| { | |
| "best_metric": 33.472796184515516, | |
| "best_model_checkpoint": "./whisper-distil-v3/checkpoint-26000", | |
| "epoch": 1.4247356019507917, | |
| "eval_steps": 1000, | |
| "global_step": 26000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010959504630390707, | |
| "grad_norm": 5.807405471801758, | |
| "learning_rate": 3.6e-06, | |
| "loss": 8.6037, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0021919009260781414, | |
| "grad_norm": 5.726860523223877, | |
| "learning_rate": 7.6e-06, | |
| "loss": 8.4874, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.003287851389117212, | |
| "grad_norm": 7.615314960479736, | |
| "learning_rate": 1.16e-05, | |
| "loss": 8.0934, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.004383801852156283, | |
| "grad_norm": 7.089465618133545, | |
| "learning_rate": 1.56e-05, | |
| "loss": 7.4227, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.005479752315195353, | |
| "grad_norm": 5.158086776733398, | |
| "learning_rate": 1.9600000000000002e-05, | |
| "loss": 6.3418, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.006575702778234424, | |
| "grad_norm": 3.288583278656006, | |
| "learning_rate": 2.3400000000000003e-05, | |
| "loss": 5.2747, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.007671653241273495, | |
| "grad_norm": 3.1715681552886963, | |
| "learning_rate": 2.7400000000000002e-05, | |
| "loss": 4.3075, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.008767603704312565, | |
| "grad_norm": 3.033198833465576, | |
| "learning_rate": 3.1400000000000004e-05, | |
| "loss": 3.631, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.009863554167351636, | |
| "grad_norm": 3.027251720428467, | |
| "learning_rate": 3.54e-05, | |
| "loss": 3.2186, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.010959504630390707, | |
| "grad_norm": 2.9063901901245117, | |
| "learning_rate": 3.94e-05, | |
| "loss": 2.9226, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.012055455093429777, | |
| "grad_norm": 3.1165690422058105, | |
| "learning_rate": 4.3400000000000005e-05, | |
| "loss": 2.8402, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.013151405556468848, | |
| "grad_norm": 2.7977383136749268, | |
| "learning_rate": 4.74e-05, | |
| "loss": 2.613, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.014247356019507919, | |
| "grad_norm": 3.7818286418914795, | |
| "learning_rate": 5.14e-05, | |
| "loss": 2.377, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.01534330648254699, | |
| "grad_norm": 3.2088804244995117, | |
| "learning_rate": 5.5400000000000005e-05, | |
| "loss": 2.3204, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.01643925694558606, | |
| "grad_norm": 3.2518157958984375, | |
| "learning_rate": 5.94e-05, | |
| "loss": 2.1812, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01753520740862513, | |
| "grad_norm": 3.725226640701294, | |
| "learning_rate": 6.340000000000001e-05, | |
| "loss": 2.158, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.0186311578716642, | |
| "grad_norm": 3.5929486751556396, | |
| "learning_rate": 6.740000000000001e-05, | |
| "loss": 2.1241, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.019727108334703272, | |
| "grad_norm": 4.1317572593688965, | |
| "learning_rate": 7.14e-05, | |
| "loss": 2.2284, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.020823058797742343, | |
| "grad_norm": 3.4276161193847656, | |
| "learning_rate": 7.54e-05, | |
| "loss": 1.9655, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.021919009260781414, | |
| "grad_norm": 3.9775540828704834, | |
| "learning_rate": 7.94e-05, | |
| "loss": 1.9407, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.023014959723820484, | |
| "grad_norm": 3.67511248588562, | |
| "learning_rate": 8.34e-05, | |
| "loss": 1.9312, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.024110910186859555, | |
| "grad_norm": 4.781565189361572, | |
| "learning_rate": 8.740000000000001e-05, | |
| "loss": 1.9218, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.025206860649898626, | |
| "grad_norm": 5.2797698974609375, | |
| "learning_rate": 9.140000000000001e-05, | |
| "loss": 1.8729, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.026302811112937696, | |
| "grad_norm": 6.1737284660339355, | |
| "learning_rate": 9.54e-05, | |
| "loss": 1.6848, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.027398761575976767, | |
| "grad_norm": 4.926702976226807, | |
| "learning_rate": 9.94e-05, | |
| "loss": 1.8866, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.028494712039015838, | |
| "grad_norm": 4.043098449707031, | |
| "learning_rate": 0.0001, | |
| "loss": 1.749, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.02959066250205491, | |
| "grad_norm": 4.022521495819092, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7654, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.03068661296509398, | |
| "grad_norm": 3.1964547634124756, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7496, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.03178256342813305, | |
| "grad_norm": 3.5182583332061768, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7312, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.03287851389117212, | |
| "grad_norm": 3.529665231704712, | |
| "learning_rate": 0.0001, | |
| "loss": 1.6307, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.03397446435421119, | |
| "grad_norm": 3.329401969909668, | |
| "learning_rate": 0.0001, | |
| "loss": 1.7613, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.03507041481725026, | |
| "grad_norm": 3.4481399059295654, | |
| "learning_rate": 0.0001, | |
| "loss": 1.6204, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.03616636528028933, | |
| "grad_norm": 3.3551902770996094, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5846, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.0372623157433284, | |
| "grad_norm": 3.591031074523926, | |
| "learning_rate": 0.0001, | |
| "loss": 1.6077, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.038358266206367474, | |
| "grad_norm": 3.8630764484405518, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5275, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.039454216669406544, | |
| "grad_norm": 3.77461838722229, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5386, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.040550167132445615, | |
| "grad_norm": 2.9158153533935547, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5536, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.041646117595484686, | |
| "grad_norm": 3.761077642440796, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5607, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.042742068058523756, | |
| "grad_norm": 3.5758230686187744, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5145, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.04383801852156283, | |
| "grad_norm": 3.01175856590271, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5639, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.0449339689846019, | |
| "grad_norm": 3.8395230770111084, | |
| "learning_rate": 0.0001, | |
| "loss": 1.6478, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.04602991944764097, | |
| "grad_norm": 2.9240541458129883, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5303, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.04712586991068004, | |
| "grad_norm": 3.603835344314575, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4436, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.04822182037371911, | |
| "grad_norm": 3.1701183319091797, | |
| "learning_rate": 0.0001, | |
| "loss": 1.5622, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.04931777083675818, | |
| "grad_norm": 4.054835796356201, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4354, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.05041372129979725, | |
| "grad_norm": 2.9651615619659424, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4676, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.05150967176283632, | |
| "grad_norm": 3.2480218410491943, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3769, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.05260562222587539, | |
| "grad_norm": 3.494356155395508, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4928, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.05370157268891446, | |
| "grad_norm": 3.394205331802368, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4045, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.054797523151953534, | |
| "grad_norm": 3.333587646484375, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4953, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.054797523151953534, | |
| "eval_loss": 1.4313914775848389, | |
| "eval_runtime": 30911.9498, | |
| "eval_samples_per_second": 2.099, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 70.88587442180551, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.055893473614992605, | |
| "grad_norm": 3.2317609786987305, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4376, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.056989424078031675, | |
| "grad_norm": 2.9077706336975098, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4398, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.058085374541070746, | |
| "grad_norm": 3.0054707527160645, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4326, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.05918132500410982, | |
| "grad_norm": 3.7243480682373047, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4915, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.06027727546714889, | |
| "grad_norm": 2.9608402252197266, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4595, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.06137322593018796, | |
| "grad_norm": 2.652829885482788, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4378, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.06246917639322703, | |
| "grad_norm": 2.9017295837402344, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4257, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.0635651268562661, | |
| "grad_norm": 3.0610859394073486, | |
| "learning_rate": 0.0001, | |
| "loss": 1.348, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.06466107731930516, | |
| "grad_norm": 2.701765775680542, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3853, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.06575702778234424, | |
| "grad_norm": 3.83376145362854, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4708, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.0668529782453833, | |
| "grad_norm": 3.178449869155884, | |
| "learning_rate": 0.0001, | |
| "loss": 1.362, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.06794892870842238, | |
| "grad_norm": 3.796205997467041, | |
| "learning_rate": 0.0001, | |
| "loss": 1.4331, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.06904487917146145, | |
| "grad_norm": 2.8163928985595703, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2835, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.07014082963450052, | |
| "grad_norm": 2.698793888092041, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3444, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.07123678009753959, | |
| "grad_norm": 2.584484815597534, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2145, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.07233273056057866, | |
| "grad_norm": 2.696967363357544, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2855, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.07342868102361773, | |
| "grad_norm": 3.382924795150757, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2164, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.0745246314866568, | |
| "grad_norm": 2.8127260208129883, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2873, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.07562058194969587, | |
| "grad_norm": 2.631011724472046, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3759, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.07671653241273495, | |
| "grad_norm": 2.913276433944702, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2688, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.07781248287577401, | |
| "grad_norm": 2.811455488204956, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2179, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.07890843333881309, | |
| "grad_norm": 2.8242247104644775, | |
| "learning_rate": 0.0001, | |
| "loss": 1.142, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.08000438380185215, | |
| "grad_norm": 3.1733341217041016, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2934, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.08110033426489123, | |
| "grad_norm": 2.491945743560791, | |
| "learning_rate": 0.0001, | |
| "loss": 1.3274, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.0821962847279303, | |
| "grad_norm": 2.717165470123291, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2484, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.08329223519096937, | |
| "grad_norm": 2.3187918663024902, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2038, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.08438818565400844, | |
| "grad_norm": 2.9296529293060303, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1962, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.08548413611704751, | |
| "grad_norm": 2.5763330459594727, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2122, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.08658008658008658, | |
| "grad_norm": 3.4159390926361084, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2302, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.08767603704312565, | |
| "grad_norm": 2.893261432647705, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2106, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.08877198750616472, | |
| "grad_norm": 2.1891727447509766, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2282, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.0898679379692038, | |
| "grad_norm": 2.4100029468536377, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2039, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.09096388843224286, | |
| "grad_norm": 2.5420494079589844, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2201, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.09205983889528194, | |
| "grad_norm": 3.1885313987731934, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2446, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.093155789358321, | |
| "grad_norm": 3.120586633682251, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2308, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.09425173982136008, | |
| "grad_norm": 2.4548628330230713, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1777, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.09534769028439914, | |
| "grad_norm": 3.101803779602051, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2123, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.09644364074743822, | |
| "grad_norm": 2.536121368408203, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1914, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.09753959121047728, | |
| "grad_norm": 2.3796801567077637, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1848, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.09863554167351636, | |
| "grad_norm": 2.67964243888855, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1973, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.09973149213655542, | |
| "grad_norm": 3.160212755203247, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2472, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.1008274425995945, | |
| "grad_norm": 2.7035927772521973, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1844, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.10192339306263357, | |
| "grad_norm": 2.7725090980529785, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1262, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.10301934352567264, | |
| "grad_norm": 2.2705016136169434, | |
| "learning_rate": 0.0001, | |
| "loss": 1.182, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.10411529398871171, | |
| "grad_norm": 3.0717403888702393, | |
| "learning_rate": 0.0001, | |
| "loss": 1.137, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.10521124445175078, | |
| "grad_norm": 2.9270904064178467, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2556, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.10630719491478985, | |
| "grad_norm": 2.4564895629882812, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1812, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.10740314537782893, | |
| "grad_norm": 2.983851909637451, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1445, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.10849909584086799, | |
| "grad_norm": 2.772733688354492, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1968, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.10959504630390707, | |
| "grad_norm": 2.9768126010894775, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0942, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.10959504630390707, | |
| "eval_loss": 1.1446514129638672, | |
| "eval_runtime": 30634.8587, | |
| "eval_samples_per_second": 2.118, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 61.28519240053001, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.11069099676694613, | |
| "grad_norm": 2.806312322616577, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1924, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.11178694722998521, | |
| "grad_norm": 2.639443874359131, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0572, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.11288289769302427, | |
| "grad_norm": 2.2005367279052734, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1337, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.11397884815606335, | |
| "grad_norm": 2.4102020263671875, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1297, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.11507479861910241, | |
| "grad_norm": 3.410691976547241, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1354, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.11617074908214149, | |
| "grad_norm": 2.1337172985076904, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1725, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.11726669954518056, | |
| "grad_norm": 2.627319097518921, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1006, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.11836265000821963, | |
| "grad_norm": 2.6450726985931396, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0985, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.1194586004712587, | |
| "grad_norm": 2.3205084800720215, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1634, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.12055455093429777, | |
| "grad_norm": 2.51177978515625, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1697, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.12165050139733684, | |
| "grad_norm": 2.6632323265075684, | |
| "learning_rate": 0.0001, | |
| "loss": 1.071, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.12274645186037592, | |
| "grad_norm": 2.8322274684906006, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0983, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.12384240232341498, | |
| "grad_norm": 2.547708749771118, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0629, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.12493835278645406, | |
| "grad_norm": 2.6638150215148926, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1985, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.12603430324949313, | |
| "grad_norm": 2.980463743209839, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1885, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.1271302537125322, | |
| "grad_norm": 1.9924368858337402, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0971, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.12822620417557126, | |
| "grad_norm": 2.2847180366516113, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1149, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.12932215463861033, | |
| "grad_norm": 2.4860479831695557, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0927, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.13041810510164942, | |
| "grad_norm": 2.3988494873046875, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1918, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.13151405556468848, | |
| "grad_norm": 2.5361902713775635, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0603, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.13261000602772754, | |
| "grad_norm": 2.4060215950012207, | |
| "learning_rate": 0.0001, | |
| "loss": 1.056, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.1337059564907666, | |
| "grad_norm": 2.4094231128692627, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0787, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.1348019069538057, | |
| "grad_norm": 2.5207912921905518, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0901, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.13589785741684476, | |
| "grad_norm": 2.1340293884277344, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1691, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.13699380787988383, | |
| "grad_norm": 2.312554359436035, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9791, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.1380897583429229, | |
| "grad_norm": 2.2881298065185547, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9998, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.13918570880596198, | |
| "grad_norm": 2.2146573066711426, | |
| "learning_rate": 0.0001, | |
| "loss": 1.094, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.14028165926900105, | |
| "grad_norm": 2.3992650508880615, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0667, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.1413776097320401, | |
| "grad_norm": 2.7630209922790527, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1541, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.14247356019507917, | |
| "grad_norm": 2.9216675758361816, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0463, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.14356951065811827, | |
| "grad_norm": 2.366373062133789, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0557, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.14466546112115733, | |
| "grad_norm": 2.7161865234375, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1066, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.1457614115841964, | |
| "grad_norm": 2.046992778778076, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9786, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.14685736204723546, | |
| "grad_norm": 2.6320793628692627, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9564, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.14795331251027455, | |
| "grad_norm": 2.485445737838745, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0283, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.1490492629733136, | |
| "grad_norm": 2.267420768737793, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0092, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.15014521343635268, | |
| "grad_norm": 2.618067502975464, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0369, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.15124116389939174, | |
| "grad_norm": 2.502471685409546, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9982, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.15233711436243083, | |
| "grad_norm": 2.936964273452759, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1122, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.1534330648254699, | |
| "grad_norm": 2.5342159271240234, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0409, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.15452901528850896, | |
| "grad_norm": 2.88598895072937, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0259, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.15562496575154802, | |
| "grad_norm": 2.6327946186065674, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9829, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.1567209162145871, | |
| "grad_norm": 2.4873671531677246, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0472, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.15781686667762618, | |
| "grad_norm": 2.1543166637420654, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0157, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.15891281714066524, | |
| "grad_norm": 1.9687381982803345, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0465, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.1600087676037043, | |
| "grad_norm": 2.868544816970825, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9835, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.1611047180667434, | |
| "grad_norm": 2.3211984634399414, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1204, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.16220066852978246, | |
| "grad_norm": 2.631458282470703, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0175, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.16329661899282152, | |
| "grad_norm": 2.7994022369384766, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0828, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.1643925694558606, | |
| "grad_norm": 2.051626443862915, | |
| "learning_rate": 0.0001, | |
| "loss": 0.97, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.1643925694558606, | |
| "eval_loss": 1.0072325468063354, | |
| "eval_runtime": 30710.9249, | |
| "eval_samples_per_second": 2.113, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 55.08434535201816, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.16548851991889968, | |
| "grad_norm": 2.6088364124298096, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9803, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.16658447038193874, | |
| "grad_norm": 2.234034299850464, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0757, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.1676804208449778, | |
| "grad_norm": 2.3472328186035156, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9408, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.16877637130801687, | |
| "grad_norm": 2.5871200561523438, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9269, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.16987232177105596, | |
| "grad_norm": 2.0150465965270996, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0547, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.17096827223409503, | |
| "grad_norm": 2.5823395252227783, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0559, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.1720642226971341, | |
| "grad_norm": 2.8252885341644287, | |
| "learning_rate": 0.0001, | |
| "loss": 1.1219, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.17316017316017315, | |
| "grad_norm": 2.1086535453796387, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0089, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.17425612362321224, | |
| "grad_norm": 2.2288014888763428, | |
| "learning_rate": 0.0001, | |
| "loss": 1.136, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.1753520740862513, | |
| "grad_norm": 2.6622703075408936, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0395, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.17644802454929037, | |
| "grad_norm": 1.9478541612625122, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0658, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.17754397501232944, | |
| "grad_norm": 2.55828857421875, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9904, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.1786399254753685, | |
| "grad_norm": 2.533651828765869, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9733, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.1797358759384076, | |
| "grad_norm": 1.8745101690292358, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9903, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.18083182640144665, | |
| "grad_norm": 1.8459206819534302, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9095, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.18192777686448572, | |
| "grad_norm": 2.6654012203216553, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9854, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.18302372732752478, | |
| "grad_norm": 2.6444480419158936, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8857, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.18411967779056387, | |
| "grad_norm": 2.190462827682495, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9375, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.18521562825360294, | |
| "grad_norm": 2.8208882808685303, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9646, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.186311578716642, | |
| "grad_norm": 2.4978795051574707, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9724, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.18740752917968106, | |
| "grad_norm": 2.4202938079833984, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9659, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.18850347964272016, | |
| "grad_norm": 1.9026118516921997, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0321, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.18959943010575922, | |
| "grad_norm": 2.6031651496887207, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9622, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.19069538056879828, | |
| "grad_norm": 1.962509274482727, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0262, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.19179133103183735, | |
| "grad_norm": 2.794633626937866, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0626, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.19288728149487644, | |
| "grad_norm": 2.4276185035705566, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9961, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.1939832319579155, | |
| "grad_norm": 2.0747737884521484, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8945, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.19507918242095457, | |
| "grad_norm": 1.9151681661605835, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0664, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.19617513288399363, | |
| "grad_norm": 2.11547589302063, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9865, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.19727108334703272, | |
| "grad_norm": 2.359848737716675, | |
| "learning_rate": 0.0001, | |
| "loss": 0.95, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.19836703381007179, | |
| "grad_norm": 1.9854378700256348, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9992, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.19946298427311085, | |
| "grad_norm": 2.476423978805542, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9097, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.2005589347361499, | |
| "grad_norm": 2.420011281967163, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0167, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.201654885199189, | |
| "grad_norm": 2.12312388420105, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9298, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.20275083566222807, | |
| "grad_norm": 1.9679986238479614, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0064, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.20384678612526713, | |
| "grad_norm": 2.608135461807251, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9396, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.2049427365883062, | |
| "grad_norm": 2.542102098464966, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0868, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.2060386870513453, | |
| "grad_norm": 2.5252091884613037, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0417, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.20713463751438435, | |
| "grad_norm": 1.98774254322052, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9949, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.20823058797742341, | |
| "grad_norm": 1.9502965211868286, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9862, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.20932653844046248, | |
| "grad_norm": 2.2537944316864014, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9087, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.21042248890350157, | |
| "grad_norm": 2.2866523265838623, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0128, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.21151843936654063, | |
| "grad_norm": 2.2907001972198486, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9654, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.2126143898295797, | |
| "grad_norm": 2.5648560523986816, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0269, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.21371034029261876, | |
| "grad_norm": 2.198974847793579, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9823, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.21480629075565785, | |
| "grad_norm": 2.1045591831207275, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9139, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.21590224121869692, | |
| "grad_norm": 2.1462857723236084, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9406, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.21699819168173598, | |
| "grad_norm": 2.3216285705566406, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8597, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.21809414214477504, | |
| "grad_norm": 1.867150068283081, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9776, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.21919009260781414, | |
| "grad_norm": 2.3432791233062744, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9546, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.21919009260781414, | |
| "eval_loss": 0.9323587417602539, | |
| "eval_runtime": 30935.2713, | |
| "eval_samples_per_second": 2.098, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 63.836951720973865, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.2202860430708532, | |
| "grad_norm": 1.9426536560058594, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9291, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.22138199353389226, | |
| "grad_norm": 2.693723201751709, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9072, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.22247794399693133, | |
| "grad_norm": 2.237900972366333, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8571, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.22357389445997042, | |
| "grad_norm": 2.739129066467285, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9132, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.22466984492300948, | |
| "grad_norm": 1.886438012123108, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9646, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.22576579538604855, | |
| "grad_norm": 2.3505897521972656, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0479, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 0.2268617458490876, | |
| "grad_norm": 2.4302868843078613, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9956, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 0.2279576963121267, | |
| "grad_norm": 2.2747528553009033, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9621, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.22905364677516576, | |
| "grad_norm": 2.312248945236206, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9292, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 0.23014959723820483, | |
| "grad_norm": 2.0439066886901855, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8804, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.2312455477012439, | |
| "grad_norm": 2.615898609161377, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9302, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 0.23234149816428298, | |
| "grad_norm": 2.306796073913574, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0401, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 0.23343744862732205, | |
| "grad_norm": 2.4527432918548584, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9195, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 0.2345333990903611, | |
| "grad_norm": 1.8589290380477905, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9284, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 0.23562934955340017, | |
| "grad_norm": 1.8492025136947632, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8898, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.23672530001643927, | |
| "grad_norm": 2.574871063232422, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0026, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.23782125047947833, | |
| "grad_norm": 2.2600936889648438, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0738, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 0.2389172009425174, | |
| "grad_norm": 2.35066556930542, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8573, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 0.24001315140555646, | |
| "grad_norm": 2.165745496749878, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8989, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 0.24110910186859555, | |
| "grad_norm": 2.1494085788726807, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8292, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.2422050523316346, | |
| "grad_norm": 2.185359239578247, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8954, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 0.24330100279467368, | |
| "grad_norm": 2.193904161453247, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8944, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.24439695325771274, | |
| "grad_norm": 2.1101438999176025, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9059, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 0.24549290372075183, | |
| "grad_norm": 2.026642084121704, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8978, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.2465888541837909, | |
| "grad_norm": 2.0481228828430176, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8835, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.24768480464682996, | |
| "grad_norm": 2.201350688934326, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9519, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 0.24878075510986902, | |
| "grad_norm": 1.852100133895874, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8458, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 0.24987670557290811, | |
| "grad_norm": 2.1303794384002686, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9092, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.25097265603594715, | |
| "grad_norm": 2.2715415954589844, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8931, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 0.25206860649898627, | |
| "grad_norm": 2.091785192489624, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8645, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.25316455696202533, | |
| "grad_norm": 2.108103036880493, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8387, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 0.2542605074250644, | |
| "grad_norm": 2.083848237991333, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8315, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.25535645788810346, | |
| "grad_norm": 1.570475459098816, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9355, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 0.2564524083511425, | |
| "grad_norm": 1.90199875831604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8308, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.2575483588141816, | |
| "grad_norm": 2.1952812671661377, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8618, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.25864430927722065, | |
| "grad_norm": 2.0530431270599365, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7951, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 0.2597402597402597, | |
| "grad_norm": 2.202252149581909, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8858, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 0.26083621020329883, | |
| "grad_norm": 1.9541796445846558, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8466, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 0.2619321606663379, | |
| "grad_norm": 1.9440534114837646, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8488, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 0.26302811112937696, | |
| "grad_norm": 2.569821834564209, | |
| "learning_rate": 0.0001, | |
| "loss": 0.963, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.264124061592416, | |
| "grad_norm": 1.8896031379699707, | |
| "learning_rate": 0.0001, | |
| "loss": 0.837, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 0.2652200120554551, | |
| "grad_norm": 1.9390859603881836, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8855, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 0.26631596251849415, | |
| "grad_norm": 2.2261974811553955, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8901, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.2674119129815332, | |
| "grad_norm": 2.0486056804656982, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8073, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 0.2685078634445723, | |
| "grad_norm": 2.292015314102173, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9492, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.2696038139076114, | |
| "grad_norm": 2.0762240886688232, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8528, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.27069976437065046, | |
| "grad_norm": 1.870642066001892, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9482, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 0.27179571483368953, | |
| "grad_norm": 2.436768054962158, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9299, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.2728916652967286, | |
| "grad_norm": 2.505880832672119, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9259, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 0.27398761575976766, | |
| "grad_norm": 1.717252492904663, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8134, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.27398761575976766, | |
| "eval_loss": 0.8726964592933655, | |
| "eval_runtime": 30710.3822, | |
| "eval_samples_per_second": 2.113, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 52.213316533880224, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.2750835662228067, | |
| "grad_norm": 2.28765869140625, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0229, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 0.2761795166858458, | |
| "grad_norm": 2.2264580726623535, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8291, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 0.27727546714888485, | |
| "grad_norm": 1.9387757778167725, | |
| "learning_rate": 0.0001, | |
| "loss": 0.821, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 0.27837141761192397, | |
| "grad_norm": 2.8628933429718018, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9521, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 0.27946736807496303, | |
| "grad_norm": 2.2691447734832764, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8182, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.2805633185380021, | |
| "grad_norm": 1.9515260457992554, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9342, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.28165926900104116, | |
| "grad_norm": 2.1714837551116943, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9663, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 0.2827552194640802, | |
| "grad_norm": 2.0159664154052734, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8294, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 0.2838511699271193, | |
| "grad_norm": 2.024634599685669, | |
| "learning_rate": 0.0001, | |
| "loss": 0.896, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 0.28494712039015835, | |
| "grad_norm": 2.0035595893859863, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8446, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.2860430708531974, | |
| "grad_norm": 2.4142866134643555, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8835, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 0.28713902131623653, | |
| "grad_norm": 2.070338010787964, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8687, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 0.2882349717792756, | |
| "grad_norm": 1.9818578958511353, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8296, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 0.28933092224231466, | |
| "grad_norm": 1.8923412561416626, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8999, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.2904268727053537, | |
| "grad_norm": 2.200206995010376, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8662, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.2915228231683928, | |
| "grad_norm": 1.982446551322937, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8301, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 0.29261877363143185, | |
| "grad_norm": 1.934844732284546, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8219, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 0.2937147240944709, | |
| "grad_norm": 2.2790510654449463, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8666, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 0.29481067455751, | |
| "grad_norm": 1.771672248840332, | |
| "learning_rate": 0.0001, | |
| "loss": 0.843, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 0.2959066250205491, | |
| "grad_norm": 2.3459877967834473, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8516, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.29700257548358816, | |
| "grad_norm": 2.156458854675293, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8425, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 0.2980985259466272, | |
| "grad_norm": 1.9492950439453125, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8445, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.2991944764096663, | |
| "grad_norm": 2.1061997413635254, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8858, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 0.30029042687270535, | |
| "grad_norm": 2.3567299842834473, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8376, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 0.3013863773357444, | |
| "grad_norm": 2.1302335262298584, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8272, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.3024823277987835, | |
| "grad_norm": 2.2098424434661865, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8742, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 0.30357827826182254, | |
| "grad_norm": 1.7558562755584717, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8863, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 0.30467422872486166, | |
| "grad_norm": 1.8461397886276245, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8792, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 0.3057701791879007, | |
| "grad_norm": 2.0006344318389893, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8263, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 0.3068661296509398, | |
| "grad_norm": 1.6772565841674805, | |
| "learning_rate": 0.0001, | |
| "loss": 0.789, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.30796208011397885, | |
| "grad_norm": 1.9263228178024292, | |
| "learning_rate": 0.0001, | |
| "loss": 0.842, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 0.3090580305770179, | |
| "grad_norm": 1.8888592720031738, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8475, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 0.310153981040057, | |
| "grad_norm": 2.2354602813720703, | |
| "learning_rate": 0.0001, | |
| "loss": 1.0036, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 0.31124993150309604, | |
| "grad_norm": 1.9634332656860352, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8517, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 0.3123458819661351, | |
| "grad_norm": 2.348825216293335, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8731, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.3134418324291742, | |
| "grad_norm": 2.487741708755493, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8556, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 0.3145377828922133, | |
| "grad_norm": 1.999516248703003, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7969, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 0.31563373335525235, | |
| "grad_norm": 1.9654616117477417, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7843, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.3167296838182914, | |
| "grad_norm": 2.1070950031280518, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8399, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 0.3178256342813305, | |
| "grad_norm": 2.257129192352295, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8224, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.31892158474436955, | |
| "grad_norm": 1.8256118297576904, | |
| "learning_rate": 0.0001, | |
| "loss": 0.794, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 0.3200175352074086, | |
| "grad_norm": 1.8899625539779663, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8614, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 0.3211134856704477, | |
| "grad_norm": 2.221484661102295, | |
| "learning_rate": 0.0001, | |
| "loss": 0.765, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 0.3222094361334868, | |
| "grad_norm": 1.796877384185791, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8359, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 0.32330538659652586, | |
| "grad_norm": 1.7495447397232056, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8688, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.3244013370595649, | |
| "grad_norm": 2.136664628982544, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9163, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 0.325497287522604, | |
| "grad_norm": 1.8508238792419434, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7975, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 0.32659323798564305, | |
| "grad_norm": 2.144523859024048, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7749, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 0.3276891884486821, | |
| "grad_norm": 2.208815336227417, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8148, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 0.3287851389117212, | |
| "grad_norm": 2.0617401599884033, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8884, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.3287851389117212, | |
| "eval_loss": 0.8316722512245178, | |
| "eval_runtime": 30850.8589, | |
| "eval_samples_per_second": 2.103, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 45.9960352377659, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.32988108937476024, | |
| "grad_norm": 2.0406434535980225, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8504, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 0.33097703983779936, | |
| "grad_norm": 2.1899139881134033, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7782, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 0.3320729903008384, | |
| "grad_norm": 2.650421380996704, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7823, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 0.3331689407638775, | |
| "grad_norm": 2.085683584213257, | |
| "learning_rate": 0.0001, | |
| "loss": 0.754, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.33426489122691655, | |
| "grad_norm": 2.1783502101898193, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8819, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.3353608416899556, | |
| "grad_norm": 2.096208333969116, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8702, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 0.3364567921529947, | |
| "grad_norm": 2.005629062652588, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8827, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 0.33755274261603374, | |
| "grad_norm": 2.1545634269714355, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8496, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 0.3386486930790728, | |
| "grad_norm": 1.8190851211547852, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7622, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 0.3397446435421119, | |
| "grad_norm": 1.9555623531341553, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8338, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.340840594005151, | |
| "grad_norm": 1.8530341386795044, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8017, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 0.34193654446819005, | |
| "grad_norm": 1.8724114894866943, | |
| "learning_rate": 0.0001, | |
| "loss": 0.848, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 0.3430324949312291, | |
| "grad_norm": 1.8598796129226685, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8074, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 0.3441284453942682, | |
| "grad_norm": 2.1442923545837402, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8473, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 0.34522439585730724, | |
| "grad_norm": 2.3083174228668213, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9016, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.3463203463203463, | |
| "grad_norm": 1.8194735050201416, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8267, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 0.34741629678338537, | |
| "grad_norm": 2.063523054122925, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7841, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 0.3485122472464245, | |
| "grad_norm": 2.17594051361084, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8318, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 0.34960819770946355, | |
| "grad_norm": 1.665189504623413, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7983, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 0.3507041481725026, | |
| "grad_norm": 2.2596445083618164, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8421, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.3518000986355417, | |
| "grad_norm": 1.7096545696258545, | |
| "learning_rate": 0.0001, | |
| "loss": 0.889, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 0.35289604909858074, | |
| "grad_norm": 1.7475535869598389, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8006, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 0.3539919995616198, | |
| "grad_norm": 1.8176007270812988, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8632, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 0.35508795002465887, | |
| "grad_norm": 2.6806535720825195, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8427, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 0.35618390048769794, | |
| "grad_norm": 2.094172477722168, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7812, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.357279850950737, | |
| "grad_norm": 1.8341765403747559, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8051, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 0.3583758014137761, | |
| "grad_norm": 2.2341349124908447, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8001, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 0.3594717518768152, | |
| "grad_norm": 2.1017801761627197, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8142, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 0.36056770233985425, | |
| "grad_norm": 1.9903994798660278, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8117, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 0.3616636528028933, | |
| "grad_norm": 2.273465394973755, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8864, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.3627596032659324, | |
| "grad_norm": 2.0767428874969482, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7687, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 0.36385555372897144, | |
| "grad_norm": 2.559774398803711, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8181, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 0.3649515041920105, | |
| "grad_norm": 2.1393582820892334, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7936, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 0.36604745465504956, | |
| "grad_norm": 2.06675386428833, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8263, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 0.3671434051180887, | |
| "grad_norm": 1.7674784660339355, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7818, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.36823935558112775, | |
| "grad_norm": 1.765442132949829, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8335, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.3693353060441668, | |
| "grad_norm": 2.044288158416748, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8742, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 0.3704312565072059, | |
| "grad_norm": 1.9821726083755493, | |
| "learning_rate": 0.0001, | |
| "loss": 0.928, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 0.37152720697024494, | |
| "grad_norm": 2.0798370838165283, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7627, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 0.372623157433284, | |
| "grad_norm": 1.6817582845687866, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7985, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.37371910789632307, | |
| "grad_norm": 1.872247576713562, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8102, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 0.37481505835936213, | |
| "grad_norm": 1.7761516571044922, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8435, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 0.37591100882240125, | |
| "grad_norm": 1.739585518836975, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8706, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 0.3770069592854403, | |
| "grad_norm": 2.0503687858581543, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8354, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 0.3781029097484794, | |
| "grad_norm": 2.283393621444702, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7476, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.37919886021151844, | |
| "grad_norm": 1.801018238067627, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7817, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 0.3802948106745575, | |
| "grad_norm": 2.5343267917633057, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7628, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 0.38139076113759657, | |
| "grad_norm": 2.010507822036743, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7931, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 0.38248671160063563, | |
| "grad_norm": 1.7228796482086182, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7517, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 0.3835826620636747, | |
| "grad_norm": 1.967822551727295, | |
| "learning_rate": 0.0001, | |
| "loss": 0.804, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.3835826620636747, | |
| "eval_loss": 0.7978512644767761, | |
| "eval_runtime": 30977.7517, | |
| "eval_samples_per_second": 2.095, | |
| "eval_steps_per_second": 0.065, | |
| "eval_wer": 61.261910549759826, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.3846786125267138, | |
| "grad_norm": 1.9999229907989502, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7634, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 0.3857745629897529, | |
| "grad_norm": 1.956128716468811, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8102, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 0.38687051345279194, | |
| "grad_norm": 2.0134966373443604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7957, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 0.387966463915831, | |
| "grad_norm": 2.0373167991638184, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8251, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 0.38906241437887007, | |
| "grad_norm": 1.7772964239120483, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8128, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.39015836484190913, | |
| "grad_norm": 1.7618379592895508, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8345, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 0.3912543153049482, | |
| "grad_norm": 2.181671380996704, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8345, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 0.39235026576798726, | |
| "grad_norm": 1.8794726133346558, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7615, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 0.3934462162310264, | |
| "grad_norm": 1.9297798871994019, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7618, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 0.39454216669406544, | |
| "grad_norm": 1.9441471099853516, | |
| "learning_rate": 0.0001, | |
| "loss": 0.859, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.3956381171571045, | |
| "grad_norm": 2.2561404705047607, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7877, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 0.39673406762014357, | |
| "grad_norm": 1.8441416025161743, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7734, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 0.39783001808318263, | |
| "grad_norm": 1.686120867729187, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7066, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 0.3989259685462217, | |
| "grad_norm": 1.9456263780593872, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7469, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 0.40002191900926076, | |
| "grad_norm": 1.9112725257873535, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7607, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.4011178694722998, | |
| "grad_norm": 2.5668513774871826, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7859, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 0.40221381993533895, | |
| "grad_norm": 1.9502942562103271, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7607, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 0.403309770398378, | |
| "grad_norm": 1.6973525285720825, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8313, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 0.4044057208614171, | |
| "grad_norm": 2.3962297439575195, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7806, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 0.40550167132445614, | |
| "grad_norm": 1.887536883354187, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7524, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.4065976217874952, | |
| "grad_norm": 1.999687910079956, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7349, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 0.40769357225053426, | |
| "grad_norm": 1.7444576025009155, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8156, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 0.40878952271357333, | |
| "grad_norm": 1.7175132036209106, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7419, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 0.4098854731766124, | |
| "grad_norm": 2.23638653755188, | |
| "learning_rate": 0.0001, | |
| "loss": 0.666, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 0.4109814236396515, | |
| "grad_norm": 2.024102210998535, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7541, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.4120773741026906, | |
| "grad_norm": 2.042541265487671, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7915, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 0.41317332456572964, | |
| "grad_norm": 1.9140897989273071, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8712, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 0.4142692750287687, | |
| "grad_norm": 1.8435416221618652, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8241, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 0.41536522549180777, | |
| "grad_norm": 2.027944803237915, | |
| "learning_rate": 0.0001, | |
| "loss": 0.9422, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 0.41646117595484683, | |
| "grad_norm": 2.07381534576416, | |
| "learning_rate": 0.0001, | |
| "loss": 0.812, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.4175571264178859, | |
| "grad_norm": 1.9762136936187744, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7852, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 0.41865307688092496, | |
| "grad_norm": 1.8222426176071167, | |
| "learning_rate": 0.0001, | |
| "loss": 0.752, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 0.4197490273439641, | |
| "grad_norm": 2.0519089698791504, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8031, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 0.42084497780700314, | |
| "grad_norm": 1.8777110576629639, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8173, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.4219409282700422, | |
| "grad_norm": 2.323411703109741, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8479, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.42303687873308127, | |
| "grad_norm": 1.6403400897979736, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7567, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 0.42413282919612033, | |
| "grad_norm": 1.6627925634384155, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7734, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 0.4252287796591594, | |
| "grad_norm": 1.8771709203720093, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7652, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 0.42632473012219846, | |
| "grad_norm": 1.9806597232818604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7699, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 0.4274206805852375, | |
| "grad_norm": 2.1376988887786865, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7825, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.42851663104827664, | |
| "grad_norm": 1.5566449165344238, | |
| "learning_rate": 0.0001, | |
| "loss": 0.704, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 0.4296125815113157, | |
| "grad_norm": 2.1835947036743164, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8101, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 0.43070853197435477, | |
| "grad_norm": 2.055119037628174, | |
| "learning_rate": 0.0001, | |
| "loss": 0.703, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 0.43180448243739383, | |
| "grad_norm": 1.9324967861175537, | |
| "learning_rate": 0.0001, | |
| "loss": 0.81, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 0.4329004329004329, | |
| "grad_norm": 2.1087846755981445, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7676, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.43399638336347196, | |
| "grad_norm": 1.8521897792816162, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7546, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 0.435092333826511, | |
| "grad_norm": 2.145947217941284, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7992, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 0.4361882842895501, | |
| "grad_norm": 1.7739931344985962, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7133, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 0.4372842347525892, | |
| "grad_norm": 1.6032921075820923, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8207, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 0.43838018521562827, | |
| "grad_norm": 2.1895668506622314, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7638, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.43838018521562827, | |
| "eval_loss": 0.770411491394043, | |
| "eval_runtime": 30675.7059, | |
| "eval_samples_per_second": 2.115, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 43.10069742838263, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.43947613567866733, | |
| "grad_norm": 1.9759962558746338, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7792, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 0.4405720861417064, | |
| "grad_norm": 1.845012903213501, | |
| "learning_rate": 0.0001, | |
| "loss": 0.847, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 0.44166803660474546, | |
| "grad_norm": 1.9666188955307007, | |
| "learning_rate": 0.0001, | |
| "loss": 0.767, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 0.4427639870677845, | |
| "grad_norm": 2.1448235511779785, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7924, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 0.4438599375308236, | |
| "grad_norm": 1.9017919301986694, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7239, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.44495588799386265, | |
| "grad_norm": 1.8005828857421875, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7202, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 0.4460518384569018, | |
| "grad_norm": 1.7341022491455078, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7045, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 0.44714778891994084, | |
| "grad_norm": 2.094618320465088, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8067, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 0.4482437393829799, | |
| "grad_norm": 2.0414187908172607, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6888, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 0.44933968984601896, | |
| "grad_norm": 1.8842118978500366, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7125, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.450435640309058, | |
| "grad_norm": 1.9878696203231812, | |
| "learning_rate": 0.0001, | |
| "loss": 0.723, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 0.4515315907720971, | |
| "grad_norm": 1.94351065158844, | |
| "learning_rate": 0.0001, | |
| "loss": 0.727, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 0.45262754123513615, | |
| "grad_norm": 1.900718331336975, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7306, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 0.4537234916981752, | |
| "grad_norm": 2.5974204540252686, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7968, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 0.45481944216121434, | |
| "grad_norm": 1.9214075803756714, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7767, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.4559153926242534, | |
| "grad_norm": 2.6079931259155273, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7787, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 0.45701134308729247, | |
| "grad_norm": 1.8398691415786743, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7941, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 0.45810729355033153, | |
| "grad_norm": 1.740376591682434, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7714, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 0.4592032440133706, | |
| "grad_norm": 2.109416961669922, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8015, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 0.46029919447640966, | |
| "grad_norm": 1.9565001726150513, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7473, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.4613951449394487, | |
| "grad_norm": 1.88534414768219, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7828, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 0.4624910954024878, | |
| "grad_norm": 1.7713934183120728, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7289, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 0.4635870458655269, | |
| "grad_norm": 1.9173312187194824, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7478, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 0.46468299632856597, | |
| "grad_norm": 1.6866717338562012, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8235, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 0.46577894679160503, | |
| "grad_norm": 1.6713476181030273, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7216, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.4668748972546441, | |
| "grad_norm": 1.9601606130599976, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6994, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 0.46797084771768316, | |
| "grad_norm": 1.7472949028015137, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7694, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 0.4690667981807222, | |
| "grad_norm": 1.8540037870407104, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7253, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 0.4701627486437613, | |
| "grad_norm": 2.0671746730804443, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7514, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 0.47125869910680035, | |
| "grad_norm": 1.900918960571289, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7871, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.47235464956983947, | |
| "grad_norm": 1.7465757131576538, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8009, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 0.47345060003287853, | |
| "grad_norm": 2.3400652408599854, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7741, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 0.4745465504959176, | |
| "grad_norm": 2.1384716033935547, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7577, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 0.47564250095895666, | |
| "grad_norm": 2.7113006114959717, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6968, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 0.4767384514219957, | |
| "grad_norm": 1.6666728258132935, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7307, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.4778344018850348, | |
| "grad_norm": 1.8394851684570312, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7353, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 0.47893035234807385, | |
| "grad_norm": 2.0569512844085693, | |
| "learning_rate": 0.0001, | |
| "loss": 0.814, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 0.4800263028111129, | |
| "grad_norm": 1.6457910537719727, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7521, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 0.48112225327415203, | |
| "grad_norm": 2.010711908340454, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7101, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 0.4822182037371911, | |
| "grad_norm": 2.422718048095703, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7867, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.48331415420023016, | |
| "grad_norm": 1.5170652866363525, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8042, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 0.4844101046632692, | |
| "grad_norm": 1.9751352071762085, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7408, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 0.4855060551263083, | |
| "grad_norm": 1.8477592468261719, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7675, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 0.48660200558934735, | |
| "grad_norm": 1.9999114274978638, | |
| "learning_rate": 0.0001, | |
| "loss": 0.745, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 0.4876979560523864, | |
| "grad_norm": 1.7456104755401611, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7713, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.4887939065154255, | |
| "grad_norm": 1.9687026739120483, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7349, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 0.4898898569784646, | |
| "grad_norm": 1.8585296869277954, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7369, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 0.49098580744150366, | |
| "grad_norm": 2.7875003814697266, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7002, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 0.4920817579045427, | |
| "grad_norm": 2.01347017288208, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7598, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 0.4931777083675818, | |
| "grad_norm": 1.8863261938095093, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7617, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.4931777083675818, | |
| "eval_loss": 0.7443549036979675, | |
| "eval_runtime": 31204.044, | |
| "eval_samples_per_second": 2.079, | |
| "eval_steps_per_second": 0.065, | |
| "eval_wer": 44.168238762227254, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.49427365883062085, | |
| "grad_norm": 2.0402464866638184, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8118, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 0.4953696092936599, | |
| "grad_norm": 2.072380304336548, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6703, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 0.496465559756699, | |
| "grad_norm": 1.9627012014389038, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7547, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 0.49756151021973805, | |
| "grad_norm": 1.904860496520996, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8141, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 0.49865746068277716, | |
| "grad_norm": 2.153672933578491, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8167, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.49975341114581623, | |
| "grad_norm": 2.0599303245544434, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8632, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 0.5008493616088553, | |
| "grad_norm": 1.9562146663665771, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7477, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 0.5019453120718943, | |
| "grad_norm": 2.086508274078369, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7973, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 0.5030412625349334, | |
| "grad_norm": 1.9192993640899658, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8359, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 0.5041372129979725, | |
| "grad_norm": 1.9085866212844849, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7942, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.5052331634610115, | |
| "grad_norm": 1.901637315750122, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7608, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 2.145914316177368, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6919, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 0.5074250643870897, | |
| "grad_norm": 1.9005271196365356, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7506, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 0.5085210148501288, | |
| "grad_norm": 1.6468952894210815, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7843, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 0.5096169653131678, | |
| "grad_norm": 1.7703279256820679, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7192, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.5107129157762069, | |
| "grad_norm": 2.0094175338745117, | |
| "learning_rate": 0.0001, | |
| "loss": 0.847, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 0.511808866239246, | |
| "grad_norm": 2.0970561504364014, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7679, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 0.512904816702285, | |
| "grad_norm": 1.757664680480957, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7391, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 0.5140007671653242, | |
| "grad_norm": 1.8297368288040161, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8382, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 0.5150967176283632, | |
| "grad_norm": 1.9832725524902344, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7226, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.5161926680914023, | |
| "grad_norm": 1.7083086967468262, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7798, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 0.5172886185544413, | |
| "grad_norm": 1.7105575799942017, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7118, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 0.5183845690174804, | |
| "grad_norm": 1.547608494758606, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7259, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 0.5194805194805194, | |
| "grad_norm": 2.0215799808502197, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7648, | |
| "step": 9480 | |
| }, | |
| { | |
| "epoch": 0.5205764699435586, | |
| "grad_norm": 1.5999863147735596, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7863, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.5216724204065977, | |
| "grad_norm": 2.0813591480255127, | |
| "learning_rate": 0.0001, | |
| "loss": 0.758, | |
| "step": 9520 | |
| }, | |
| { | |
| "epoch": 0.5227683708696367, | |
| "grad_norm": 1.6513686180114746, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7735, | |
| "step": 9540 | |
| }, | |
| { | |
| "epoch": 0.5238643213326758, | |
| "grad_norm": 1.49434232711792, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6547, | |
| "step": 9560 | |
| }, | |
| { | |
| "epoch": 0.5249602717957148, | |
| "grad_norm": 1.8316184282302856, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7428, | |
| "step": 9580 | |
| }, | |
| { | |
| "epoch": 0.5260562222587539, | |
| "grad_norm": 2.0041682720184326, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7058, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.5271521727217929, | |
| "grad_norm": 1.9916651248931885, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7049, | |
| "step": 9620 | |
| }, | |
| { | |
| "epoch": 0.528248123184832, | |
| "grad_norm": 1.8289718627929688, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7179, | |
| "step": 9640 | |
| }, | |
| { | |
| "epoch": 0.5293440736478712, | |
| "grad_norm": 1.7447452545166016, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7432, | |
| "step": 9660 | |
| }, | |
| { | |
| "epoch": 0.5304400241109102, | |
| "grad_norm": 2.375234365463257, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6676, | |
| "step": 9680 | |
| }, | |
| { | |
| "epoch": 0.5315359745739493, | |
| "grad_norm": 1.683435320854187, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7269, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.5326319250369883, | |
| "grad_norm": 1.6535717248916626, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7315, | |
| "step": 9720 | |
| }, | |
| { | |
| "epoch": 0.5337278755000274, | |
| "grad_norm": 1.5276830196380615, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7382, | |
| "step": 9740 | |
| }, | |
| { | |
| "epoch": 0.5348238259630664, | |
| "grad_norm": 1.8443965911865234, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7471, | |
| "step": 9760 | |
| }, | |
| { | |
| "epoch": 0.5359197764261056, | |
| "grad_norm": 2.0346148014068604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7268, | |
| "step": 9780 | |
| }, | |
| { | |
| "epoch": 0.5370157268891446, | |
| "grad_norm": 1.750613808631897, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8444, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.5381116773521837, | |
| "grad_norm": 1.9546024799346924, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6968, | |
| "step": 9820 | |
| }, | |
| { | |
| "epoch": 0.5392076278152228, | |
| "grad_norm": 1.6618010997772217, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7222, | |
| "step": 9840 | |
| }, | |
| { | |
| "epoch": 0.5403035782782618, | |
| "grad_norm": 1.6404950618743896, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6896, | |
| "step": 9860 | |
| }, | |
| { | |
| "epoch": 0.5413995287413009, | |
| "grad_norm": 1.7741234302520752, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7412, | |
| "step": 9880 | |
| }, | |
| { | |
| "epoch": 0.5424954792043399, | |
| "grad_norm": 1.8278882503509521, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7385, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.5435914296673791, | |
| "grad_norm": 1.6102566719055176, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7461, | |
| "step": 9920 | |
| }, | |
| { | |
| "epoch": 0.5446873801304181, | |
| "grad_norm": 1.7899205684661865, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6349, | |
| "step": 9940 | |
| }, | |
| { | |
| "epoch": 0.5457833305934572, | |
| "grad_norm": 1.9663938283920288, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8028, | |
| "step": 9960 | |
| }, | |
| { | |
| "epoch": 0.5468792810564962, | |
| "grad_norm": 1.841476559638977, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7503, | |
| "step": 9980 | |
| }, | |
| { | |
| "epoch": 0.5479752315195353, | |
| "grad_norm": 1.9106056690216064, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7097, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.5479752315195353, | |
| "eval_loss": 0.7265371084213257, | |
| "eval_runtime": 30536.1813, | |
| "eval_samples_per_second": 2.125, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 42.517110448415295, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.5490711819825744, | |
| "grad_norm": 1.680649995803833, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6445, | |
| "step": 10020 | |
| }, | |
| { | |
| "epoch": 0.5501671324456134, | |
| "grad_norm": 2.079050064086914, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7648, | |
| "step": 10040 | |
| }, | |
| { | |
| "epoch": 0.5512630829086526, | |
| "grad_norm": 1.4419294595718384, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6953, | |
| "step": 10060 | |
| }, | |
| { | |
| "epoch": 0.5523590333716916, | |
| "grad_norm": 1.9906927347183228, | |
| "learning_rate": 0.0001, | |
| "loss": 0.749, | |
| "step": 10080 | |
| }, | |
| { | |
| "epoch": 0.5534549838347307, | |
| "grad_norm": 1.7384852170944214, | |
| "learning_rate": 0.0001, | |
| "loss": 0.745, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.5545509342977697, | |
| "grad_norm": 1.7342479228973389, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7687, | |
| "step": 10120 | |
| }, | |
| { | |
| "epoch": 0.5556468847608088, | |
| "grad_norm": 1.887969970703125, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7662, | |
| "step": 10140 | |
| }, | |
| { | |
| "epoch": 0.5567428352238479, | |
| "grad_norm": 1.6345020532608032, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7843, | |
| "step": 10160 | |
| }, | |
| { | |
| "epoch": 0.5578387856868869, | |
| "grad_norm": 1.5596251487731934, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6983, | |
| "step": 10180 | |
| }, | |
| { | |
| "epoch": 0.5589347361499261, | |
| "grad_norm": 1.6423192024230957, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6622, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.5600306866129651, | |
| "grad_norm": 1.7268792390823364, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8409, | |
| "step": 10220 | |
| }, | |
| { | |
| "epoch": 0.5611266370760042, | |
| "grad_norm": 1.6870604753494263, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7801, | |
| "step": 10240 | |
| }, | |
| { | |
| "epoch": 0.5622225875390432, | |
| "grad_norm": 1.5945113897323608, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6695, | |
| "step": 10260 | |
| }, | |
| { | |
| "epoch": 0.5633185380020823, | |
| "grad_norm": 1.7995914220809937, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7088, | |
| "step": 10280 | |
| }, | |
| { | |
| "epoch": 0.5644144884651213, | |
| "grad_norm": 1.8924362659454346, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7621, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.5655104389281604, | |
| "grad_norm": 1.5099490880966187, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6923, | |
| "step": 10320 | |
| }, | |
| { | |
| "epoch": 0.5666063893911996, | |
| "grad_norm": 1.481195092201233, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6801, | |
| "step": 10340 | |
| }, | |
| { | |
| "epoch": 0.5677023398542386, | |
| "grad_norm": 1.9247808456420898, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7247, | |
| "step": 10360 | |
| }, | |
| { | |
| "epoch": 0.5687982903172777, | |
| "grad_norm": 1.721666693687439, | |
| "learning_rate": 0.0001, | |
| "loss": 0.85, | |
| "step": 10380 | |
| }, | |
| { | |
| "epoch": 0.5698942407803167, | |
| "grad_norm": 1.981312870979309, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6894, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.5709901912433558, | |
| "grad_norm": 1.825363039970398, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7017, | |
| "step": 10420 | |
| }, | |
| { | |
| "epoch": 0.5720861417063948, | |
| "grad_norm": 2.021385669708252, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7996, | |
| "step": 10440 | |
| }, | |
| { | |
| "epoch": 0.5731820921694339, | |
| "grad_norm": 1.9287372827529907, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7387, | |
| "step": 10460 | |
| }, | |
| { | |
| "epoch": 0.5742780426324731, | |
| "grad_norm": 2.0109355449676514, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7359, | |
| "step": 10480 | |
| }, | |
| { | |
| "epoch": 0.5753739930955121, | |
| "grad_norm": 1.7715758085250854, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7126, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.5764699435585512, | |
| "grad_norm": 1.5866303443908691, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6808, | |
| "step": 10520 | |
| }, | |
| { | |
| "epoch": 0.5775658940215902, | |
| "grad_norm": 1.3831912279129028, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7251, | |
| "step": 10540 | |
| }, | |
| { | |
| "epoch": 0.5786618444846293, | |
| "grad_norm": 1.603388786315918, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6497, | |
| "step": 10560 | |
| }, | |
| { | |
| "epoch": 0.5797577949476683, | |
| "grad_norm": 1.8507051467895508, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7247, | |
| "step": 10580 | |
| }, | |
| { | |
| "epoch": 0.5808537454107074, | |
| "grad_norm": 2.240337610244751, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7879, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.5819496958737465, | |
| "grad_norm": 1.858344316482544, | |
| "learning_rate": 0.0001, | |
| "loss": 0.647, | |
| "step": 10620 | |
| }, | |
| { | |
| "epoch": 0.5830456463367856, | |
| "grad_norm": 1.840640664100647, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6924, | |
| "step": 10640 | |
| }, | |
| { | |
| "epoch": 0.5841415967998247, | |
| "grad_norm": 2.0423295497894287, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6762, | |
| "step": 10660 | |
| }, | |
| { | |
| "epoch": 0.5852375472628637, | |
| "grad_norm": 1.7426679134368896, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7824, | |
| "step": 10680 | |
| }, | |
| { | |
| "epoch": 0.5863334977259028, | |
| "grad_norm": 1.5974029302597046, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6874, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.5874294481889418, | |
| "grad_norm": 1.6082810163497925, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6916, | |
| "step": 10720 | |
| }, | |
| { | |
| "epoch": 0.588525398651981, | |
| "grad_norm": 1.6124242544174194, | |
| "learning_rate": 0.0001, | |
| "loss": 0.676, | |
| "step": 10740 | |
| }, | |
| { | |
| "epoch": 0.58962134911502, | |
| "grad_norm": 1.9140983819961548, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6281, | |
| "step": 10760 | |
| }, | |
| { | |
| "epoch": 0.5907172995780591, | |
| "grad_norm": 1.708742618560791, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7245, | |
| "step": 10780 | |
| }, | |
| { | |
| "epoch": 0.5918132500410982, | |
| "grad_norm": 2.36368989944458, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6934, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.5929092005041372, | |
| "grad_norm": 1.9806820154190063, | |
| "learning_rate": 0.0001, | |
| "loss": 0.677, | |
| "step": 10820 | |
| }, | |
| { | |
| "epoch": 0.5940051509671763, | |
| "grad_norm": 1.893801212310791, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7629, | |
| "step": 10840 | |
| }, | |
| { | |
| "epoch": 0.5951011014302153, | |
| "grad_norm": 1.917204737663269, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7836, | |
| "step": 10860 | |
| }, | |
| { | |
| "epoch": 0.5961970518932544, | |
| "grad_norm": 1.5599673986434937, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7132, | |
| "step": 10880 | |
| }, | |
| { | |
| "epoch": 0.5972930023562935, | |
| "grad_norm": 1.9569772481918335, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7466, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.5983889528193326, | |
| "grad_norm": 1.8709198236465454, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6456, | |
| "step": 10920 | |
| }, | |
| { | |
| "epoch": 0.5994849032823716, | |
| "grad_norm": 1.8249480724334717, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7159, | |
| "step": 10940 | |
| }, | |
| { | |
| "epoch": 0.6005808537454107, | |
| "grad_norm": 1.7063779830932617, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7093, | |
| "step": 10960 | |
| }, | |
| { | |
| "epoch": 0.6016768042084498, | |
| "grad_norm": 1.681219220161438, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6995, | |
| "step": 10980 | |
| }, | |
| { | |
| "epoch": 0.6027727546714888, | |
| "grad_norm": 1.640663504600525, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7045, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.6027727546714888, | |
| "eval_loss": 0.7112395763397217, | |
| "eval_runtime": 30901.8839, | |
| "eval_samples_per_second": 2.1, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 47.33354332649714, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.603868705134528, | |
| "grad_norm": 1.7204805612564087, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7217, | |
| "step": 11020 | |
| }, | |
| { | |
| "epoch": 0.604964655597567, | |
| "grad_norm": 1.507012128829956, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7482, | |
| "step": 11040 | |
| }, | |
| { | |
| "epoch": 0.6060606060606061, | |
| "grad_norm": 1.7084465026855469, | |
| "learning_rate": 0.0001, | |
| "loss": 0.645, | |
| "step": 11060 | |
| }, | |
| { | |
| "epoch": 0.6071565565236451, | |
| "grad_norm": 1.73207426071167, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7687, | |
| "step": 11080 | |
| }, | |
| { | |
| "epoch": 0.6082525069866842, | |
| "grad_norm": 2.2146365642547607, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7771, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.6093484574497233, | |
| "grad_norm": 1.6794184446334839, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6613, | |
| "step": 11120 | |
| }, | |
| { | |
| "epoch": 0.6104444079127623, | |
| "grad_norm": 1.8254398107528687, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6787, | |
| "step": 11140 | |
| }, | |
| { | |
| "epoch": 0.6115403583758015, | |
| "grad_norm": 1.8397271633148193, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7119, | |
| "step": 11160 | |
| }, | |
| { | |
| "epoch": 0.6126363088388405, | |
| "grad_norm": 1.8676248788833618, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7294, | |
| "step": 11180 | |
| }, | |
| { | |
| "epoch": 0.6137322593018796, | |
| "grad_norm": 1.4971026182174683, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6312, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.6148282097649186, | |
| "grad_norm": 1.8128615617752075, | |
| "learning_rate": 0.0001, | |
| "loss": 0.653, | |
| "step": 11220 | |
| }, | |
| { | |
| "epoch": 0.6159241602279577, | |
| "grad_norm": 1.426620364189148, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7087, | |
| "step": 11240 | |
| }, | |
| { | |
| "epoch": 0.6170201106909967, | |
| "grad_norm": 1.4840887784957886, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6665, | |
| "step": 11260 | |
| }, | |
| { | |
| "epoch": 0.6181160611540358, | |
| "grad_norm": 1.7882121801376343, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7236, | |
| "step": 11280 | |
| }, | |
| { | |
| "epoch": 0.619212011617075, | |
| "grad_norm": 1.8195546865463257, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7998, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.620307962080114, | |
| "grad_norm": 1.9482252597808838, | |
| "learning_rate": 0.0001, | |
| "loss": 0.751, | |
| "step": 11320 | |
| }, | |
| { | |
| "epoch": 0.6214039125431531, | |
| "grad_norm": 2.1224782466888428, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7518, | |
| "step": 11340 | |
| }, | |
| { | |
| "epoch": 0.6224998630061921, | |
| "grad_norm": 1.811909556388855, | |
| "learning_rate": 0.0001, | |
| "loss": 0.679, | |
| "step": 11360 | |
| }, | |
| { | |
| "epoch": 0.6235958134692312, | |
| "grad_norm": 2.0843353271484375, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7381, | |
| "step": 11380 | |
| }, | |
| { | |
| "epoch": 0.6246917639322702, | |
| "grad_norm": 1.5517933368682861, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7318, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.6257877143953093, | |
| "grad_norm": 1.3482716083526611, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6999, | |
| "step": 11420 | |
| }, | |
| { | |
| "epoch": 0.6268836648583485, | |
| "grad_norm": 1.548904299736023, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8772, | |
| "step": 11440 | |
| }, | |
| { | |
| "epoch": 0.6279796153213875, | |
| "grad_norm": 1.553775429725647, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6479, | |
| "step": 11460 | |
| }, | |
| { | |
| "epoch": 0.6290755657844266, | |
| "grad_norm": 2.0762696266174316, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6457, | |
| "step": 11480 | |
| }, | |
| { | |
| "epoch": 0.6301715162474656, | |
| "grad_norm": 1.9620105028152466, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7098, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.6312674667105047, | |
| "grad_norm": 1.382176399230957, | |
| "learning_rate": 0.0001, | |
| "loss": 0.686, | |
| "step": 11520 | |
| }, | |
| { | |
| "epoch": 0.6323634171735437, | |
| "grad_norm": 1.9390108585357666, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6882, | |
| "step": 11540 | |
| }, | |
| { | |
| "epoch": 0.6334593676365828, | |
| "grad_norm": 1.7750768661499023, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6942, | |
| "step": 11560 | |
| }, | |
| { | |
| "epoch": 0.6345553180996218, | |
| "grad_norm": 1.8459293842315674, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7118, | |
| "step": 11580 | |
| }, | |
| { | |
| "epoch": 0.635651268562661, | |
| "grad_norm": 1.8210084438323975, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7166, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.6367472190257001, | |
| "grad_norm": 1.7728508710861206, | |
| "learning_rate": 0.0001, | |
| "loss": 0.692, | |
| "step": 11620 | |
| }, | |
| { | |
| "epoch": 0.6378431694887391, | |
| "grad_norm": 1.7886627912521362, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7185, | |
| "step": 11640 | |
| }, | |
| { | |
| "epoch": 0.6389391199517782, | |
| "grad_norm": 1.895150065422058, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7174, | |
| "step": 11660 | |
| }, | |
| { | |
| "epoch": 0.6400350704148172, | |
| "grad_norm": 1.8740530014038086, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6893, | |
| "step": 11680 | |
| }, | |
| { | |
| "epoch": 0.6411310208778563, | |
| "grad_norm": 1.6588834524154663, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7073, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.6422269713408953, | |
| "grad_norm": 1.9573453664779663, | |
| "learning_rate": 0.0001, | |
| "loss": 0.671, | |
| "step": 11720 | |
| }, | |
| { | |
| "epoch": 0.6433229218039345, | |
| "grad_norm": 1.7064661979675293, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7401, | |
| "step": 11740 | |
| }, | |
| { | |
| "epoch": 0.6444188722669736, | |
| "grad_norm": 1.8850706815719604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7397, | |
| "step": 11760 | |
| }, | |
| { | |
| "epoch": 0.6455148227300126, | |
| "grad_norm": 1.7744836807250977, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7204, | |
| "step": 11780 | |
| }, | |
| { | |
| "epoch": 0.6466107731930517, | |
| "grad_norm": 1.5768756866455078, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7868, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.6477067236560907, | |
| "grad_norm": 2.0770552158355713, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7434, | |
| "step": 11820 | |
| }, | |
| { | |
| "epoch": 0.6488026741191298, | |
| "grad_norm": 2.0797810554504395, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7342, | |
| "step": 11840 | |
| }, | |
| { | |
| "epoch": 0.6498986245821688, | |
| "grad_norm": 1.8984261751174927, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6642, | |
| "step": 11860 | |
| }, | |
| { | |
| "epoch": 0.650994575045208, | |
| "grad_norm": 2.00124192237854, | |
| "learning_rate": 0.0001, | |
| "loss": 0.749, | |
| "step": 11880 | |
| }, | |
| { | |
| "epoch": 0.652090525508247, | |
| "grad_norm": 1.5575506687164307, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8315, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.6531864759712861, | |
| "grad_norm": 2.6183197498321533, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7533, | |
| "step": 11920 | |
| }, | |
| { | |
| "epoch": 0.6542824264343252, | |
| "grad_norm": 1.7211464643478394, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7073, | |
| "step": 11940 | |
| }, | |
| { | |
| "epoch": 0.6553783768973642, | |
| "grad_norm": 1.9105095863342285, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6526, | |
| "step": 11960 | |
| }, | |
| { | |
| "epoch": 0.6564743273604033, | |
| "grad_norm": 1.9578741788864136, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6746, | |
| "step": 11980 | |
| }, | |
| { | |
| "epoch": 0.6575702778234424, | |
| "grad_norm": 1.8473331928253174, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6808, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.6575702778234424, | |
| "eval_loss": 0.6952778100967407, | |
| "eval_runtime": 30743.2328, | |
| "eval_samples_per_second": 2.111, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 51.05538683822195, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.6586662282864815, | |
| "grad_norm": 1.7486096620559692, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7417, | |
| "step": 12020 | |
| }, | |
| { | |
| "epoch": 0.6597621787495205, | |
| "grad_norm": 1.6540303230285645, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7118, | |
| "step": 12040 | |
| }, | |
| { | |
| "epoch": 0.6608581292125596, | |
| "grad_norm": 1.89935302734375, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6315, | |
| "step": 12060 | |
| }, | |
| { | |
| "epoch": 0.6619540796755987, | |
| "grad_norm": 1.8266342878341675, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7434, | |
| "step": 12080 | |
| }, | |
| { | |
| "epoch": 0.6630500301386377, | |
| "grad_norm": 1.8254984617233276, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7832, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.6641459806016768, | |
| "grad_norm": 2.0791878700256348, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6694, | |
| "step": 12120 | |
| }, | |
| { | |
| "epoch": 0.6652419310647159, | |
| "grad_norm": 1.8277227878570557, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6686, | |
| "step": 12140 | |
| }, | |
| { | |
| "epoch": 0.666337881527755, | |
| "grad_norm": 1.697810411453247, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6895, | |
| "step": 12160 | |
| }, | |
| { | |
| "epoch": 0.667433831990794, | |
| "grad_norm": 1.6084686517715454, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7431, | |
| "step": 12180 | |
| }, | |
| { | |
| "epoch": 0.6685297824538331, | |
| "grad_norm": 1.7437437772750854, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6851, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.6696257329168721, | |
| "grad_norm": 1.849237322807312, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6926, | |
| "step": 12220 | |
| }, | |
| { | |
| "epoch": 0.6707216833799112, | |
| "grad_norm": 1.8398326635360718, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7282, | |
| "step": 12240 | |
| }, | |
| { | |
| "epoch": 0.6718176338429503, | |
| "grad_norm": 2.056136131286621, | |
| "learning_rate": 0.0001, | |
| "loss": 0.76, | |
| "step": 12260 | |
| }, | |
| { | |
| "epoch": 0.6729135843059894, | |
| "grad_norm": 1.8255378007888794, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7155, | |
| "step": 12280 | |
| }, | |
| { | |
| "epoch": 0.6740095347690285, | |
| "grad_norm": 1.6555898189544678, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6333, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.6751054852320675, | |
| "grad_norm": 1.825000286102295, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6603, | |
| "step": 12320 | |
| }, | |
| { | |
| "epoch": 0.6762014356951066, | |
| "grad_norm": 1.5000559091567993, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6861, | |
| "step": 12340 | |
| }, | |
| { | |
| "epoch": 0.6772973861581456, | |
| "grad_norm": 1.826874017715454, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7337, | |
| "step": 12360 | |
| }, | |
| { | |
| "epoch": 0.6783933366211847, | |
| "grad_norm": 2.042325735092163, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7566, | |
| "step": 12380 | |
| }, | |
| { | |
| "epoch": 0.6794892870842238, | |
| "grad_norm": 1.6419124603271484, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6825, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.6805852375472629, | |
| "grad_norm": 2.1221911907196045, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7013, | |
| "step": 12420 | |
| }, | |
| { | |
| "epoch": 0.681681188010302, | |
| "grad_norm": 1.598191738128662, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6976, | |
| "step": 12440 | |
| }, | |
| { | |
| "epoch": 0.682777138473341, | |
| "grad_norm": 1.8890109062194824, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6792, | |
| "step": 12460 | |
| }, | |
| { | |
| "epoch": 0.6838730889363801, | |
| "grad_norm": 1.7647831439971924, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6878, | |
| "step": 12480 | |
| }, | |
| { | |
| "epoch": 0.6849690393994191, | |
| "grad_norm": 1.358193278312683, | |
| "learning_rate": 0.0001, | |
| "loss": 0.754, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.6860649898624582, | |
| "grad_norm": 1.9739768505096436, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6799, | |
| "step": 12520 | |
| }, | |
| { | |
| "epoch": 0.6871609403254972, | |
| "grad_norm": 1.532867670059204, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7063, | |
| "step": 12540 | |
| }, | |
| { | |
| "epoch": 0.6882568907885364, | |
| "grad_norm": 1.3203604221343994, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6725, | |
| "step": 12560 | |
| }, | |
| { | |
| "epoch": 0.6893528412515755, | |
| "grad_norm": 1.835530161857605, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6447, | |
| "step": 12580 | |
| }, | |
| { | |
| "epoch": 0.6904487917146145, | |
| "grad_norm": 1.4508098363876343, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7137, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.6915447421776536, | |
| "grad_norm": 1.520942211151123, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6454, | |
| "step": 12620 | |
| }, | |
| { | |
| "epoch": 0.6926406926406926, | |
| "grad_norm": 1.655716061592102, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6659, | |
| "step": 12640 | |
| }, | |
| { | |
| "epoch": 0.6937366431037317, | |
| "grad_norm": 1.5934149026870728, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7084, | |
| "step": 12660 | |
| }, | |
| { | |
| "epoch": 0.6948325935667707, | |
| "grad_norm": 1.961393117904663, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7553, | |
| "step": 12680 | |
| }, | |
| { | |
| "epoch": 0.6959285440298099, | |
| "grad_norm": 1.4186025857925415, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7205, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.697024494492849, | |
| "grad_norm": 1.6756350994110107, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6166, | |
| "step": 12720 | |
| }, | |
| { | |
| "epoch": 0.698120444955888, | |
| "grad_norm": 1.8438879251480103, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6837, | |
| "step": 12740 | |
| }, | |
| { | |
| "epoch": 0.6992163954189271, | |
| "grad_norm": 1.5732409954071045, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7777, | |
| "step": 12760 | |
| }, | |
| { | |
| "epoch": 0.7003123458819661, | |
| "grad_norm": 1.8927737474441528, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6978, | |
| "step": 12780 | |
| }, | |
| { | |
| "epoch": 0.7014082963450052, | |
| "grad_norm": 1.4720592498779297, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7043, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.7025042468080442, | |
| "grad_norm": 1.8671678304672241, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7245, | |
| "step": 12820 | |
| }, | |
| { | |
| "epoch": 0.7036001972710834, | |
| "grad_norm": 1.5541017055511475, | |
| "learning_rate": 0.0001, | |
| "loss": 0.728, | |
| "step": 12840 | |
| }, | |
| { | |
| "epoch": 0.7046961477341224, | |
| "grad_norm": 1.6623157262802124, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7211, | |
| "step": 12860 | |
| }, | |
| { | |
| "epoch": 0.7057920981971615, | |
| "grad_norm": 2.1644530296325684, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7958, | |
| "step": 12880 | |
| }, | |
| { | |
| "epoch": 0.7068880486602006, | |
| "grad_norm": 1.4526203870773315, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6385, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.7079839991232396, | |
| "grad_norm": 1.586296796798706, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7759, | |
| "step": 12920 | |
| }, | |
| { | |
| "epoch": 0.7090799495862787, | |
| "grad_norm": 1.8547158241271973, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7478, | |
| "step": 12940 | |
| }, | |
| { | |
| "epoch": 0.7101759000493177, | |
| "grad_norm": 1.46295964717865, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5804, | |
| "step": 12960 | |
| }, | |
| { | |
| "epoch": 0.7112718505123569, | |
| "grad_norm": 1.8653600215911865, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6874, | |
| "step": 12980 | |
| }, | |
| { | |
| "epoch": 0.7123678009753959, | |
| "grad_norm": 1.8301453590393066, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6886, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.7123678009753959, | |
| "eval_loss": 0.6790329217910767, | |
| "eval_runtime": 30731.9197, | |
| "eval_samples_per_second": 2.111, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 52.619379401724906, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.713463751438435, | |
| "grad_norm": 2.6560330390930176, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6422, | |
| "step": 13020 | |
| }, | |
| { | |
| "epoch": 0.714559701901474, | |
| "grad_norm": 1.9559868574142456, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7321, | |
| "step": 13040 | |
| }, | |
| { | |
| "epoch": 0.7156556523645131, | |
| "grad_norm": 1.8091590404510498, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6303, | |
| "step": 13060 | |
| }, | |
| { | |
| "epoch": 0.7167516028275522, | |
| "grad_norm": 1.428688406944275, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6489, | |
| "step": 13080 | |
| }, | |
| { | |
| "epoch": 0.7178475532905912, | |
| "grad_norm": 1.6543529033660889, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6793, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.7189435037536304, | |
| "grad_norm": 2.012596368789673, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6502, | |
| "step": 13120 | |
| }, | |
| { | |
| "epoch": 0.7200394542166694, | |
| "grad_norm": 2.0701732635498047, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6746, | |
| "step": 13140 | |
| }, | |
| { | |
| "epoch": 0.7211354046797085, | |
| "grad_norm": 1.5318336486816406, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6142, | |
| "step": 13160 | |
| }, | |
| { | |
| "epoch": 0.7222313551427475, | |
| "grad_norm": 1.7924253940582275, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6536, | |
| "step": 13180 | |
| }, | |
| { | |
| "epoch": 0.7233273056057866, | |
| "grad_norm": 1.8197805881500244, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7804, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.7244232560688257, | |
| "grad_norm": 1.5444835424423218, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6019, | |
| "step": 13220 | |
| }, | |
| { | |
| "epoch": 0.7255192065318647, | |
| "grad_norm": 1.735474705696106, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5891, | |
| "step": 13240 | |
| }, | |
| { | |
| "epoch": 0.7266151569949039, | |
| "grad_norm": 1.9891881942749023, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6498, | |
| "step": 13260 | |
| }, | |
| { | |
| "epoch": 0.7277111074579429, | |
| "grad_norm": 1.6917784214019775, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7044, | |
| "step": 13280 | |
| }, | |
| { | |
| "epoch": 0.728807057920982, | |
| "grad_norm": 1.661033034324646, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6185, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.729903008384021, | |
| "grad_norm": 2.326937198638916, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7274, | |
| "step": 13320 | |
| }, | |
| { | |
| "epoch": 0.7309989588470601, | |
| "grad_norm": 1.6929740905761719, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6452, | |
| "step": 13340 | |
| }, | |
| { | |
| "epoch": 0.7320949093100991, | |
| "grad_norm": 1.7399369478225708, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6243, | |
| "step": 13360 | |
| }, | |
| { | |
| "epoch": 0.7331908597731382, | |
| "grad_norm": 1.7071975469589233, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6578, | |
| "step": 13380 | |
| }, | |
| { | |
| "epoch": 0.7342868102361774, | |
| "grad_norm": 1.7368084192276, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6337, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.7353827606992164, | |
| "grad_norm": 2.0973663330078125, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7101, | |
| "step": 13420 | |
| }, | |
| { | |
| "epoch": 0.7364787111622555, | |
| "grad_norm": 1.636421799659729, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6958, | |
| "step": 13440 | |
| }, | |
| { | |
| "epoch": 0.7375746616252945, | |
| "grad_norm": 1.6134982109069824, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7336, | |
| "step": 13460 | |
| }, | |
| { | |
| "epoch": 0.7386706120883336, | |
| "grad_norm": 1.8911906480789185, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6723, | |
| "step": 13480 | |
| }, | |
| { | |
| "epoch": 0.7397665625513726, | |
| "grad_norm": 1.8372421264648438, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6748, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.7408625130144117, | |
| "grad_norm": 1.8735203742980957, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7481, | |
| "step": 13520 | |
| }, | |
| { | |
| "epoch": 0.7419584634774509, | |
| "grad_norm": 1.7684818506240845, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6395, | |
| "step": 13540 | |
| }, | |
| { | |
| "epoch": 0.7430544139404899, | |
| "grad_norm": 1.8018254041671753, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8523, | |
| "step": 13560 | |
| }, | |
| { | |
| "epoch": 0.744150364403529, | |
| "grad_norm": 1.7703465223312378, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6938, | |
| "step": 13580 | |
| }, | |
| { | |
| "epoch": 0.745246314866568, | |
| "grad_norm": 1.6299625635147095, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6341, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.7463422653296071, | |
| "grad_norm": 2.0545101165771484, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7248, | |
| "step": 13620 | |
| }, | |
| { | |
| "epoch": 0.7474382157926461, | |
| "grad_norm": 1.5883153676986694, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6917, | |
| "step": 13640 | |
| }, | |
| { | |
| "epoch": 0.7485341662556853, | |
| "grad_norm": 1.5751030445098877, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6789, | |
| "step": 13660 | |
| }, | |
| { | |
| "epoch": 0.7496301167187243, | |
| "grad_norm": 1.53587806224823, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6462, | |
| "step": 13680 | |
| }, | |
| { | |
| "epoch": 0.7507260671817634, | |
| "grad_norm": 1.5108363628387451, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7585, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.7518220176448025, | |
| "grad_norm": 1.5622588396072388, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6466, | |
| "step": 13720 | |
| }, | |
| { | |
| "epoch": 0.7529179681078415, | |
| "grad_norm": 1.9326175451278687, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7086, | |
| "step": 13740 | |
| }, | |
| { | |
| "epoch": 0.7540139185708806, | |
| "grad_norm": 1.7847191095352173, | |
| "learning_rate": 0.0001, | |
| "loss": 0.661, | |
| "step": 13760 | |
| }, | |
| { | |
| "epoch": 0.7551098690339196, | |
| "grad_norm": 2.1520116329193115, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5861, | |
| "step": 13780 | |
| }, | |
| { | |
| "epoch": 0.7562058194969588, | |
| "grad_norm": 1.9346301555633545, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7735, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.7573017699599978, | |
| "grad_norm": 1.5564959049224854, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6997, | |
| "step": 13820 | |
| }, | |
| { | |
| "epoch": 0.7583977204230369, | |
| "grad_norm": 1.848569393157959, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6836, | |
| "step": 13840 | |
| }, | |
| { | |
| "epoch": 0.759493670886076, | |
| "grad_norm": 1.5552887916564941, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6014, | |
| "step": 13860 | |
| }, | |
| { | |
| "epoch": 0.760589621349115, | |
| "grad_norm": 1.5576545000076294, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7034, | |
| "step": 13880 | |
| }, | |
| { | |
| "epoch": 0.7616855718121541, | |
| "grad_norm": 1.795949935913086, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7322, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.7627815222751931, | |
| "grad_norm": 1.498818039894104, | |
| "learning_rate": 0.0001, | |
| "loss": 0.697, | |
| "step": 13920 | |
| }, | |
| { | |
| "epoch": 0.7638774727382323, | |
| "grad_norm": 1.7154011726379395, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7057, | |
| "step": 13940 | |
| }, | |
| { | |
| "epoch": 0.7649734232012713, | |
| "grad_norm": 1.693199872970581, | |
| "learning_rate": 0.0001, | |
| "loss": 0.722, | |
| "step": 13960 | |
| }, | |
| { | |
| "epoch": 0.7660693736643104, | |
| "grad_norm": 1.7617517709732056, | |
| "learning_rate": 0.0001, | |
| "loss": 0.727, | |
| "step": 13980 | |
| }, | |
| { | |
| "epoch": 0.7671653241273494, | |
| "grad_norm": 1.7693978548049927, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6552, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.7671653241273494, | |
| "eval_loss": 0.6688939929008484, | |
| "eval_runtime": 30450.1084, | |
| "eval_samples_per_second": 2.131, | |
| "eval_steps_per_second": 0.067, | |
| "eval_wer": 38.925884967114385, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.7682612745903885, | |
| "grad_norm": 1.531043529510498, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6237, | |
| "step": 14020 | |
| }, | |
| { | |
| "epoch": 0.7693572250534276, | |
| "grad_norm": 1.7747310400009155, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6769, | |
| "step": 14040 | |
| }, | |
| { | |
| "epoch": 0.7704531755164666, | |
| "grad_norm": 1.457766056060791, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6406, | |
| "step": 14060 | |
| }, | |
| { | |
| "epoch": 0.7715491259795058, | |
| "grad_norm": 1.478061318397522, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6797, | |
| "step": 14080 | |
| }, | |
| { | |
| "epoch": 0.7726450764425448, | |
| "grad_norm": 1.462485909461975, | |
| "learning_rate": 0.0001, | |
| "loss": 0.654, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.7737410269055839, | |
| "grad_norm": 2.1201417446136475, | |
| "learning_rate": 0.0001, | |
| "loss": 0.676, | |
| "step": 14120 | |
| }, | |
| { | |
| "epoch": 0.7748369773686229, | |
| "grad_norm": 1.6672828197479248, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6402, | |
| "step": 14140 | |
| }, | |
| { | |
| "epoch": 0.775932927831662, | |
| "grad_norm": 1.848254680633545, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6623, | |
| "step": 14160 | |
| }, | |
| { | |
| "epoch": 0.7770288782947011, | |
| "grad_norm": 1.8868560791015625, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6598, | |
| "step": 14180 | |
| }, | |
| { | |
| "epoch": 0.7781248287577401, | |
| "grad_norm": 2.0615594387054443, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6786, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.7792207792207793, | |
| "grad_norm": 1.6596072912216187, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6448, | |
| "step": 14220 | |
| }, | |
| { | |
| "epoch": 0.7803167296838183, | |
| "grad_norm": 2.0829083919525146, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6558, | |
| "step": 14240 | |
| }, | |
| { | |
| "epoch": 0.7814126801468574, | |
| "grad_norm": 1.7660095691680908, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7399, | |
| "step": 14260 | |
| }, | |
| { | |
| "epoch": 0.7825086306098964, | |
| "grad_norm": 1.7068332433700562, | |
| "learning_rate": 0.0001, | |
| "loss": 0.637, | |
| "step": 14280 | |
| }, | |
| { | |
| "epoch": 0.7836045810729355, | |
| "grad_norm": 1.5040172338485718, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5708, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.7847005315359745, | |
| "grad_norm": 1.7479969263076782, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7348, | |
| "step": 14320 | |
| }, | |
| { | |
| "epoch": 0.7857964819990136, | |
| "grad_norm": 1.7886347770690918, | |
| "learning_rate": 0.0001, | |
| "loss": 0.72, | |
| "step": 14340 | |
| }, | |
| { | |
| "epoch": 0.7868924324620528, | |
| "grad_norm": 1.6001741886138916, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6512, | |
| "step": 14360 | |
| }, | |
| { | |
| "epoch": 0.7879883829250918, | |
| "grad_norm": 1.7489492893218994, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6714, | |
| "step": 14380 | |
| }, | |
| { | |
| "epoch": 0.7890843333881309, | |
| "grad_norm": 1.9967806339263916, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6651, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.7901802838511699, | |
| "grad_norm": 1.6555088758468628, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6584, | |
| "step": 14420 | |
| }, | |
| { | |
| "epoch": 0.791276234314209, | |
| "grad_norm": 1.589168667793274, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6448, | |
| "step": 14440 | |
| }, | |
| { | |
| "epoch": 0.792372184777248, | |
| "grad_norm": 1.2876309156417847, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6612, | |
| "step": 14460 | |
| }, | |
| { | |
| "epoch": 0.7934681352402871, | |
| "grad_norm": 1.6673985719680786, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5769, | |
| "step": 14480 | |
| }, | |
| { | |
| "epoch": 0.7945640857033263, | |
| "grad_norm": 1.6478184461593628, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6457, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.7956600361663653, | |
| "grad_norm": 1.5702099800109863, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6866, | |
| "step": 14520 | |
| }, | |
| { | |
| "epoch": 0.7967559866294044, | |
| "grad_norm": 1.850900411605835, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6475, | |
| "step": 14540 | |
| }, | |
| { | |
| "epoch": 0.7978519370924434, | |
| "grad_norm": 1.2784024477005005, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6366, | |
| "step": 14560 | |
| }, | |
| { | |
| "epoch": 0.7989478875554825, | |
| "grad_norm": 2.2533817291259766, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6164, | |
| "step": 14580 | |
| }, | |
| { | |
| "epoch": 0.8000438380185215, | |
| "grad_norm": 1.442713737487793, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6853, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.8011397884815606, | |
| "grad_norm": 1.594449520111084, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6535, | |
| "step": 14620 | |
| }, | |
| { | |
| "epoch": 0.8022357389445997, | |
| "grad_norm": 1.4961411952972412, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6696, | |
| "step": 14640 | |
| }, | |
| { | |
| "epoch": 0.8033316894076388, | |
| "grad_norm": 2.1010756492614746, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6607, | |
| "step": 14660 | |
| }, | |
| { | |
| "epoch": 0.8044276398706779, | |
| "grad_norm": 2.134493589401245, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6512, | |
| "step": 14680 | |
| }, | |
| { | |
| "epoch": 0.8055235903337169, | |
| "grad_norm": 1.6435072422027588, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6094, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.806619540796756, | |
| "grad_norm": 1.8982771635055542, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6761, | |
| "step": 14720 | |
| }, | |
| { | |
| "epoch": 0.807715491259795, | |
| "grad_norm": 1.968770146369934, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7178, | |
| "step": 14740 | |
| }, | |
| { | |
| "epoch": 0.8088114417228341, | |
| "grad_norm": 2.19568133354187, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6246, | |
| "step": 14760 | |
| }, | |
| { | |
| "epoch": 0.8099073921858732, | |
| "grad_norm": 1.6024566888809204, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6174, | |
| "step": 14780 | |
| }, | |
| { | |
| "epoch": 0.8110033426489123, | |
| "grad_norm": 1.4896485805511475, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5843, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.8120992931119514, | |
| "grad_norm": 1.502487301826477, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7747, | |
| "step": 14820 | |
| }, | |
| { | |
| "epoch": 0.8131952435749904, | |
| "grad_norm": 1.5037872791290283, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6745, | |
| "step": 14840 | |
| }, | |
| { | |
| "epoch": 0.8142911940380295, | |
| "grad_norm": 1.3984043598175049, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6938, | |
| "step": 14860 | |
| }, | |
| { | |
| "epoch": 0.8153871445010685, | |
| "grad_norm": 1.7627023458480835, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6407, | |
| "step": 14880 | |
| }, | |
| { | |
| "epoch": 0.8164830949641076, | |
| "grad_norm": 1.5276484489440918, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6142, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.8175790454271467, | |
| "grad_norm": 1.598743200302124, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6534, | |
| "step": 14920 | |
| }, | |
| { | |
| "epoch": 0.8186749958901858, | |
| "grad_norm": 1.5528680086135864, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6272, | |
| "step": 14940 | |
| }, | |
| { | |
| "epoch": 0.8197709463532248, | |
| "grad_norm": 1.71839439868927, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7259, | |
| "step": 14960 | |
| }, | |
| { | |
| "epoch": 0.8208668968162639, | |
| "grad_norm": 1.5527739524841309, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6387, | |
| "step": 14980 | |
| }, | |
| { | |
| "epoch": 0.821962847279303, | |
| "grad_norm": 1.7775479555130005, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6963, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.821962847279303, | |
| "eval_loss": 0.6593644618988037, | |
| "eval_runtime": 30360.7703, | |
| "eval_samples_per_second": 2.137, | |
| "eval_steps_per_second": 0.067, | |
| "eval_wer": 42.68128173436093, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.823058797742342, | |
| "grad_norm": 1.772290825843811, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6439, | |
| "step": 15020 | |
| }, | |
| { | |
| "epoch": 0.8241547482053811, | |
| "grad_norm": 1.655604600906372, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7521, | |
| "step": 15040 | |
| }, | |
| { | |
| "epoch": 0.8252506986684202, | |
| "grad_norm": 1.5305246114730835, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5623, | |
| "step": 15060 | |
| }, | |
| { | |
| "epoch": 0.8263466491314593, | |
| "grad_norm": 1.399568796157837, | |
| "learning_rate": 0.0001, | |
| "loss": 0.634, | |
| "step": 15080 | |
| }, | |
| { | |
| "epoch": 0.8274425995944983, | |
| "grad_norm": 1.412463903427124, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7272, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.8285385500575374, | |
| "grad_norm": 1.793396234512329, | |
| "learning_rate": 0.0001, | |
| "loss": 0.684, | |
| "step": 15120 | |
| }, | |
| { | |
| "epoch": 0.8296345005205765, | |
| "grad_norm": 1.9623442888259888, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6247, | |
| "step": 15140 | |
| }, | |
| { | |
| "epoch": 0.8307304509836155, | |
| "grad_norm": 1.4576257467269897, | |
| "learning_rate": 0.0001, | |
| "loss": 0.675, | |
| "step": 15160 | |
| }, | |
| { | |
| "epoch": 0.8318264014466547, | |
| "grad_norm": 1.6135623455047607, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7046, | |
| "step": 15180 | |
| }, | |
| { | |
| "epoch": 0.8329223519096937, | |
| "grad_norm": 1.5553112030029297, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7246, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.8340183023727328, | |
| "grad_norm": 1.4521915912628174, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7288, | |
| "step": 15220 | |
| }, | |
| { | |
| "epoch": 0.8351142528357718, | |
| "grad_norm": 1.429190754890442, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6357, | |
| "step": 15240 | |
| }, | |
| { | |
| "epoch": 0.8362102032988109, | |
| "grad_norm": 1.80194890499115, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6462, | |
| "step": 15260 | |
| }, | |
| { | |
| "epoch": 0.8373061537618499, | |
| "grad_norm": 1.833225131034851, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6942, | |
| "step": 15280 | |
| }, | |
| { | |
| "epoch": 0.838402104224889, | |
| "grad_norm": 1.8329098224639893, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6525, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.8394980546879282, | |
| "grad_norm": 1.5729244947433472, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6721, | |
| "step": 15320 | |
| }, | |
| { | |
| "epoch": 0.8405940051509672, | |
| "grad_norm": 1.8156899213790894, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6777, | |
| "step": 15340 | |
| }, | |
| { | |
| "epoch": 0.8416899556140063, | |
| "grad_norm": 1.7255985736846924, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6653, | |
| "step": 15360 | |
| }, | |
| { | |
| "epoch": 0.8427859060770453, | |
| "grad_norm": 1.8051388263702393, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6429, | |
| "step": 15380 | |
| }, | |
| { | |
| "epoch": 0.8438818565400844, | |
| "grad_norm": 1.4799489974975586, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6219, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.8449778070031234, | |
| "grad_norm": 1.5661497116088867, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7042, | |
| "step": 15420 | |
| }, | |
| { | |
| "epoch": 0.8460737574661625, | |
| "grad_norm": 1.7842859029769897, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5657, | |
| "step": 15440 | |
| }, | |
| { | |
| "epoch": 0.8471697079292017, | |
| "grad_norm": 2.036591053009033, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7448, | |
| "step": 15460 | |
| }, | |
| { | |
| "epoch": 0.8482656583922407, | |
| "grad_norm": 1.5923106670379639, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6937, | |
| "step": 15480 | |
| }, | |
| { | |
| "epoch": 0.8493616088552798, | |
| "grad_norm": 1.7609819173812866, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6282, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.8504575593183188, | |
| "grad_norm": 1.627193570137024, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6593, | |
| "step": 15520 | |
| }, | |
| { | |
| "epoch": 0.8515535097813579, | |
| "grad_norm": 1.5199600458145142, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6493, | |
| "step": 15540 | |
| }, | |
| { | |
| "epoch": 0.8526494602443969, | |
| "grad_norm": 1.8375046253204346, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7139, | |
| "step": 15560 | |
| }, | |
| { | |
| "epoch": 0.853745410707436, | |
| "grad_norm": 1.7061831951141357, | |
| "learning_rate": 0.0001, | |
| "loss": 0.645, | |
| "step": 15580 | |
| }, | |
| { | |
| "epoch": 0.854841361170475, | |
| "grad_norm": 1.5046154260635376, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7154, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.8559373116335142, | |
| "grad_norm": 2.0937325954437256, | |
| "learning_rate": 0.0001, | |
| "loss": 0.716, | |
| "step": 15620 | |
| }, | |
| { | |
| "epoch": 0.8570332620965533, | |
| "grad_norm": 1.502930760383606, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6476, | |
| "step": 15640 | |
| }, | |
| { | |
| "epoch": 0.8581292125595923, | |
| "grad_norm": 1.832287073135376, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6279, | |
| "step": 15660 | |
| }, | |
| { | |
| "epoch": 0.8592251630226314, | |
| "grad_norm": 1.9679219722747803, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7048, | |
| "step": 15680 | |
| }, | |
| { | |
| "epoch": 0.8603211134856704, | |
| "grad_norm": 1.4660624265670776, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6217, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.8614170639487095, | |
| "grad_norm": 1.6641209125518799, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5692, | |
| "step": 15720 | |
| }, | |
| { | |
| "epoch": 0.8625130144117485, | |
| "grad_norm": 1.6354645490646362, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7019, | |
| "step": 15740 | |
| }, | |
| { | |
| "epoch": 0.8636089648747877, | |
| "grad_norm": 1.5404868125915527, | |
| "learning_rate": 0.0001, | |
| "loss": 0.667, | |
| "step": 15760 | |
| }, | |
| { | |
| "epoch": 0.8647049153378268, | |
| "grad_norm": 1.759466528892517, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7413, | |
| "step": 15780 | |
| }, | |
| { | |
| "epoch": 0.8658008658008658, | |
| "grad_norm": 1.289501667022705, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6696, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.8668968162639049, | |
| "grad_norm": 1.516506552696228, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6385, | |
| "step": 15820 | |
| }, | |
| { | |
| "epoch": 0.8679927667269439, | |
| "grad_norm": 1.602023959159851, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6034, | |
| "step": 15840 | |
| }, | |
| { | |
| "epoch": 0.869088717189983, | |
| "grad_norm": 1.6681197881698608, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6524, | |
| "step": 15860 | |
| }, | |
| { | |
| "epoch": 0.870184667653022, | |
| "grad_norm": 1.7448092699050903, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5845, | |
| "step": 15880 | |
| }, | |
| { | |
| "epoch": 0.8712806181160612, | |
| "grad_norm": 1.763609766960144, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6078, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.8723765685791002, | |
| "grad_norm": 1.8752708435058594, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6415, | |
| "step": 15920 | |
| }, | |
| { | |
| "epoch": 0.8734725190421393, | |
| "grad_norm": 1.4633687734603882, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6509, | |
| "step": 15940 | |
| }, | |
| { | |
| "epoch": 0.8745684695051784, | |
| "grad_norm": 1.630188226699829, | |
| "learning_rate": 0.0001, | |
| "loss": 0.678, | |
| "step": 15960 | |
| }, | |
| { | |
| "epoch": 0.8756644199682174, | |
| "grad_norm": 1.746390461921692, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6883, | |
| "step": 15980 | |
| }, | |
| { | |
| "epoch": 0.8767603704312565, | |
| "grad_norm": 1.8357354402542114, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5674, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.8767603704312565, | |
| "eval_loss": 0.6495629549026489, | |
| "eval_runtime": 30578.7623, | |
| "eval_samples_per_second": 2.122, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 46.744478263995646, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.8778563208942955, | |
| "grad_norm": 1.3959294557571411, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7334, | |
| "step": 16020 | |
| }, | |
| { | |
| "epoch": 0.8789522713573347, | |
| "grad_norm": 1.7587610483169556, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6916, | |
| "step": 16040 | |
| }, | |
| { | |
| "epoch": 0.8800482218203737, | |
| "grad_norm": 2.157567024230957, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6288, | |
| "step": 16060 | |
| }, | |
| { | |
| "epoch": 0.8811441722834128, | |
| "grad_norm": 1.927071452140808, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6596, | |
| "step": 16080 | |
| }, | |
| { | |
| "epoch": 0.8822401227464519, | |
| "grad_norm": 1.7229890823364258, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6351, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.8833360732094909, | |
| "grad_norm": 1.4584635496139526, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6347, | |
| "step": 16120 | |
| }, | |
| { | |
| "epoch": 0.88443202367253, | |
| "grad_norm": 1.4768098592758179, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6326, | |
| "step": 16140 | |
| }, | |
| { | |
| "epoch": 0.885527974135569, | |
| "grad_norm": 1.6411234140396118, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6935, | |
| "step": 16160 | |
| }, | |
| { | |
| "epoch": 0.8866239245986082, | |
| "grad_norm": 1.4742987155914307, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6188, | |
| "step": 16180 | |
| }, | |
| { | |
| "epoch": 0.8877198750616472, | |
| "grad_norm": 2.1708977222442627, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6837, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.8888158255246863, | |
| "grad_norm": 1.5142560005187988, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6218, | |
| "step": 16220 | |
| }, | |
| { | |
| "epoch": 0.8899117759877253, | |
| "grad_norm": 1.5650640726089478, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5991, | |
| "step": 16240 | |
| }, | |
| { | |
| "epoch": 0.8910077264507644, | |
| "grad_norm": 1.5553919076919556, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6081, | |
| "step": 16260 | |
| }, | |
| { | |
| "epoch": 0.8921036769138035, | |
| "grad_norm": 1.813482642173767, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6599, | |
| "step": 16280 | |
| }, | |
| { | |
| "epoch": 0.8931996273768426, | |
| "grad_norm": 1.6864385604858398, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6337, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.8942955778398817, | |
| "grad_norm": 1.5707799196243286, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7029, | |
| "step": 16320 | |
| }, | |
| { | |
| "epoch": 0.8953915283029207, | |
| "grad_norm": 1.3465133905410767, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6519, | |
| "step": 16340 | |
| }, | |
| { | |
| "epoch": 0.8964874787659598, | |
| "grad_norm": 1.5546880960464478, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6111, | |
| "step": 16360 | |
| }, | |
| { | |
| "epoch": 0.8975834292289988, | |
| "grad_norm": 1.6297564506530762, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6825, | |
| "step": 16380 | |
| }, | |
| { | |
| "epoch": 0.8986793796920379, | |
| "grad_norm": 1.5396370887756348, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6454, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.8997753301550769, | |
| "grad_norm": 1.3082808256149292, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6489, | |
| "step": 16420 | |
| }, | |
| { | |
| "epoch": 0.900871280618116, | |
| "grad_norm": 1.68564772605896, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6688, | |
| "step": 16440 | |
| }, | |
| { | |
| "epoch": 0.9019672310811552, | |
| "grad_norm": 1.6919423341751099, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6353, | |
| "step": 16460 | |
| }, | |
| { | |
| "epoch": 0.9030631815441942, | |
| "grad_norm": 1.4040336608886719, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6286, | |
| "step": 16480 | |
| }, | |
| { | |
| "epoch": 0.9041591320072333, | |
| "grad_norm": 1.5394583940505981, | |
| "learning_rate": 0.0001, | |
| "loss": 0.648, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.9052550824702723, | |
| "grad_norm": 1.8135911226272583, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6315, | |
| "step": 16520 | |
| }, | |
| { | |
| "epoch": 0.9063510329333114, | |
| "grad_norm": 1.6827434301376343, | |
| "learning_rate": 0.0001, | |
| "loss": 0.637, | |
| "step": 16540 | |
| }, | |
| { | |
| "epoch": 0.9074469833963504, | |
| "grad_norm": 1.3692152500152588, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7015, | |
| "step": 16560 | |
| }, | |
| { | |
| "epoch": 0.9085429338593896, | |
| "grad_norm": 1.6391196250915527, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6479, | |
| "step": 16580 | |
| }, | |
| { | |
| "epoch": 0.9096388843224287, | |
| "grad_norm": 2.5071117877960205, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6746, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.9107348347854677, | |
| "grad_norm": 1.7680779695510864, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6543, | |
| "step": 16620 | |
| }, | |
| { | |
| "epoch": 0.9118307852485068, | |
| "grad_norm": 1.487269639968872, | |
| "learning_rate": 0.0001, | |
| "loss": 0.618, | |
| "step": 16640 | |
| }, | |
| { | |
| "epoch": 0.9129267357115458, | |
| "grad_norm": 1.322325348854065, | |
| "learning_rate": 0.0001, | |
| "loss": 0.635, | |
| "step": 16660 | |
| }, | |
| { | |
| "epoch": 0.9140226861745849, | |
| "grad_norm": 2.054997682571411, | |
| "learning_rate": 0.0001, | |
| "loss": 0.645, | |
| "step": 16680 | |
| }, | |
| { | |
| "epoch": 0.9151186366376239, | |
| "grad_norm": 1.7619165182113647, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6405, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 0.9162145871006631, | |
| "grad_norm": 1.3276571035385132, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5797, | |
| "step": 16720 | |
| }, | |
| { | |
| "epoch": 0.9173105375637021, | |
| "grad_norm": 2.2796542644500732, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7342, | |
| "step": 16740 | |
| }, | |
| { | |
| "epoch": 0.9184064880267412, | |
| "grad_norm": 1.637654423713684, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6149, | |
| "step": 16760 | |
| }, | |
| { | |
| "epoch": 0.9195024384897803, | |
| "grad_norm": 1.4013864994049072, | |
| "learning_rate": 0.0001, | |
| "loss": 0.627, | |
| "step": 16780 | |
| }, | |
| { | |
| "epoch": 0.9205983889528193, | |
| "grad_norm": 1.5173211097717285, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5449, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.9216943394158584, | |
| "grad_norm": 1.5530805587768555, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5416, | |
| "step": 16820 | |
| }, | |
| { | |
| "epoch": 0.9227902898788974, | |
| "grad_norm": 1.5294363498687744, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6146, | |
| "step": 16840 | |
| }, | |
| { | |
| "epoch": 0.9238862403419366, | |
| "grad_norm": 1.7312266826629639, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6952, | |
| "step": 16860 | |
| }, | |
| { | |
| "epoch": 0.9249821908049756, | |
| "grad_norm": 1.301459789276123, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6258, | |
| "step": 16880 | |
| }, | |
| { | |
| "epoch": 0.9260781412680147, | |
| "grad_norm": 1.915128469467163, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6955, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 0.9271740917310538, | |
| "grad_norm": 1.3437505960464478, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7044, | |
| "step": 16920 | |
| }, | |
| { | |
| "epoch": 0.9282700421940928, | |
| "grad_norm": 1.5920603275299072, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5871, | |
| "step": 16940 | |
| }, | |
| { | |
| "epoch": 0.9293659926571319, | |
| "grad_norm": 1.2615900039672852, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6311, | |
| "step": 16960 | |
| }, | |
| { | |
| "epoch": 0.9304619431201709, | |
| "grad_norm": 1.6863378286361694, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5746, | |
| "step": 16980 | |
| }, | |
| { | |
| "epoch": 0.9315578935832101, | |
| "grad_norm": 1.3633450269699097, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6354, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.9315578935832101, | |
| "eval_loss": 0.6385661959648132, | |
| "eval_runtime": 30462.2265, | |
| "eval_samples_per_second": 2.13, | |
| "eval_steps_per_second": 0.067, | |
| "eval_wer": 36.07830918982583, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.9326538440462491, | |
| "grad_norm": 1.7412103414535522, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6635, | |
| "step": 17020 | |
| }, | |
| { | |
| "epoch": 0.9337497945092882, | |
| "grad_norm": 2.0697691440582275, | |
| "learning_rate": 0.0001, | |
| "loss": 0.678, | |
| "step": 17040 | |
| }, | |
| { | |
| "epoch": 0.9348457449723272, | |
| "grad_norm": 1.6238869428634644, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6427, | |
| "step": 17060 | |
| }, | |
| { | |
| "epoch": 0.9359416954353663, | |
| "grad_norm": 1.498334288597107, | |
| "learning_rate": 0.0001, | |
| "loss": 0.614, | |
| "step": 17080 | |
| }, | |
| { | |
| "epoch": 0.9370376458984054, | |
| "grad_norm": 1.4905815124511719, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5759, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 0.9381335963614444, | |
| "grad_norm": 1.433747410774231, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6754, | |
| "step": 17120 | |
| }, | |
| { | |
| "epoch": 0.9392295468244836, | |
| "grad_norm": 1.8419586420059204, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6392, | |
| "step": 17140 | |
| }, | |
| { | |
| "epoch": 0.9403254972875226, | |
| "grad_norm": 1.5990883111953735, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6204, | |
| "step": 17160 | |
| }, | |
| { | |
| "epoch": 0.9414214477505617, | |
| "grad_norm": 1.482010841369629, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6887, | |
| "step": 17180 | |
| }, | |
| { | |
| "epoch": 0.9425173982136007, | |
| "grad_norm": 1.6629010438919067, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6457, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.9436133486766398, | |
| "grad_norm": 1.4538336992263794, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6401, | |
| "step": 17220 | |
| }, | |
| { | |
| "epoch": 0.9447092991396789, | |
| "grad_norm": 1.5684305429458618, | |
| "learning_rate": 0.0001, | |
| "loss": 0.676, | |
| "step": 17240 | |
| }, | |
| { | |
| "epoch": 0.945805249602718, | |
| "grad_norm": 1.4637812376022339, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5963, | |
| "step": 17260 | |
| }, | |
| { | |
| "epoch": 0.9469012000657571, | |
| "grad_norm": 2.155348300933838, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6624, | |
| "step": 17280 | |
| }, | |
| { | |
| "epoch": 0.9479971505287961, | |
| "grad_norm": 1.6532953977584839, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6784, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 0.9490931009918352, | |
| "grad_norm": 1.934787392616272, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6876, | |
| "step": 17320 | |
| }, | |
| { | |
| "epoch": 0.9501890514548742, | |
| "grad_norm": 2.319920063018799, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7676, | |
| "step": 17340 | |
| }, | |
| { | |
| "epoch": 0.9512850019179133, | |
| "grad_norm": 1.5026947259902954, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5634, | |
| "step": 17360 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 1.9578672647476196, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5959, | |
| "step": 17380 | |
| }, | |
| { | |
| "epoch": 0.9534769028439914, | |
| "grad_norm": 1.9930877685546875, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5947, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.9545728533070306, | |
| "grad_norm": 1.6241062879562378, | |
| "learning_rate": 0.0001, | |
| "loss": 0.587, | |
| "step": 17420 | |
| }, | |
| { | |
| "epoch": 0.9556688037700696, | |
| "grad_norm": 1.7155011892318726, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6352, | |
| "step": 17440 | |
| }, | |
| { | |
| "epoch": 0.9567647542331087, | |
| "grad_norm": 1.7239856719970703, | |
| "learning_rate": 0.0001, | |
| "loss": 0.647, | |
| "step": 17460 | |
| }, | |
| { | |
| "epoch": 0.9578607046961477, | |
| "grad_norm": 1.6342066526412964, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6017, | |
| "step": 17480 | |
| }, | |
| { | |
| "epoch": 0.9589566551591868, | |
| "grad_norm": 1.4042915105819702, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6479, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.9600526056222258, | |
| "grad_norm": 1.5023634433746338, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6008, | |
| "step": 17520 | |
| }, | |
| { | |
| "epoch": 0.961148556085265, | |
| "grad_norm": 1.5713409185409546, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6643, | |
| "step": 17540 | |
| }, | |
| { | |
| "epoch": 0.9622445065483041, | |
| "grad_norm": 1.8917444944381714, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6992, | |
| "step": 17560 | |
| }, | |
| { | |
| "epoch": 0.9633404570113431, | |
| "grad_norm": 1.918900728225708, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6004, | |
| "step": 17580 | |
| }, | |
| { | |
| "epoch": 0.9644364074743822, | |
| "grad_norm": 1.7599738836288452, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6479, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.9655323579374212, | |
| "grad_norm": 1.4554500579833984, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6367, | |
| "step": 17620 | |
| }, | |
| { | |
| "epoch": 0.9666283084004603, | |
| "grad_norm": 1.6860467195510864, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5769, | |
| "step": 17640 | |
| }, | |
| { | |
| "epoch": 0.9677242588634993, | |
| "grad_norm": 1.6800360679626465, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6402, | |
| "step": 17660 | |
| }, | |
| { | |
| "epoch": 0.9688202093265385, | |
| "grad_norm": 1.3988690376281738, | |
| "learning_rate": 0.0001, | |
| "loss": 0.589, | |
| "step": 17680 | |
| }, | |
| { | |
| "epoch": 0.9699161597895775, | |
| "grad_norm": 1.6789034605026245, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6469, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 0.9710121102526166, | |
| "grad_norm": 1.4465025663375854, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6362, | |
| "step": 17720 | |
| }, | |
| { | |
| "epoch": 0.9721080607156557, | |
| "grad_norm": 1.6466797590255737, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5882, | |
| "step": 17740 | |
| }, | |
| { | |
| "epoch": 0.9732040111786947, | |
| "grad_norm": 1.4487119913101196, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6481, | |
| "step": 17760 | |
| }, | |
| { | |
| "epoch": 0.9742999616417338, | |
| "grad_norm": 1.769286870956421, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7198, | |
| "step": 17780 | |
| }, | |
| { | |
| "epoch": 0.9753959121047728, | |
| "grad_norm": 1.8515903949737549, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6257, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.976491862567812, | |
| "grad_norm": 1.9811028242111206, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6154, | |
| "step": 17820 | |
| }, | |
| { | |
| "epoch": 0.977587813030851, | |
| "grad_norm": 1.6197538375854492, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6196, | |
| "step": 17840 | |
| }, | |
| { | |
| "epoch": 0.9786837634938901, | |
| "grad_norm": 1.605971336364746, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5549, | |
| "step": 17860 | |
| }, | |
| { | |
| "epoch": 0.9797797139569292, | |
| "grad_norm": 1.5246946811676025, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5979, | |
| "step": 17880 | |
| }, | |
| { | |
| "epoch": 0.9808756644199682, | |
| "grad_norm": 1.8534538745880127, | |
| "learning_rate": 0.0001, | |
| "loss": 0.663, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 0.9819716148830073, | |
| "grad_norm": 1.6625508069992065, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5863, | |
| "step": 17920 | |
| }, | |
| { | |
| "epoch": 0.9830675653460463, | |
| "grad_norm": 1.704788088798523, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6529, | |
| "step": 17940 | |
| }, | |
| { | |
| "epoch": 0.9841635158090855, | |
| "grad_norm": 1.3650102615356445, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6317, | |
| "step": 17960 | |
| }, | |
| { | |
| "epoch": 0.9852594662721245, | |
| "grad_norm": 1.7923402786254883, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6023, | |
| "step": 17980 | |
| }, | |
| { | |
| "epoch": 0.9863554167351636, | |
| "grad_norm": 2.0028254985809326, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6569, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.9863554167351636, | |
| "eval_loss": 0.6309429407119751, | |
| "eval_runtime": 30654.7333, | |
| "eval_samples_per_second": 2.117, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 36.64375002139876, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.9874513671982026, | |
| "grad_norm": 1.66196870803833, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6687, | |
| "step": 18020 | |
| }, | |
| { | |
| "epoch": 0.9885473176612417, | |
| "grad_norm": 2.031445264816284, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6032, | |
| "step": 18040 | |
| }, | |
| { | |
| "epoch": 0.9896432681242808, | |
| "grad_norm": 1.3204675912857056, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5913, | |
| "step": 18060 | |
| }, | |
| { | |
| "epoch": 0.9907392185873198, | |
| "grad_norm": 1.7389861345291138, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6678, | |
| "step": 18080 | |
| }, | |
| { | |
| "epoch": 0.991835169050359, | |
| "grad_norm": 1.9815995693206787, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6566, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 0.992931119513398, | |
| "grad_norm": 1.5902196168899536, | |
| "learning_rate": 0.0001, | |
| "loss": 0.617, | |
| "step": 18120 | |
| }, | |
| { | |
| "epoch": 0.9940270699764371, | |
| "grad_norm": 1.4741644859313965, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5921, | |
| "step": 18140 | |
| }, | |
| { | |
| "epoch": 0.9951230204394761, | |
| "grad_norm": 1.419965147972107, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6031, | |
| "step": 18160 | |
| }, | |
| { | |
| "epoch": 0.9962189709025152, | |
| "grad_norm": 1.5964018106460571, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6015, | |
| "step": 18180 | |
| }, | |
| { | |
| "epoch": 0.9973149213655543, | |
| "grad_norm": 1.4470981359481812, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6898, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.9984108718285933, | |
| "grad_norm": 1.54426908493042, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6328, | |
| "step": 18220 | |
| }, | |
| { | |
| "epoch": 0.9995068222916325, | |
| "grad_norm": 1.778437614440918, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6381, | |
| "step": 18240 | |
| }, | |
| { | |
| "epoch": 1.0006027727546716, | |
| "grad_norm": 1.5332226753234863, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5782, | |
| "step": 18260 | |
| }, | |
| { | |
| "epoch": 1.0016987232177106, | |
| "grad_norm": 1.785762906074524, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6779, | |
| "step": 18280 | |
| }, | |
| { | |
| "epoch": 1.0027946736807496, | |
| "grad_norm": 1.684708595275879, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5786, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 1.0038906241437886, | |
| "grad_norm": 1.5112040042877197, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5989, | |
| "step": 18320 | |
| }, | |
| { | |
| "epoch": 1.0049865746068278, | |
| "grad_norm": 1.5711543560028076, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6525, | |
| "step": 18340 | |
| }, | |
| { | |
| "epoch": 1.0060825250698668, | |
| "grad_norm": 1.4053367376327515, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5913, | |
| "step": 18360 | |
| }, | |
| { | |
| "epoch": 1.0071784755329058, | |
| "grad_norm": 1.7791050672531128, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6212, | |
| "step": 18380 | |
| }, | |
| { | |
| "epoch": 1.008274425995945, | |
| "grad_norm": 1.4202812910079956, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6168, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 1.009370376458984, | |
| "grad_norm": 1.7469732761383057, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6425, | |
| "step": 18420 | |
| }, | |
| { | |
| "epoch": 1.010466326922023, | |
| "grad_norm": 2.1195449829101562, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6564, | |
| "step": 18440 | |
| }, | |
| { | |
| "epoch": 1.011562277385062, | |
| "grad_norm": 1.4056214094161987, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6809, | |
| "step": 18460 | |
| }, | |
| { | |
| "epoch": 1.0126582278481013, | |
| "grad_norm": 2.07029128074646, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6119, | |
| "step": 18480 | |
| }, | |
| { | |
| "epoch": 1.0137541783111403, | |
| "grad_norm": 1.6518419981002808, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5954, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.0148501287741793, | |
| "grad_norm": 1.7785189151763916, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5959, | |
| "step": 18520 | |
| }, | |
| { | |
| "epoch": 1.0159460792372186, | |
| "grad_norm": 1.4902641773223877, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6066, | |
| "step": 18540 | |
| }, | |
| { | |
| "epoch": 1.0170420297002576, | |
| "grad_norm": 1.6291300058364868, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5781, | |
| "step": 18560 | |
| }, | |
| { | |
| "epoch": 1.0181379801632966, | |
| "grad_norm": 1.5571300983428955, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5622, | |
| "step": 18580 | |
| }, | |
| { | |
| "epoch": 1.0192339306263356, | |
| "grad_norm": 1.5963464975357056, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6797, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 1.0203298810893748, | |
| "grad_norm": 1.4604226350784302, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6712, | |
| "step": 18620 | |
| }, | |
| { | |
| "epoch": 1.0214258315524138, | |
| "grad_norm": 1.6052221059799194, | |
| "learning_rate": 0.0001, | |
| "loss": 0.595, | |
| "step": 18640 | |
| }, | |
| { | |
| "epoch": 1.0225217820154529, | |
| "grad_norm": 1.4075971841812134, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6123, | |
| "step": 18660 | |
| }, | |
| { | |
| "epoch": 1.0236177324784919, | |
| "grad_norm": 1.910475730895996, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7016, | |
| "step": 18680 | |
| }, | |
| { | |
| "epoch": 1.024713682941531, | |
| "grad_norm": 1.946268081665039, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6178, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 1.02580963340457, | |
| "grad_norm": 1.5547478199005127, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6068, | |
| "step": 18720 | |
| }, | |
| { | |
| "epoch": 1.026905583867609, | |
| "grad_norm": 1.5006910562515259, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5818, | |
| "step": 18740 | |
| }, | |
| { | |
| "epoch": 1.0280015343306483, | |
| "grad_norm": 1.5395736694335938, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6614, | |
| "step": 18760 | |
| }, | |
| { | |
| "epoch": 1.0290974847936873, | |
| "grad_norm": 1.5935709476470947, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5593, | |
| "step": 18780 | |
| }, | |
| { | |
| "epoch": 1.0301934352567264, | |
| "grad_norm": 1.6643317937850952, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6441, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 1.0312893857197654, | |
| "grad_norm": 1.6811660528182983, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5663, | |
| "step": 18820 | |
| }, | |
| { | |
| "epoch": 1.0323853361828046, | |
| "grad_norm": 1.4203201532363892, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6622, | |
| "step": 18840 | |
| }, | |
| { | |
| "epoch": 1.0334812866458436, | |
| "grad_norm": 1.9712319374084473, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5818, | |
| "step": 18860 | |
| }, | |
| { | |
| "epoch": 1.0345772371088826, | |
| "grad_norm": 2.0921614170074463, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6489, | |
| "step": 18880 | |
| }, | |
| { | |
| "epoch": 1.0356731875719218, | |
| "grad_norm": 1.3215636014938354, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5784, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 1.0367691380349608, | |
| "grad_norm": 1.6520031690597534, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6069, | |
| "step": 18920 | |
| }, | |
| { | |
| "epoch": 1.0378650884979999, | |
| "grad_norm": 1.8051795959472656, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6596, | |
| "step": 18940 | |
| }, | |
| { | |
| "epoch": 1.0389610389610389, | |
| "grad_norm": 1.7375438213348389, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5161, | |
| "step": 18960 | |
| }, | |
| { | |
| "epoch": 1.040056989424078, | |
| "grad_norm": 1.4865177869796753, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5436, | |
| "step": 18980 | |
| }, | |
| { | |
| "epoch": 1.041152939887117, | |
| "grad_norm": 1.4444303512573242, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5906, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.041152939887117, | |
| "eval_loss": 0.622921347618103, | |
| "eval_runtime": 30997.4799, | |
| "eval_samples_per_second": 2.093, | |
| "eval_steps_per_second": 0.065, | |
| "eval_wer": 37.84105343527132, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.042248890350156, | |
| "grad_norm": 1.4964603185653687, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6014, | |
| "step": 19020 | |
| }, | |
| { | |
| "epoch": 1.0433448408131953, | |
| "grad_norm": 2.2886295318603516, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6472, | |
| "step": 19040 | |
| }, | |
| { | |
| "epoch": 1.0444407912762343, | |
| "grad_norm": 1.7065175771713257, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5561, | |
| "step": 19060 | |
| }, | |
| { | |
| "epoch": 1.0455367417392734, | |
| "grad_norm": 1.6928189992904663, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5224, | |
| "step": 19080 | |
| }, | |
| { | |
| "epoch": 1.0466326922023124, | |
| "grad_norm": 1.4461798667907715, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5549, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 1.0477286426653516, | |
| "grad_norm": 3.3619306087493896, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6065, | |
| "step": 19120 | |
| }, | |
| { | |
| "epoch": 1.0488245931283906, | |
| "grad_norm": 1.8468629121780396, | |
| "learning_rate": 0.0001, | |
| "loss": 0.716, | |
| "step": 19140 | |
| }, | |
| { | |
| "epoch": 1.0499205435914296, | |
| "grad_norm": 1.7207484245300293, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6078, | |
| "step": 19160 | |
| }, | |
| { | |
| "epoch": 1.0510164940544688, | |
| "grad_norm": 1.6650727987289429, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6347, | |
| "step": 19180 | |
| }, | |
| { | |
| "epoch": 1.0521124445175078, | |
| "grad_norm": 1.3957616090774536, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6265, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 1.0532083949805469, | |
| "grad_norm": 1.4066413640975952, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6118, | |
| "step": 19220 | |
| }, | |
| { | |
| "epoch": 1.0543043454435859, | |
| "grad_norm": 1.8007709980010986, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6989, | |
| "step": 19240 | |
| }, | |
| { | |
| "epoch": 1.055400295906625, | |
| "grad_norm": 1.4759665727615356, | |
| "learning_rate": 0.0001, | |
| "loss": 0.621, | |
| "step": 19260 | |
| }, | |
| { | |
| "epoch": 1.056496246369664, | |
| "grad_norm": 1.7062383890151978, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6413, | |
| "step": 19280 | |
| }, | |
| { | |
| "epoch": 1.0575921968327031, | |
| "grad_norm": 1.5822961330413818, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5877, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 1.0586881472957423, | |
| "grad_norm": 1.7339930534362793, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6121, | |
| "step": 19320 | |
| }, | |
| { | |
| "epoch": 1.0597840977587814, | |
| "grad_norm": 1.3066824674606323, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6278, | |
| "step": 19340 | |
| }, | |
| { | |
| "epoch": 1.0608800482218204, | |
| "grad_norm": 1.6953777074813843, | |
| "learning_rate": 0.0001, | |
| "loss": 0.624, | |
| "step": 19360 | |
| }, | |
| { | |
| "epoch": 1.0619759986848594, | |
| "grad_norm": 1.5192081928253174, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6078, | |
| "step": 19380 | |
| }, | |
| { | |
| "epoch": 1.0630719491478986, | |
| "grad_norm": 1.5474629402160645, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7147, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 1.0641678996109376, | |
| "grad_norm": 1.5060781240463257, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6342, | |
| "step": 19420 | |
| }, | |
| { | |
| "epoch": 1.0652638500739766, | |
| "grad_norm": 1.4227863550186157, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5967, | |
| "step": 19440 | |
| }, | |
| { | |
| "epoch": 1.0663598005370156, | |
| "grad_norm": 1.4965442419052124, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6344, | |
| "step": 19460 | |
| }, | |
| { | |
| "epoch": 1.0674557510000549, | |
| "grad_norm": 1.4466912746429443, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6276, | |
| "step": 19480 | |
| }, | |
| { | |
| "epoch": 1.0685517014630939, | |
| "grad_norm": 1.7357358932495117, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5664, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.0696476519261329, | |
| "grad_norm": 1.6092090606689453, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6207, | |
| "step": 19520 | |
| }, | |
| { | |
| "epoch": 1.070743602389172, | |
| "grad_norm": 1.5991522073745728, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6113, | |
| "step": 19540 | |
| }, | |
| { | |
| "epoch": 1.071839552852211, | |
| "grad_norm": 1.320917010307312, | |
| "learning_rate": 0.0001, | |
| "loss": 0.577, | |
| "step": 19560 | |
| }, | |
| { | |
| "epoch": 1.0729355033152501, | |
| "grad_norm": 1.4419164657592773, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6622, | |
| "step": 19580 | |
| }, | |
| { | |
| "epoch": 1.0740314537782891, | |
| "grad_norm": 1.7298556566238403, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5721, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 1.0751274042413284, | |
| "grad_norm": 1.8013224601745605, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6079, | |
| "step": 19620 | |
| }, | |
| { | |
| "epoch": 1.0762233547043674, | |
| "grad_norm": 1.5226448774337769, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6681, | |
| "step": 19640 | |
| }, | |
| { | |
| "epoch": 1.0773193051674064, | |
| "grad_norm": 1.894225835800171, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6599, | |
| "step": 19660 | |
| }, | |
| { | |
| "epoch": 1.0784152556304456, | |
| "grad_norm": 1.58738112449646, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5796, | |
| "step": 19680 | |
| }, | |
| { | |
| "epoch": 1.0795112060934846, | |
| "grad_norm": 1.880391240119934, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6305, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 1.0806071565565236, | |
| "grad_norm": 1.4821720123291016, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5675, | |
| "step": 19720 | |
| }, | |
| { | |
| "epoch": 1.0817031070195626, | |
| "grad_norm": 1.504714846611023, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5888, | |
| "step": 19740 | |
| }, | |
| { | |
| "epoch": 1.0827990574826019, | |
| "grad_norm": 1.6745513677597046, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6199, | |
| "step": 19760 | |
| }, | |
| { | |
| "epoch": 1.0838950079456409, | |
| "grad_norm": 1.827014684677124, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5889, | |
| "step": 19780 | |
| }, | |
| { | |
| "epoch": 1.0849909584086799, | |
| "grad_norm": 1.6627857685089111, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5649, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 1.086086908871719, | |
| "grad_norm": 1.5660628080368042, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6814, | |
| "step": 19820 | |
| }, | |
| { | |
| "epoch": 1.0871828593347581, | |
| "grad_norm": 1.4713698625564575, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5789, | |
| "step": 19840 | |
| }, | |
| { | |
| "epoch": 1.0882788097977971, | |
| "grad_norm": 1.5290462970733643, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5894, | |
| "step": 19860 | |
| }, | |
| { | |
| "epoch": 1.0893747602608361, | |
| "grad_norm": 1.7101823091506958, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5732, | |
| "step": 19880 | |
| }, | |
| { | |
| "epoch": 1.0904707107238754, | |
| "grad_norm": 1.8752964735031128, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6358, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 1.0915666611869144, | |
| "grad_norm": 1.7808418273925781, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5942, | |
| "step": 19920 | |
| }, | |
| { | |
| "epoch": 1.0926626116499534, | |
| "grad_norm": 1.4797093868255615, | |
| "learning_rate": 0.0001, | |
| "loss": 0.605, | |
| "step": 19940 | |
| }, | |
| { | |
| "epoch": 1.0937585621129924, | |
| "grad_norm": 2.443544387817383, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7019, | |
| "step": 19960 | |
| }, | |
| { | |
| "epoch": 1.0948545125760316, | |
| "grad_norm": 1.4167999029159546, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5969, | |
| "step": 19980 | |
| }, | |
| { | |
| "epoch": 1.0959504630390706, | |
| "grad_norm": 1.3823322057724, | |
| "learning_rate": 0.0001, | |
| "loss": 0.634, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.0959504630390706, | |
| "eval_loss": 0.6163960099220276, | |
| "eval_runtime": 30555.9795, | |
| "eval_samples_per_second": 2.124, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 36.777449473248126, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.0970464135021096, | |
| "grad_norm": 1.5388779640197754, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5836, | |
| "step": 20020 | |
| }, | |
| { | |
| "epoch": 1.0981423639651489, | |
| "grad_norm": 1.9624749422073364, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5574, | |
| "step": 20040 | |
| }, | |
| { | |
| "epoch": 1.0992383144281879, | |
| "grad_norm": 1.4712483882904053, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6284, | |
| "step": 20060 | |
| }, | |
| { | |
| "epoch": 1.1003342648912269, | |
| "grad_norm": 1.4846380949020386, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5815, | |
| "step": 20080 | |
| }, | |
| { | |
| "epoch": 1.1014302153542659, | |
| "grad_norm": 1.4635918140411377, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6003, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 1.1025261658173051, | |
| "grad_norm": 1.775586485862732, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5997, | |
| "step": 20120 | |
| }, | |
| { | |
| "epoch": 1.1036221162803441, | |
| "grad_norm": 1.5434575080871582, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5924, | |
| "step": 20140 | |
| }, | |
| { | |
| "epoch": 1.1047180667433831, | |
| "grad_norm": 1.661482572555542, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6288, | |
| "step": 20160 | |
| }, | |
| { | |
| "epoch": 1.1058140172064224, | |
| "grad_norm": 1.706123948097229, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5624, | |
| "step": 20180 | |
| }, | |
| { | |
| "epoch": 1.1069099676694614, | |
| "grad_norm": 1.9980905055999756, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5577, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 1.1080059181325004, | |
| "grad_norm": 1.60072922706604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5947, | |
| "step": 20220 | |
| }, | |
| { | |
| "epoch": 1.1091018685955394, | |
| "grad_norm": 1.6381016969680786, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5183, | |
| "step": 20240 | |
| }, | |
| { | |
| "epoch": 1.1101978190585786, | |
| "grad_norm": 1.7384517192840576, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5752, | |
| "step": 20260 | |
| }, | |
| { | |
| "epoch": 1.1112937695216176, | |
| "grad_norm": 1.9785683155059814, | |
| "learning_rate": 0.0001, | |
| "loss": 0.576, | |
| "step": 20280 | |
| }, | |
| { | |
| "epoch": 1.1123897199846566, | |
| "grad_norm": 1.2318958044052124, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5853, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 1.1134856704476959, | |
| "grad_norm": 1.5406831502914429, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6059, | |
| "step": 20320 | |
| }, | |
| { | |
| "epoch": 1.1145816209107349, | |
| "grad_norm": 1.6655981540679932, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6834, | |
| "step": 20340 | |
| }, | |
| { | |
| "epoch": 1.1156775713737739, | |
| "grad_norm": 1.513757586479187, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6158, | |
| "step": 20360 | |
| }, | |
| { | |
| "epoch": 1.1167735218368129, | |
| "grad_norm": 2.1243133544921875, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5769, | |
| "step": 20380 | |
| }, | |
| { | |
| "epoch": 1.1178694722998521, | |
| "grad_norm": 1.6118996143341064, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5197, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 1.1189654227628911, | |
| "grad_norm": 1.8882734775543213, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5459, | |
| "step": 20420 | |
| }, | |
| { | |
| "epoch": 1.1200613732259301, | |
| "grad_norm": 1.8385454416275024, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5589, | |
| "step": 20440 | |
| }, | |
| { | |
| "epoch": 1.1211573236889691, | |
| "grad_norm": 1.7789567708969116, | |
| "learning_rate": 0.0001, | |
| "loss": 0.589, | |
| "step": 20460 | |
| }, | |
| { | |
| "epoch": 1.1222532741520084, | |
| "grad_norm": 1.839416265487671, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5979, | |
| "step": 20480 | |
| }, | |
| { | |
| "epoch": 1.1233492246150474, | |
| "grad_norm": 1.8148174285888672, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6179, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.1244451750780864, | |
| "grad_norm": 1.594193696975708, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6427, | |
| "step": 20520 | |
| }, | |
| { | |
| "epoch": 1.1255411255411256, | |
| "grad_norm": 1.5754518508911133, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5933, | |
| "step": 20540 | |
| }, | |
| { | |
| "epoch": 1.1266370760041646, | |
| "grad_norm": 1.7265543937683105, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6021, | |
| "step": 20560 | |
| }, | |
| { | |
| "epoch": 1.1277330264672036, | |
| "grad_norm": 1.4708410501480103, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5824, | |
| "step": 20580 | |
| }, | |
| { | |
| "epoch": 1.1288289769302429, | |
| "grad_norm": 1.7831743955612183, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5806, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 1.1299249273932819, | |
| "grad_norm": 1.530446171760559, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6523, | |
| "step": 20620 | |
| }, | |
| { | |
| "epoch": 1.1310208778563209, | |
| "grad_norm": 2.154409408569336, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5898, | |
| "step": 20640 | |
| }, | |
| { | |
| "epoch": 1.13211682831936, | |
| "grad_norm": 1.3791108131408691, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6321, | |
| "step": 20660 | |
| }, | |
| { | |
| "epoch": 1.1332127787823991, | |
| "grad_norm": 2.259727954864502, | |
| "learning_rate": 0.0001, | |
| "loss": 0.67, | |
| "step": 20680 | |
| }, | |
| { | |
| "epoch": 1.1343087292454381, | |
| "grad_norm": 1.7098194360733032, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5554, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 1.1354046797084771, | |
| "grad_norm": 1.5874308347702026, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5596, | |
| "step": 20720 | |
| }, | |
| { | |
| "epoch": 1.1365006301715161, | |
| "grad_norm": 1.9818806648254395, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6618, | |
| "step": 20740 | |
| }, | |
| { | |
| "epoch": 1.1375965806345554, | |
| "grad_norm": 1.4672831296920776, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5771, | |
| "step": 20760 | |
| }, | |
| { | |
| "epoch": 1.1386925310975944, | |
| "grad_norm": 1.746772050857544, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6085, | |
| "step": 20780 | |
| }, | |
| { | |
| "epoch": 1.1397884815606334, | |
| "grad_norm": 1.7367818355560303, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6401, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 1.1408844320236726, | |
| "grad_norm": 1.5269123315811157, | |
| "learning_rate": 0.0001, | |
| "loss": 0.678, | |
| "step": 20820 | |
| }, | |
| { | |
| "epoch": 1.1419803824867116, | |
| "grad_norm": 1.6921030282974243, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6292, | |
| "step": 20840 | |
| }, | |
| { | |
| "epoch": 1.1430763329497506, | |
| "grad_norm": 1.4888218641281128, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6034, | |
| "step": 20860 | |
| }, | |
| { | |
| "epoch": 1.1441722834127896, | |
| "grad_norm": 1.9028393030166626, | |
| "learning_rate": 0.0001, | |
| "loss": 0.604, | |
| "step": 20880 | |
| }, | |
| { | |
| "epoch": 1.1452682338758289, | |
| "grad_norm": 1.4702014923095703, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6375, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 1.1463641843388679, | |
| "grad_norm": 1.9169687032699585, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6173, | |
| "step": 20920 | |
| }, | |
| { | |
| "epoch": 1.147460134801907, | |
| "grad_norm": 1.6540066003799438, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5366, | |
| "step": 20940 | |
| }, | |
| { | |
| "epoch": 1.148556085264946, | |
| "grad_norm": 1.564468502998352, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5614, | |
| "step": 20960 | |
| }, | |
| { | |
| "epoch": 1.1496520357279851, | |
| "grad_norm": 1.75001060962677, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6408, | |
| "step": 20980 | |
| }, | |
| { | |
| "epoch": 1.1507479861910241, | |
| "grad_norm": 1.759342074394226, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6267, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.1507479861910241, | |
| "eval_loss": 0.6101906895637512, | |
| "eval_runtime": 30794.7182, | |
| "eval_samples_per_second": 2.107, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 46.360156536208414, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.1518439366540631, | |
| "grad_norm": 1.5255391597747803, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6105, | |
| "step": 21020 | |
| }, | |
| { | |
| "epoch": 1.1529398871171024, | |
| "grad_norm": 1.8833086490631104, | |
| "learning_rate": 0.0001, | |
| "loss": 0.628, | |
| "step": 21040 | |
| }, | |
| { | |
| "epoch": 1.1540358375801414, | |
| "grad_norm": 1.6667803525924683, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5758, | |
| "step": 21060 | |
| }, | |
| { | |
| "epoch": 1.1551317880431804, | |
| "grad_norm": 1.6798675060272217, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5869, | |
| "step": 21080 | |
| }, | |
| { | |
| "epoch": 1.1562277385062196, | |
| "grad_norm": 1.3532921075820923, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5978, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 1.1573236889692586, | |
| "grad_norm": 1.9075069427490234, | |
| "learning_rate": 0.0001, | |
| "loss": 0.636, | |
| "step": 21120 | |
| }, | |
| { | |
| "epoch": 1.1584196394322976, | |
| "grad_norm": 1.6051623821258545, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6624, | |
| "step": 21140 | |
| }, | |
| { | |
| "epoch": 1.1595155898953367, | |
| "grad_norm": 1.6509668827056885, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6098, | |
| "step": 21160 | |
| }, | |
| { | |
| "epoch": 1.1606115403583759, | |
| "grad_norm": 1.8513017892837524, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5753, | |
| "step": 21180 | |
| }, | |
| { | |
| "epoch": 1.161707490821415, | |
| "grad_norm": 1.7680573463439941, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5971, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 1.162803441284454, | |
| "grad_norm": 1.7778613567352295, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6127, | |
| "step": 21220 | |
| }, | |
| { | |
| "epoch": 1.163899391747493, | |
| "grad_norm": 1.524961233139038, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6181, | |
| "step": 21240 | |
| }, | |
| { | |
| "epoch": 1.1649953422105321, | |
| "grad_norm": 1.7401758432388306, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6575, | |
| "step": 21260 | |
| }, | |
| { | |
| "epoch": 1.1660912926735711, | |
| "grad_norm": 1.7773276567459106, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6264, | |
| "step": 21280 | |
| }, | |
| { | |
| "epoch": 1.1671872431366102, | |
| "grad_norm": 1.987033486366272, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5309, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 1.1682831935996494, | |
| "grad_norm": 1.7425816059112549, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6644, | |
| "step": 21320 | |
| }, | |
| { | |
| "epoch": 1.1693791440626884, | |
| "grad_norm": 1.6333574056625366, | |
| "learning_rate": 0.0001, | |
| "loss": 0.586, | |
| "step": 21340 | |
| }, | |
| { | |
| "epoch": 1.1704750945257274, | |
| "grad_norm": 1.6959367990493774, | |
| "learning_rate": 0.0001, | |
| "loss": 0.566, | |
| "step": 21360 | |
| }, | |
| { | |
| "epoch": 1.1715710449887666, | |
| "grad_norm": 1.6892461776733398, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5788, | |
| "step": 21380 | |
| }, | |
| { | |
| "epoch": 1.1726669954518056, | |
| "grad_norm": 1.805227279663086, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5984, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 1.1737629459148446, | |
| "grad_norm": 1.5054594278335571, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5851, | |
| "step": 21420 | |
| }, | |
| { | |
| "epoch": 1.1748588963778837, | |
| "grad_norm": 1.7826286554336548, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5872, | |
| "step": 21440 | |
| }, | |
| { | |
| "epoch": 1.1759548468409229, | |
| "grad_norm": 1.1451081037521362, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5533, | |
| "step": 21460 | |
| }, | |
| { | |
| "epoch": 1.177050797303962, | |
| "grad_norm": 1.425122618675232, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5934, | |
| "step": 21480 | |
| }, | |
| { | |
| "epoch": 1.178146747767001, | |
| "grad_norm": 1.6452502012252808, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6008, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.17924269823004, | |
| "grad_norm": 1.6565967798233032, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5945, | |
| "step": 21520 | |
| }, | |
| { | |
| "epoch": 1.1803386486930791, | |
| "grad_norm": 1.7541433572769165, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6243, | |
| "step": 21540 | |
| }, | |
| { | |
| "epoch": 1.1814345991561181, | |
| "grad_norm": 1.5369079113006592, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5968, | |
| "step": 21560 | |
| }, | |
| { | |
| "epoch": 1.1825305496191572, | |
| "grad_norm": 2.0941789150238037, | |
| "learning_rate": 0.0001, | |
| "loss": 0.599, | |
| "step": 21580 | |
| }, | |
| { | |
| "epoch": 1.1836265000821964, | |
| "grad_norm": 1.6641647815704346, | |
| "learning_rate": 0.0001, | |
| "loss": 0.634, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 1.1847224505452354, | |
| "grad_norm": 1.5256268978118896, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5939, | |
| "step": 21620 | |
| }, | |
| { | |
| "epoch": 1.1858184010082744, | |
| "grad_norm": 1.5720540285110474, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6005, | |
| "step": 21640 | |
| }, | |
| { | |
| "epoch": 1.1869143514713134, | |
| "grad_norm": 1.773973822593689, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5607, | |
| "step": 21660 | |
| }, | |
| { | |
| "epoch": 1.1880103019343526, | |
| "grad_norm": 1.6868877410888672, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6013, | |
| "step": 21680 | |
| }, | |
| { | |
| "epoch": 1.1891062523973916, | |
| "grad_norm": 1.518250584602356, | |
| "learning_rate": 0.0001, | |
| "loss": 0.612, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 1.1902022028604307, | |
| "grad_norm": 1.4469574689865112, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5684, | |
| "step": 21720 | |
| }, | |
| { | |
| "epoch": 1.1912981533234697, | |
| "grad_norm": 1.3651134967803955, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6275, | |
| "step": 21740 | |
| }, | |
| { | |
| "epoch": 1.192394103786509, | |
| "grad_norm": 1.1910673379898071, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6281, | |
| "step": 21760 | |
| }, | |
| { | |
| "epoch": 1.193490054249548, | |
| "grad_norm": 1.5071038007736206, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5909, | |
| "step": 21780 | |
| }, | |
| { | |
| "epoch": 1.194586004712587, | |
| "grad_norm": 1.3401362895965576, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6273, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 1.1956819551756261, | |
| "grad_norm": 1.3563752174377441, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5634, | |
| "step": 21820 | |
| }, | |
| { | |
| "epoch": 1.1967779056386652, | |
| "grad_norm": 1.5860759019851685, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5996, | |
| "step": 21840 | |
| }, | |
| { | |
| "epoch": 1.1978738561017042, | |
| "grad_norm": 1.6106479167938232, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6038, | |
| "step": 21860 | |
| }, | |
| { | |
| "epoch": 1.1989698065647434, | |
| "grad_norm": 1.2792236804962158, | |
| "learning_rate": 0.0001, | |
| "loss": 0.569, | |
| "step": 21880 | |
| }, | |
| { | |
| "epoch": 1.2000657570277824, | |
| "grad_norm": 1.7770174741744995, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6159, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 1.2011617074908214, | |
| "grad_norm": 1.522647738456726, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5756, | |
| "step": 21920 | |
| }, | |
| { | |
| "epoch": 1.2022576579538604, | |
| "grad_norm": 1.4393162727355957, | |
| "learning_rate": 0.0001, | |
| "loss": 0.598, | |
| "step": 21940 | |
| }, | |
| { | |
| "epoch": 1.2033536084168996, | |
| "grad_norm": 1.9221006631851196, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5736, | |
| "step": 21960 | |
| }, | |
| { | |
| "epoch": 1.2044495588799387, | |
| "grad_norm": 1.4930051565170288, | |
| "learning_rate": 0.0001, | |
| "loss": 0.591, | |
| "step": 21980 | |
| }, | |
| { | |
| "epoch": 1.2055455093429777, | |
| "grad_norm": 1.1012686491012573, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6139, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.2055455093429777, | |
| "eval_loss": 0.6032226085662842, | |
| "eval_runtime": 30533.3597, | |
| "eval_samples_per_second": 2.125, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 36.777449473248126, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.2066414598060167, | |
| "grad_norm": 1.501633644104004, | |
| "learning_rate": 0.0001, | |
| "loss": 0.55, | |
| "step": 22020 | |
| }, | |
| { | |
| "epoch": 1.207737410269056, | |
| "grad_norm": 1.7444618940353394, | |
| "learning_rate": 0.0001, | |
| "loss": 0.62, | |
| "step": 22040 | |
| }, | |
| { | |
| "epoch": 1.208833360732095, | |
| "grad_norm": 1.3666551113128662, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6087, | |
| "step": 22060 | |
| }, | |
| { | |
| "epoch": 1.209929311195134, | |
| "grad_norm": 1.5576726198196411, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5958, | |
| "step": 22080 | |
| }, | |
| { | |
| "epoch": 1.2110252616581731, | |
| "grad_norm": 1.457824945449829, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5588, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 1.2121212121212122, | |
| "grad_norm": 1.621120810508728, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5574, | |
| "step": 22120 | |
| }, | |
| { | |
| "epoch": 1.2132171625842512, | |
| "grad_norm": 1.1742050647735596, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5498, | |
| "step": 22140 | |
| }, | |
| { | |
| "epoch": 1.2143131130472902, | |
| "grad_norm": 1.3734312057495117, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5899, | |
| "step": 22160 | |
| }, | |
| { | |
| "epoch": 1.2154090635103294, | |
| "grad_norm": 2.046262741088867, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6574, | |
| "step": 22180 | |
| }, | |
| { | |
| "epoch": 1.2165050139733684, | |
| "grad_norm": 1.3114126920700073, | |
| "learning_rate": 0.0001, | |
| "loss": 0.607, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 1.2176009644364074, | |
| "grad_norm": 1.5335580110549927, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5828, | |
| "step": 22220 | |
| }, | |
| { | |
| "epoch": 1.2186969148994464, | |
| "grad_norm": 1.7492777109146118, | |
| "learning_rate": 0.0001, | |
| "loss": 0.557, | |
| "step": 22240 | |
| }, | |
| { | |
| "epoch": 1.2197928653624857, | |
| "grad_norm": 1.3159027099609375, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6923, | |
| "step": 22260 | |
| }, | |
| { | |
| "epoch": 1.2208888158255247, | |
| "grad_norm": 1.5170766115188599, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7194, | |
| "step": 22280 | |
| }, | |
| { | |
| "epoch": 1.2219847662885637, | |
| "grad_norm": 1.336846113204956, | |
| "learning_rate": 0.0001, | |
| "loss": 0.583, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 1.223080716751603, | |
| "grad_norm": 1.768999457359314, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7009, | |
| "step": 22320 | |
| }, | |
| { | |
| "epoch": 1.224176667214642, | |
| "grad_norm": 1.6113883256912231, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5747, | |
| "step": 22340 | |
| }, | |
| { | |
| "epoch": 1.225272617677681, | |
| "grad_norm": 1.4983850717544556, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5477, | |
| "step": 22360 | |
| }, | |
| { | |
| "epoch": 1.2263685681407201, | |
| "grad_norm": 1.380181908607483, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6003, | |
| "step": 22380 | |
| }, | |
| { | |
| "epoch": 1.2274645186037592, | |
| "grad_norm": 1.4921499490737915, | |
| "learning_rate": 0.0001, | |
| "loss": 0.586, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 1.2285604690667982, | |
| "grad_norm": 1.3056907653808594, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5393, | |
| "step": 22420 | |
| }, | |
| { | |
| "epoch": 1.2296564195298372, | |
| "grad_norm": 1.702541470527649, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6474, | |
| "step": 22440 | |
| }, | |
| { | |
| "epoch": 1.2307523699928764, | |
| "grad_norm": 1.703065276145935, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6102, | |
| "step": 22460 | |
| }, | |
| { | |
| "epoch": 1.2318483204559154, | |
| "grad_norm": 1.7823582887649536, | |
| "learning_rate": 0.0001, | |
| "loss": 0.529, | |
| "step": 22480 | |
| }, | |
| { | |
| "epoch": 1.2329442709189544, | |
| "grad_norm": 1.5001643896102905, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5936, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.2340402213819934, | |
| "grad_norm": 1.4515180587768555, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5817, | |
| "step": 22520 | |
| }, | |
| { | |
| "epoch": 1.2351361718450327, | |
| "grad_norm": 1.6166529655456543, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5901, | |
| "step": 22540 | |
| }, | |
| { | |
| "epoch": 1.2362321223080717, | |
| "grad_norm": 1.6117253303527832, | |
| "learning_rate": 0.0001, | |
| "loss": 0.668, | |
| "step": 22560 | |
| }, | |
| { | |
| "epoch": 1.2373280727711107, | |
| "grad_norm": 1.4674168825149536, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5783, | |
| "step": 22580 | |
| }, | |
| { | |
| "epoch": 1.23842402323415, | |
| "grad_norm": 1.5282671451568604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6069, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 1.239519973697189, | |
| "grad_norm": 1.446772575378418, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5645, | |
| "step": 22620 | |
| }, | |
| { | |
| "epoch": 1.240615924160228, | |
| "grad_norm": 1.7833497524261475, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5555, | |
| "step": 22640 | |
| }, | |
| { | |
| "epoch": 1.241711874623267, | |
| "grad_norm": 1.8573659658432007, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6258, | |
| "step": 22660 | |
| }, | |
| { | |
| "epoch": 1.2428078250863062, | |
| "grad_norm": 1.375735878944397, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5316, | |
| "step": 22680 | |
| }, | |
| { | |
| "epoch": 1.2439037755493452, | |
| "grad_norm": 1.4545280933380127, | |
| "learning_rate": 0.0001, | |
| "loss": 0.609, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 1.2449997260123842, | |
| "grad_norm": 1.499182105064392, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6205, | |
| "step": 22720 | |
| }, | |
| { | |
| "epoch": 1.2460956764754232, | |
| "grad_norm": 1.418739914894104, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5038, | |
| "step": 22740 | |
| }, | |
| { | |
| "epoch": 1.2471916269384624, | |
| "grad_norm": 1.4958001375198364, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5607, | |
| "step": 22760 | |
| }, | |
| { | |
| "epoch": 1.2482875774015014, | |
| "grad_norm": 1.7422837018966675, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5593, | |
| "step": 22780 | |
| }, | |
| { | |
| "epoch": 1.2493835278645404, | |
| "grad_norm": 1.536526083946228, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5518, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 1.2504794783275797, | |
| "grad_norm": 1.2415670156478882, | |
| "learning_rate": 0.0001, | |
| "loss": 0.625, | |
| "step": 22820 | |
| }, | |
| { | |
| "epoch": 1.2515754287906187, | |
| "grad_norm": 1.2609211206436157, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5213, | |
| "step": 22840 | |
| }, | |
| { | |
| "epoch": 1.2526713792536577, | |
| "grad_norm": 1.4843876361846924, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5618, | |
| "step": 22860 | |
| }, | |
| { | |
| "epoch": 1.253767329716697, | |
| "grad_norm": 1.7089099884033203, | |
| "learning_rate": 0.0001, | |
| "loss": 0.587, | |
| "step": 22880 | |
| }, | |
| { | |
| "epoch": 1.254863280179736, | |
| "grad_norm": 1.8894917964935303, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5952, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 1.255959230642775, | |
| "grad_norm": 1.3892401456832886, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5949, | |
| "step": 22920 | |
| }, | |
| { | |
| "epoch": 1.257055181105814, | |
| "grad_norm": 1.8688722848892212, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6524, | |
| "step": 22940 | |
| }, | |
| { | |
| "epoch": 1.2581511315688532, | |
| "grad_norm": 1.8726931810379028, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5547, | |
| "step": 22960 | |
| }, | |
| { | |
| "epoch": 1.2592470820318922, | |
| "grad_norm": 1.9214690923690796, | |
| "learning_rate": 0.0001, | |
| "loss": 0.618, | |
| "step": 22980 | |
| }, | |
| { | |
| "epoch": 1.2603430324949312, | |
| "grad_norm": 1.6148467063903809, | |
| "learning_rate": 0.0001, | |
| "loss": 0.4822, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.2603430324949312, | |
| "eval_loss": 0.599087655544281, | |
| "eval_runtime": 30610.1389, | |
| "eval_samples_per_second": 2.12, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 40.07337206794192, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.2614389829579702, | |
| "grad_norm": 1.536855936050415, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5852, | |
| "step": 23020 | |
| }, | |
| { | |
| "epoch": 1.2625349334210094, | |
| "grad_norm": 1.884334683418274, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6474, | |
| "step": 23040 | |
| }, | |
| { | |
| "epoch": 1.2636308838840484, | |
| "grad_norm": 1.359174132347107, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6204, | |
| "step": 23060 | |
| }, | |
| { | |
| "epoch": 1.2647268343470874, | |
| "grad_norm": 1.7376055717468262, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5524, | |
| "step": 23080 | |
| }, | |
| { | |
| "epoch": 1.2658227848101267, | |
| "grad_norm": 1.6594703197479248, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5396, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 1.2669187352731657, | |
| "grad_norm": 1.4215294122695923, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5842, | |
| "step": 23120 | |
| }, | |
| { | |
| "epoch": 1.2680146857362047, | |
| "grad_norm": 1.7680842876434326, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5625, | |
| "step": 23140 | |
| }, | |
| { | |
| "epoch": 1.269110636199244, | |
| "grad_norm": 1.3353180885314941, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5775, | |
| "step": 23160 | |
| }, | |
| { | |
| "epoch": 1.270206586662283, | |
| "grad_norm": 1.450649619102478, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5319, | |
| "step": 23180 | |
| }, | |
| { | |
| "epoch": 1.271302537125322, | |
| "grad_norm": 1.7398178577423096, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5615, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 1.272398487588361, | |
| "grad_norm": 1.6940994262695312, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6069, | |
| "step": 23220 | |
| }, | |
| { | |
| "epoch": 1.2734944380514, | |
| "grad_norm": 1.899994969367981, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6491, | |
| "step": 23240 | |
| }, | |
| { | |
| "epoch": 1.2745903885144392, | |
| "grad_norm": 1.457036018371582, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5762, | |
| "step": 23260 | |
| }, | |
| { | |
| "epoch": 1.2756863389774782, | |
| "grad_norm": 1.4215611219406128, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5966, | |
| "step": 23280 | |
| }, | |
| { | |
| "epoch": 1.2767822894405172, | |
| "grad_norm": 1.7165329456329346, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5897, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 1.2778782399035564, | |
| "grad_norm": 1.524688720703125, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6541, | |
| "step": 23320 | |
| }, | |
| { | |
| "epoch": 1.2789741903665954, | |
| "grad_norm": 1.3125251531600952, | |
| "learning_rate": 0.0001, | |
| "loss": 0.521, | |
| "step": 23340 | |
| }, | |
| { | |
| "epoch": 1.2800701408296344, | |
| "grad_norm": 1.5787118673324585, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5453, | |
| "step": 23360 | |
| }, | |
| { | |
| "epoch": 1.2811660912926737, | |
| "grad_norm": 1.636098027229309, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5606, | |
| "step": 23380 | |
| }, | |
| { | |
| "epoch": 1.2822620417557127, | |
| "grad_norm": 1.4963462352752686, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6251, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 1.2833579922187517, | |
| "grad_norm": 1.3213664293289185, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6293, | |
| "step": 23420 | |
| }, | |
| { | |
| "epoch": 1.284453942681791, | |
| "grad_norm": 1.6737782955169678, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6499, | |
| "step": 23440 | |
| }, | |
| { | |
| "epoch": 1.28554989314483, | |
| "grad_norm": 1.525976300239563, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6073, | |
| "step": 23460 | |
| }, | |
| { | |
| "epoch": 1.286645843607869, | |
| "grad_norm": 1.3534733057022095, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5434, | |
| "step": 23480 | |
| }, | |
| { | |
| "epoch": 1.287741794070908, | |
| "grad_norm": 1.8090375661849976, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6806, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.288837744533947, | |
| "grad_norm": 1.7110000848770142, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5678, | |
| "step": 23520 | |
| }, | |
| { | |
| "epoch": 1.2899336949969862, | |
| "grad_norm": 1.6300121545791626, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6674, | |
| "step": 23540 | |
| }, | |
| { | |
| "epoch": 1.2910296454600252, | |
| "grad_norm": 1.4068278074264526, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5294, | |
| "step": 23560 | |
| }, | |
| { | |
| "epoch": 1.2921255959230642, | |
| "grad_norm": 1.6015020608901978, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5791, | |
| "step": 23580 | |
| }, | |
| { | |
| "epoch": 1.2932215463861034, | |
| "grad_norm": 1.7282171249389648, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6358, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 1.2943174968491424, | |
| "grad_norm": 1.3395479917526245, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5972, | |
| "step": 23620 | |
| }, | |
| { | |
| "epoch": 1.2954134473121814, | |
| "grad_norm": 1.5393882989883423, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6243, | |
| "step": 23640 | |
| }, | |
| { | |
| "epoch": 1.2965093977752207, | |
| "grad_norm": 2.0010182857513428, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6047, | |
| "step": 23660 | |
| }, | |
| { | |
| "epoch": 1.2976053482382597, | |
| "grad_norm": 1.742031455039978, | |
| "learning_rate": 0.0001, | |
| "loss": 0.586, | |
| "step": 23680 | |
| }, | |
| { | |
| "epoch": 1.2987012987012987, | |
| "grad_norm": 1.5596591234207153, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5545, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 1.2997972491643377, | |
| "grad_norm": 1.2098394632339478, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6264, | |
| "step": 23720 | |
| }, | |
| { | |
| "epoch": 1.3008931996273767, | |
| "grad_norm": 1.5962443351745605, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5827, | |
| "step": 23740 | |
| }, | |
| { | |
| "epoch": 1.301989150090416, | |
| "grad_norm": 1.7482990026474, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6113, | |
| "step": 23760 | |
| }, | |
| { | |
| "epoch": 1.303085100553455, | |
| "grad_norm": 1.6832690238952637, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5231, | |
| "step": 23780 | |
| }, | |
| { | |
| "epoch": 1.304181051016494, | |
| "grad_norm": 1.2904006242752075, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5472, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 1.3052770014795332, | |
| "grad_norm": 1.260377287864685, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5555, | |
| "step": 23820 | |
| }, | |
| { | |
| "epoch": 1.3063729519425722, | |
| "grad_norm": 1.6346126794815063, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5247, | |
| "step": 23840 | |
| }, | |
| { | |
| "epoch": 1.3074689024056112, | |
| "grad_norm": 1.5854252576828003, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5784, | |
| "step": 23860 | |
| }, | |
| { | |
| "epoch": 1.3085648528686504, | |
| "grad_norm": 1.754293441772461, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6555, | |
| "step": 23880 | |
| }, | |
| { | |
| "epoch": 1.3096608033316894, | |
| "grad_norm": 1.642504096031189, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5822, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 1.3107567537947284, | |
| "grad_norm": 1.4986265897750854, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6246, | |
| "step": 23920 | |
| }, | |
| { | |
| "epoch": 1.3118527042577677, | |
| "grad_norm": 1.6175062656402588, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5616, | |
| "step": 23940 | |
| }, | |
| { | |
| "epoch": 1.3129486547208067, | |
| "grad_norm": 1.6189128160476685, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6108, | |
| "step": 23960 | |
| }, | |
| { | |
| "epoch": 1.3140446051838457, | |
| "grad_norm": 1.6187801361083984, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5992, | |
| "step": 23980 | |
| }, | |
| { | |
| "epoch": 1.3151405556468847, | |
| "grad_norm": 1.396136999130249, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5409, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.3151405556468847, | |
| "eval_loss": 0.5925264358520508, | |
| "eval_runtime": 30722.0761, | |
| "eval_samples_per_second": 2.112, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 50.53154519589281, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.3162365061099237, | |
| "grad_norm": 1.4604203701019287, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5795, | |
| "step": 24020 | |
| }, | |
| { | |
| "epoch": 1.317332456572963, | |
| "grad_norm": 1.667830228805542, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5799, | |
| "step": 24040 | |
| }, | |
| { | |
| "epoch": 1.318428407036002, | |
| "grad_norm": 1.4066877365112305, | |
| "learning_rate": 0.0001, | |
| "loss": 0.603, | |
| "step": 24060 | |
| }, | |
| { | |
| "epoch": 1.319524357499041, | |
| "grad_norm": 1.6986689567565918, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5943, | |
| "step": 24080 | |
| }, | |
| { | |
| "epoch": 1.3206203079620802, | |
| "grad_norm": 2.00864839553833, | |
| "learning_rate": 0.0001, | |
| "loss": 0.646, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 1.3217162584251192, | |
| "grad_norm": 1.3557894229888916, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6324, | |
| "step": 24120 | |
| }, | |
| { | |
| "epoch": 1.3228122088881582, | |
| "grad_norm": 1.442983865737915, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5326, | |
| "step": 24140 | |
| }, | |
| { | |
| "epoch": 1.3239081593511974, | |
| "grad_norm": 1.4924156665802002, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5464, | |
| "step": 24160 | |
| }, | |
| { | |
| "epoch": 1.3250041098142364, | |
| "grad_norm": 1.408599615097046, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5989, | |
| "step": 24180 | |
| }, | |
| { | |
| "epoch": 1.3261000602772754, | |
| "grad_norm": 1.6432658433914185, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6034, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 1.3271960107403145, | |
| "grad_norm": 1.5687427520751953, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5222, | |
| "step": 24220 | |
| }, | |
| { | |
| "epoch": 1.3282919612033537, | |
| "grad_norm": 1.3878777027130127, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5913, | |
| "step": 24240 | |
| }, | |
| { | |
| "epoch": 1.3293879116663927, | |
| "grad_norm": 1.276931643486023, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5509, | |
| "step": 24260 | |
| }, | |
| { | |
| "epoch": 1.3304838621294317, | |
| "grad_norm": 1.5601953268051147, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6077, | |
| "step": 24280 | |
| }, | |
| { | |
| "epoch": 1.3315798125924707, | |
| "grad_norm": 1.9250099658966064, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5629, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 1.33267576305551, | |
| "grad_norm": 1.314794659614563, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6085, | |
| "step": 24320 | |
| }, | |
| { | |
| "epoch": 1.333771713518549, | |
| "grad_norm": 1.4445682764053345, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5725, | |
| "step": 24340 | |
| }, | |
| { | |
| "epoch": 1.334867663981588, | |
| "grad_norm": 1.6029905080795288, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6448, | |
| "step": 24360 | |
| }, | |
| { | |
| "epoch": 1.3359636144446272, | |
| "grad_norm": 1.969078540802002, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5397, | |
| "step": 24380 | |
| }, | |
| { | |
| "epoch": 1.3370595649076662, | |
| "grad_norm": 1.900762677192688, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6243, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 1.3381555153707052, | |
| "grad_norm": 1.8829255104064941, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5744, | |
| "step": 24420 | |
| }, | |
| { | |
| "epoch": 1.3392514658337444, | |
| "grad_norm": 1.4927318096160889, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6023, | |
| "step": 24440 | |
| }, | |
| { | |
| "epoch": 1.3403474162967834, | |
| "grad_norm": 1.7608602046966553, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5654, | |
| "step": 24460 | |
| }, | |
| { | |
| "epoch": 1.3414433667598225, | |
| "grad_norm": 1.2257969379425049, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5953, | |
| "step": 24480 | |
| }, | |
| { | |
| "epoch": 1.3425393172228615, | |
| "grad_norm": 1.5768262147903442, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5972, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.3436352676859005, | |
| "grad_norm": 1.5148476362228394, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6313, | |
| "step": 24520 | |
| }, | |
| { | |
| "epoch": 1.3447312181489397, | |
| "grad_norm": 1.347442865371704, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5867, | |
| "step": 24540 | |
| }, | |
| { | |
| "epoch": 1.3458271686119787, | |
| "grad_norm": 1.3003042936325073, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5709, | |
| "step": 24560 | |
| }, | |
| { | |
| "epoch": 1.3469231190750177, | |
| "grad_norm": 1.5577054023742676, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5719, | |
| "step": 24580 | |
| }, | |
| { | |
| "epoch": 1.348019069538057, | |
| "grad_norm": 1.6141449213027954, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5684, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 1.349115020001096, | |
| "grad_norm": 1.515576958656311, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5421, | |
| "step": 24620 | |
| }, | |
| { | |
| "epoch": 1.350210970464135, | |
| "grad_norm": 1.62236487865448, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6338, | |
| "step": 24640 | |
| }, | |
| { | |
| "epoch": 1.3513069209271742, | |
| "grad_norm": 1.5557656288146973, | |
| "learning_rate": 0.0001, | |
| "loss": 0.55, | |
| "step": 24660 | |
| }, | |
| { | |
| "epoch": 1.3524028713902132, | |
| "grad_norm": 1.9079481363296509, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6352, | |
| "step": 24680 | |
| }, | |
| { | |
| "epoch": 1.3534988218532522, | |
| "grad_norm": 1.3640868663787842, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6484, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 1.3545947723162914, | |
| "grad_norm": 1.831858515739441, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5577, | |
| "step": 24720 | |
| }, | |
| { | |
| "epoch": 1.3556907227793304, | |
| "grad_norm": 1.5431773662567139, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5918, | |
| "step": 24740 | |
| }, | |
| { | |
| "epoch": 1.3567866732423695, | |
| "grad_norm": 1.1655539274215698, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5892, | |
| "step": 24760 | |
| }, | |
| { | |
| "epoch": 1.3578826237054085, | |
| "grad_norm": 1.7395954132080078, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5416, | |
| "step": 24780 | |
| }, | |
| { | |
| "epoch": 1.3589785741684475, | |
| "grad_norm": 1.89164400100708, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5708, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 1.3600745246314867, | |
| "grad_norm": 1.5013233423233032, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5234, | |
| "step": 24820 | |
| }, | |
| { | |
| "epoch": 1.3611704750945257, | |
| "grad_norm": 1.4959980249404907, | |
| "learning_rate": 0.0001, | |
| "loss": 0.7578, | |
| "step": 24840 | |
| }, | |
| { | |
| "epoch": 1.3622664255575647, | |
| "grad_norm": 1.3962562084197998, | |
| "learning_rate": 0.0001, | |
| "loss": 0.564, | |
| "step": 24860 | |
| }, | |
| { | |
| "epoch": 1.363362376020604, | |
| "grad_norm": 1.370695948600769, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5533, | |
| "step": 24880 | |
| }, | |
| { | |
| "epoch": 1.364458326483643, | |
| "grad_norm": 1.629671573638916, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5463, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 1.365554276946682, | |
| "grad_norm": 1.6115111112594604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5837, | |
| "step": 24920 | |
| }, | |
| { | |
| "epoch": 1.3666502274097212, | |
| "grad_norm": 1.756898283958435, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5784, | |
| "step": 24940 | |
| }, | |
| { | |
| "epoch": 1.3677461778727602, | |
| "grad_norm": 1.3999930620193481, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6007, | |
| "step": 24960 | |
| }, | |
| { | |
| "epoch": 1.3688421283357992, | |
| "grad_norm": 1.3553202152252197, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5996, | |
| "step": 24980 | |
| }, | |
| { | |
| "epoch": 1.3699380787988382, | |
| "grad_norm": 1.3020998239517212, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5006, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.3699380787988382, | |
| "eval_loss": 0.5878660678863525, | |
| "eval_runtime": 30562.5187, | |
| "eval_samples_per_second": 2.123, | |
| "eval_steps_per_second": 0.066, | |
| "eval_wer": 44.19117823283905, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.3710340292618772, | |
| "grad_norm": 1.622175931930542, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6049, | |
| "step": 25020 | |
| }, | |
| { | |
| "epoch": 1.3721299797249165, | |
| "grad_norm": 1.6178827285766602, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6272, | |
| "step": 25040 | |
| }, | |
| { | |
| "epoch": 1.3732259301879555, | |
| "grad_norm": 1.9687530994415283, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6194, | |
| "step": 25060 | |
| }, | |
| { | |
| "epoch": 1.3743218806509945, | |
| "grad_norm": 1.5638937950134277, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6033, | |
| "step": 25080 | |
| }, | |
| { | |
| "epoch": 1.3754178311140337, | |
| "grad_norm": 1.5316487550735474, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6847, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 1.3765137815770727, | |
| "grad_norm": 2.431041717529297, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5975, | |
| "step": 25120 | |
| }, | |
| { | |
| "epoch": 1.3776097320401117, | |
| "grad_norm": 1.2522825002670288, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6188, | |
| "step": 25140 | |
| }, | |
| { | |
| "epoch": 1.378705682503151, | |
| "grad_norm": 1.428832769393921, | |
| "learning_rate": 0.0001, | |
| "loss": 0.4743, | |
| "step": 25160 | |
| }, | |
| { | |
| "epoch": 1.37980163296619, | |
| "grad_norm": 1.6147892475128174, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6463, | |
| "step": 25180 | |
| }, | |
| { | |
| "epoch": 1.380897583429229, | |
| "grad_norm": 1.6648330688476562, | |
| "learning_rate": 0.0001, | |
| "loss": 0.556, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 1.3819935338922682, | |
| "grad_norm": 1.4745285511016846, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5997, | |
| "step": 25220 | |
| }, | |
| { | |
| "epoch": 1.3830894843553072, | |
| "grad_norm": 1.433664321899414, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5288, | |
| "step": 25240 | |
| }, | |
| { | |
| "epoch": 1.3841854348183462, | |
| "grad_norm": 1.264054775238037, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6075, | |
| "step": 25260 | |
| }, | |
| { | |
| "epoch": 1.3852813852813852, | |
| "grad_norm": 1.7683794498443604, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6011, | |
| "step": 25280 | |
| }, | |
| { | |
| "epoch": 1.3863773357444242, | |
| "grad_norm": 1.5316460132598877, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5799, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 1.3874732862074635, | |
| "grad_norm": 3.361269950866699, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6577, | |
| "step": 25320 | |
| }, | |
| { | |
| "epoch": 1.3885692366705025, | |
| "grad_norm": 1.7918739318847656, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6106, | |
| "step": 25340 | |
| }, | |
| { | |
| "epoch": 1.3896651871335415, | |
| "grad_norm": 1.5828975439071655, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5683, | |
| "step": 25360 | |
| }, | |
| { | |
| "epoch": 1.3907611375965807, | |
| "grad_norm": 1.9565653800964355, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5368, | |
| "step": 25380 | |
| }, | |
| { | |
| "epoch": 1.3918570880596197, | |
| "grad_norm": 1.446603775024414, | |
| "learning_rate": 0.0001, | |
| "loss": 0.551, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 1.3929530385226587, | |
| "grad_norm": 1.3377403020858765, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5889, | |
| "step": 25420 | |
| }, | |
| { | |
| "epoch": 1.394048988985698, | |
| "grad_norm": 1.5956981182098389, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5962, | |
| "step": 25440 | |
| }, | |
| { | |
| "epoch": 1.395144939448737, | |
| "grad_norm": 2.122570276260376, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6044, | |
| "step": 25460 | |
| }, | |
| { | |
| "epoch": 1.396240889911776, | |
| "grad_norm": 1.8314157724380493, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5612, | |
| "step": 25480 | |
| }, | |
| { | |
| "epoch": 1.397336840374815, | |
| "grad_norm": 1.327453851699829, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5214, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.3984327908378542, | |
| "grad_norm": 2.1016182899475098, | |
| "learning_rate": 0.0001, | |
| "loss": 0.594, | |
| "step": 25520 | |
| }, | |
| { | |
| "epoch": 1.3995287413008932, | |
| "grad_norm": 1.5892709493637085, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5927, | |
| "step": 25540 | |
| }, | |
| { | |
| "epoch": 1.4006246917639322, | |
| "grad_norm": 1.8178175687789917, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5204, | |
| "step": 25560 | |
| }, | |
| { | |
| "epoch": 1.4017206422269712, | |
| "grad_norm": 1.3808900117874146, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5434, | |
| "step": 25580 | |
| }, | |
| { | |
| "epoch": 1.4028165926900105, | |
| "grad_norm": 1.4849821329116821, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6208, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 1.4039125431530495, | |
| "grad_norm": 1.5404740571975708, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5492, | |
| "step": 25620 | |
| }, | |
| { | |
| "epoch": 1.4050084936160885, | |
| "grad_norm": 1.4663268327713013, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5275, | |
| "step": 25640 | |
| }, | |
| { | |
| "epoch": 1.4061044440791277, | |
| "grad_norm": 1.5518896579742432, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5966, | |
| "step": 25660 | |
| }, | |
| { | |
| "epoch": 1.4072003945421667, | |
| "grad_norm": 1.5495002269744873, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5451, | |
| "step": 25680 | |
| }, | |
| { | |
| "epoch": 1.4082963450052057, | |
| "grad_norm": 1.5400805473327637, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6543, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 1.409392295468245, | |
| "grad_norm": 1.5806297063827515, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5849, | |
| "step": 25720 | |
| }, | |
| { | |
| "epoch": 1.410488245931284, | |
| "grad_norm": 1.574846863746643, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5653, | |
| "step": 25740 | |
| }, | |
| { | |
| "epoch": 1.411584196394323, | |
| "grad_norm": 1.4742170572280884, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5173, | |
| "step": 25760 | |
| }, | |
| { | |
| "epoch": 1.412680146857362, | |
| "grad_norm": 2.009706735610962, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5467, | |
| "step": 25780 | |
| }, | |
| { | |
| "epoch": 1.413776097320401, | |
| "grad_norm": 1.9192876815795898, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6211, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 1.4148720477834402, | |
| "grad_norm": 1.2425312995910645, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5398, | |
| "step": 25820 | |
| }, | |
| { | |
| "epoch": 1.4159679982464792, | |
| "grad_norm": 1.8032817840576172, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6143, | |
| "step": 25840 | |
| }, | |
| { | |
| "epoch": 1.4170639487095182, | |
| "grad_norm": 1.5889620780944824, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6361, | |
| "step": 25860 | |
| }, | |
| { | |
| "epoch": 1.4181598991725575, | |
| "grad_norm": 1.3312866687774658, | |
| "learning_rate": 0.0001, | |
| "loss": 0.637, | |
| "step": 25880 | |
| }, | |
| { | |
| "epoch": 1.4192558496355965, | |
| "grad_norm": 1.4662201404571533, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6206, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 1.4203518000986355, | |
| "grad_norm": 1.7590441703796387, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5637, | |
| "step": 25920 | |
| }, | |
| { | |
| "epoch": 1.4214477505616747, | |
| "grad_norm": 1.5360428094863892, | |
| "learning_rate": 0.0001, | |
| "loss": 0.6033, | |
| "step": 25940 | |
| }, | |
| { | |
| "epoch": 1.4225437010247137, | |
| "grad_norm": 1.6273736953735352, | |
| "learning_rate": 0.0001, | |
| "loss": 0.559, | |
| "step": 25960 | |
| }, | |
| { | |
| "epoch": 1.4236396514877527, | |
| "grad_norm": 1.3700859546661377, | |
| "learning_rate": 0.0001, | |
| "loss": 0.637, | |
| "step": 25980 | |
| }, | |
| { | |
| "epoch": 1.4247356019507917, | |
| "grad_norm": 1.419872522354126, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5674, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.4247356019507917, | |
| "eval_loss": 0.5845187306404114, | |
| "eval_runtime": 30268.7098, | |
| "eval_samples_per_second": 2.144, | |
| "eval_steps_per_second": 0.067, | |
| "eval_wer": 33.472796184515516, | |
| "step": 26000 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 54747, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.278115609397166e+21, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |