 2.651440382003784, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.3738, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.08080155138978669, | |
| "grad_norm": 2.6359574794769287, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3923, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.08888170652876536, | |
| "grad_norm": 3.015693426132202, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.3597, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.09696186166774402, | |
| "grad_norm": 2.4634430408477783, | |
| "learning_rate": 6e-06, | |
| "loss": 0.3577, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.10504201680672269, | |
| "grad_norm": 2.6268017292022705, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.3397, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.11312217194570136, | |
| "grad_norm": 2.1071865558624268, | |
| "learning_rate": 7e-06, | |
| "loss": 0.344, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.12120232708468003, | |
| "grad_norm": 2.0719916820526123, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3232, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.1292824822236587, | |
| "grad_norm": 2.445434331893921, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3488, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.13736263736263737, | |
| "grad_norm": 2.1177170276641846, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.327, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.14544279250161604, | |
| "grad_norm": 2.2949392795562744, | |
| "learning_rate": 9e-06, | |
| "loss": 0.3228, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1535229476405947, | |
| "grad_norm": 1.701798439025879, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.2907, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.16160310277957338, | |
| "grad_norm": 1.857657790184021, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3021, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.16968325791855204, | |
| "grad_norm": 1.8236263990402222, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.2878, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.1777634130575307, | |
| "grad_norm": 2.30806303024292, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.2741, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.18584356819650938, | |
| "grad_norm": 2.0603034496307373, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.2953, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.19392372333548805, | |
| "grad_norm": 2.1043479442596436, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.2855, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2020038784744667, | |
| "grad_norm": 1.921908974647522, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.2709, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.21008403361344538, | |
| "grad_norm": 2.1535732746124268, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.251, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.21816418875242405, | |
| "grad_norm": 2.01731014251709, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.2817, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.22624434389140272, | |
| "grad_norm": 1.9176462888717651, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.2608, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.23432449903038138, | |
| "grad_norm": 2.2685463428497314, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.2584, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.24240465416936005, | |
| "grad_norm": 1.6904027462005615, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.2529, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.2504848093083387, | |
| "grad_norm": 2.014220714569092, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.2443, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.2585649644473174, | |
| "grad_norm": 2.1156768798828125, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.2548, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.26664511958629605, | |
| "grad_norm": 2.0581390857696533, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.2391, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.27472527472527475, | |
| "grad_norm": 1.9932626485824585, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.2326, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.2828054298642534, | |
| "grad_norm": 1.7348275184631348, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.2399, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.2908855850032321, | |
| "grad_norm": 1.7778377532958984, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.2272, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.2989657401422107, | |
| "grad_norm": 1.917076826095581, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.2498, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.3070458952811894, | |
| "grad_norm": 1.727513313293457, | |
| "learning_rate": 9e-06, | |
| "loss": 0.226, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.31512605042016806, | |
| "grad_norm": 2.011462688446045, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.2358, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.32320620555914675, | |
| "grad_norm": 2.0421934127807617, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.2469, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.32320620555914675, | |
| "eval_loss": 0.23132415115833282, | |
| "eval_runtime": 421.4559, | |
| "eval_samples_per_second": 2.373, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.17325433135828397, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3312863606981254, | |
| "grad_norm": 2.1990480422973633, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.2397, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.3393665158371041, | |
| "grad_norm": 1.6808555126190186, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.212, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3474466709760827, | |
| "grad_norm": 1.9749352931976318, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.2148, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.3555268261150614, | |
| "grad_norm": 2.0048928260803223, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.2161, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.36360698125404006, | |
| "grad_norm": 1.8271081447601318, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.237, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.37168713639301876, | |
| "grad_norm": 1.751715898513794, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.2088, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.3797672915319974, | |
| "grad_norm": 1.490201473236084, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.2207, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.3878474466709761, | |
| "grad_norm": 2.108602285385132, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.2177, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.39592760180995473, | |
| "grad_norm": 1.637518048286438, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.2134, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.4040077569489334, | |
| "grad_norm": 1.8813297748565674, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.2048, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.41208791208791207, | |
| "grad_norm": 1.61934494972229, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.1951, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.42016806722689076, | |
| "grad_norm": 1.9593381881713867, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.1989, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.4282482223658694, | |
| "grad_norm": 1.7399969100952148, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.1837, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.4363283775048481, | |
| "grad_norm": 1.9102469682693481, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.19, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.44440853264382674, | |
| "grad_norm": 1.8125574588775635, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.2015, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.45248868778280543, | |
| "grad_norm": 1.8675082921981812, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2076, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.46056884292178407, | |
| "grad_norm": 2.056074619293213, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.2054, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.46864899806076277, | |
| "grad_norm": 1.4601351022720337, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.2057, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.47672915319974146, | |
| "grad_norm": 1.6205873489379883, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.1887, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.4848093083387201, | |
| "grad_norm": 1.356102705001831, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.1767, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.4928894634776988, | |
| "grad_norm": 2.1016225814819336, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.1859, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.5009696186166774, | |
| "grad_norm": 1.3985319137573242, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.1981, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.5090497737556561, | |
| "grad_norm": 1.8482751846313477, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.1979, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.5171299288946348, | |
| "grad_norm": 1.6422449350357056, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.1858, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.5252100840336135, | |
| "grad_norm": 1.5610370635986328, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.1988, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.5332902391725921, | |
| "grad_norm": 1.5282772779464722, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.1904, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.5413703943115707, | |
| "grad_norm": 1.5836628675460815, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.1913, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 1.7465559244155884, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.1638, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5575307045895281, | |
| "grad_norm": 1.7590513229370117, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.1946, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.5656108597285068, | |
| "grad_norm": 1.7977447509765625, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.1919, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.5736910148674854, | |
| "grad_norm": 1.5070719718933105, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.1801, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.5817711700064642, | |
| "grad_norm": 1.7022117376327515, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.1831, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.5898513251454428, | |
| "grad_norm": 1.6493513584136963, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.1889, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.5979314802844214, | |
| "grad_norm": 1.4757349491119385, | |
| "learning_rate": 7e-06, | |
| "loss": 0.1834, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.6060116354234001, | |
| "grad_norm": 1.7376744747161865, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.1676, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.6140917905623788, | |
| "grad_norm": 1.9608112573623657, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.1884, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.6221719457013575, | |
| "grad_norm": 1.412118673324585, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.1949, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.6302521008403361, | |
| "grad_norm": 2.119964361190796, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.193, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.6383322559793148, | |
| "grad_norm": 1.9113435745239258, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.1662, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.6464124111182935, | |
| "grad_norm": 1.3223708868026733, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.1578, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6464124111182935, | |
| "eval_loss": 0.1784905195236206, | |
| "eval_runtime": 427.4993, | |
| "eval_samples_per_second": 2.339, | |
| "eval_steps_per_second": 0.147, | |
| "eval_wer": 0.13732843321083027, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6544925662572721, | |
| "grad_norm": 1.5342998504638672, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.1727, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.6625727213962508, | |
| "grad_norm": 1.322567105293274, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.1858, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.6706528765352294, | |
| "grad_norm": 1.5235811471939087, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.1523, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.6787330316742082, | |
| "grad_norm": 1.478602647781372, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.1667, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.6868131868131868, | |
| "grad_norm": 1.6120604276657104, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.163, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.6948933419521655, | |
| "grad_norm": 1.3464877605438232, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.1624, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.7029734970911441, | |
| "grad_norm": 1.2172194719314575, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.1483, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.7110536522301228, | |
| "grad_norm": 1.5536859035491943, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.1552, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.7191338073691015, | |
| "grad_norm": 1.394464373588562, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.1536, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.7272139625080801, | |
| "grad_norm": 1.8853507041931152, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.1673, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.7352941176470589, | |
| "grad_norm": 1.529552936553955, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.1622, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.7433742727860375, | |
| "grad_norm": 1.541446566581726, | |
| "learning_rate": 6e-06, | |
| "loss": 0.1647, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.7514544279250162, | |
| "grad_norm": 1.6299506425857544, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.172, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.7595345830639948, | |
| "grad_norm": 1.5410858392715454, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.1558, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.7676147382029735, | |
| "grad_norm": 1.5796810388565063, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.158, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.7756948933419522, | |
| "grad_norm": 1.5637125968933105, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.1583, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.7837750484809308, | |
| "grad_norm": 1.4852577447891235, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.1618, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.7918552036199095, | |
| "grad_norm": 1.6491366624832153, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.16, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.7999353587588882, | |
| "grad_norm": 1.7685949802398682, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.1706, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.8080155138978669, | |
| "grad_norm": 1.4126862287521362, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.1525, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.8160956690368455, | |
| "grad_norm": 1.4942584037780762, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.1628, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.8241758241758241, | |
| "grad_norm": 1.6680798530578613, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.1647, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.8322559793148029, | |
| "grad_norm": 1.1703355312347412, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.1519, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.8403361344537815, | |
| "grad_norm": 1.4046279191970825, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.1492, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.8484162895927602, | |
| "grad_norm": 1.4554412364959717, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.1452, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.8564964447317388, | |
| "grad_norm": 1.737576961517334, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.1629, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.8645765998707176, | |
| "grad_norm": 1.261842966079712, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.1539, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.8726567550096962, | |
| "grad_norm": 1.5773130655288696, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.1434, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.8807369101486748, | |
| "grad_norm": 1.619884967803955, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.1753, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.8888170652876535, | |
| "grad_norm": 1.175441861152649, | |
| "learning_rate": 5e-06, | |
| "loss": 0.1594, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.8968972204266322, | |
| "grad_norm": 1.8705310821533203, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.1618, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.9049773755656109, | |
| "grad_norm": 1.5610700845718384, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.1537, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.9130575307045895, | |
| "grad_norm": 1.940385103225708, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.1602, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.9211376858435681, | |
| "grad_norm": 1.4349099397659302, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.1492, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.9292178409825469, | |
| "grad_norm": 1.7276194095611572, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.1418, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.9372979961215255, | |
| "grad_norm": 1.138739824295044, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.1566, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.9453781512605042, | |
| "grad_norm": 1.1318280696868896, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.1489, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.9534583063994829, | |
| "grad_norm": 1.766103744506836, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.1567, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.9615384615384616, | |
| "grad_norm": 1.9743990898132324, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.1524, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.9696186166774402, | |
| "grad_norm": 1.7132006883621216, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.1491, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9696186166774402, | |
| "eval_loss": 0.15305228531360626, | |
| "eval_runtime": 421.6838, | |
| "eval_samples_per_second": 2.371, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.12127803195079877, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.9776987718164188, | |
| "grad_norm": 1.675523042678833, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.1359, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.9857789269553976, | |
| "grad_norm": 1.1501892805099487, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.1527, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.9938590820943762, | |
| "grad_norm": 1.8712406158447266, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.1572, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.0019392372333549, | |
| "grad_norm": 1.688395619392395, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.143, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.0100193923723335, | |
| "grad_norm": 1.719205379486084, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.1135, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.0180995475113122, | |
| "grad_norm": 1.399117112159729, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.1138, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.0261797026502908, | |
| "grad_norm": 1.3488880395889282, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.097, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.0342598577892697, | |
| "grad_norm": 1.3759610652923584, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.1073, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.0423400129282483, | |
| "grad_norm": 1.2929582595825195, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.1067, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.050420168067227, | |
| "grad_norm": 1.2151849269866943, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.1023, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.0585003232062056, | |
| "grad_norm": 1.4267935752868652, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.1098, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.0665804783451842, | |
| "grad_norm": 1.4334278106689453, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.1001, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.0746606334841629, | |
| "grad_norm": 1.3168810606002808, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.1025, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.0827407886231415, | |
| "grad_norm": 1.4960328340530396, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.109, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.0908209437621201, | |
| "grad_norm": 1.430653691291809, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.0938, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.098901098901099, | |
| "grad_norm": 1.261281132698059, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.0978, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.1069812540400776, | |
| "grad_norm": 1.2067958116531372, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.1008, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.1150614091790563, | |
| "grad_norm": 1.456865906715393, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.1122, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.123141564318035, | |
| "grad_norm": 1.0266293287277222, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.0972, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.1312217194570136, | |
| "grad_norm": 1.4198247194290161, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.11, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.1393018745959922, | |
| "grad_norm": 1.3642793893814087, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.1031, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.1473820297349708, | |
| "grad_norm": 1.474433422088623, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0999, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.1554621848739495, | |
| "grad_norm": 1.726651668548584, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.1052, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.1635423400129283, | |
| "grad_norm": 1.0963325500488281, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0889, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.171622495151907, | |
| "grad_norm": 1.0337573289871216, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.1078, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.1797026502908856, | |
| "grad_norm": 1.2517417669296265, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0983, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.1877828054298643, | |
| "grad_norm": 1.1968861818313599, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0981, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.195862960568843, | |
| "grad_norm": 1.646301031112671, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0982, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.2039431157078215, | |
| "grad_norm": 1.5741595029830933, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.0954, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.2120232708468002, | |
| "grad_norm": 0.865523099899292, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0936, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.220103425985779, | |
| "grad_norm": 1.109647512435913, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.1092, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.2281835811247577, | |
| "grad_norm": 1.2372595071792603, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.1021, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.2362637362637363, | |
| "grad_norm": 1.5520275831222534, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0972, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.244343891402715, | |
| "grad_norm": 1.1914920806884766, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0945, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.2524240465416936, | |
| "grad_norm": 1.269389033317566, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0949, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.2605042016806722, | |
| "grad_norm": 1.683188557624817, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0923, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.2685843568196509, | |
| "grad_norm": 1.4076889753341675, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0992, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.2766645119586295, | |
| "grad_norm": 1.0736989974975586, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.0927, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.2847446670976082, | |
| "grad_norm": 0.8500351905822754, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.092, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.292824822236587, | |
| "grad_norm": 1.2514774799346924, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.099, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.292824822236587, | |
| "eval_loss": 0.1433822065591812, | |
| "eval_runtime": 422.4096, | |
| "eval_samples_per_second": 2.367, | |
| "eval_steps_per_second": 0.149, | |
| "eval_wer": 0.11287782194554864, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.3009049773755657, | |
| "grad_norm": 1.5008472204208374, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0973, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.3089851325145443, | |
| "grad_norm": 1.1695759296417236, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0903, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.317065287653523, | |
| "grad_norm": 1.221191644668579, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0955, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.3251454427925016, | |
| "grad_norm": 1.3147417306900024, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.1003, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.3332255979314802, | |
| "grad_norm": 1.5507885217666626, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.1098, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.341305753070459, | |
| "grad_norm": 1.1320619583129883, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0902, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.3493859082094377, | |
| "grad_norm": 1.003406047821045, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0887, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.3574660633484164, | |
| "grad_norm": 1.2398234605789185, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.1044, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.365546218487395, | |
| "grad_norm": 1.3185756206512451, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0931, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.3736263736263736, | |
| "grad_norm": 1.8669284582138062, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0998, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.3817065287653523, | |
| "grad_norm": 1.4902080297470093, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.0944, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.389786683904331, | |
| "grad_norm": 1.2797434329986572, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.1002, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.3978668390433096, | |
| "grad_norm": 1.3823866844177246, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0848, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.4059469941822882, | |
| "grad_norm": 1.5034900903701782, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.089, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.4140271493212668, | |
| "grad_norm": 1.0645461082458496, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0814, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.4221073044602457, | |
| "grad_norm": 1.3759708404541016, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.084, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.4301874595992243, | |
| "grad_norm": 1.6364449262619019, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.1011, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.438267614738203, | |
| "grad_norm": 1.8542567491531372, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.1041, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.4463477698771816, | |
| "grad_norm": 1.2790472507476807, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0768, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.4544279250161603, | |
| "grad_norm": 1.3523430824279785, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0862, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.4625080801551391, | |
| "grad_norm": 1.2620325088500977, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0931, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.4705882352941178, | |
| "grad_norm": 1.5867211818695068, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.094, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.4786683904330964, | |
| "grad_norm": 1.3185136318206787, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0987, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.486748545572075, | |
| "grad_norm": 1.5266519784927368, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0936, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.4948287007110537, | |
| "grad_norm": 1.2671607732772827, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0968, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.5029088558500323, | |
| "grad_norm": 0.9610714912414551, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.1071, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.510989010989011, | |
| "grad_norm": 1.585331678390503, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0785, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.5190691661279896, | |
| "grad_norm": 1.5121099948883057, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0802, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.5271493212669682, | |
| "grad_norm": 1.6761611700057983, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0956, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.5352294764059469, | |
| "grad_norm": 1.0066338777542114, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0875, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.5433096315449255, | |
| "grad_norm": 0.9917683601379395, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0871, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.5513897866839044, | |
| "grad_norm": 1.4825103282928467, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.1008, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.559469941822883, | |
| "grad_norm": 1.3838367462158203, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0903, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.5675500969618616, | |
| "grad_norm": 1.0742937326431274, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0918, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.5756302521008403, | |
| "grad_norm": 1.3103052377700806, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0964, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.5837104072398192, | |
| "grad_norm": 1.6678719520568848, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.1042, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.5917905623787978, | |
| "grad_norm": 1.3988322019577026, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0925, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.5998707175177764, | |
| "grad_norm": 1.1961122751235962, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0931, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.607950872656755, | |
| "grad_norm": 1.1993706226348877, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.1012, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.6160310277957337, | |
| "grad_norm": 1.3308823108673096, | |
| "learning_rate": 0.0, | |
| "loss": 0.0874, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6160310277957337, | |
| "eval_loss": 0.1364615559577942, | |
| "eval_runtime": 437.3566, | |
| "eval_samples_per_second": 2.286, | |
| "eval_steps_per_second": 0.144, | |
| "eval_wer": 0.10755268881722042, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6160310277957337, | |
| "step": 5000, | |
| "total_flos": 5.435725490631475e+20, | |
| "train_loss": 0.17819489703178407, | |
| "train_runtime": 33619.5535, | |
| "train_samples_per_second": 4.759, | |
| "train_steps_per_second": 0.149 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.435725490631475e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |