| { | |
| "best_metric": 2.2782892974889872, | |
| "best_model_checkpoint": "./whisper-medium_new_data/checkpoint-8000", | |
| "epoch": 3.261312678353037, | |
| "eval_steps": 1000, | |
| "global_step": 8000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010191602119853241, | |
| "grad_norm": 29.000947952270508, | |
| "learning_rate": 2.1000000000000003e-07, | |
| "loss": 3.777, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.020383204239706482, | |
| "grad_norm": 16.35529136657715, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 3.379, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.030574806359559722, | |
| "grad_norm": 11.864029884338379, | |
| "learning_rate": 7.1e-07, | |
| "loss": 2.7049, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.040766408479412965, | |
| "grad_norm": 9.735739707946777, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 2.1205, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.050958010599266204, | |
| "grad_norm": 7.762232780456543, | |
| "learning_rate": 1.21e-06, | |
| "loss": 1.731, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.061149612719119444, | |
| "grad_norm": 8.16604995727539, | |
| "learning_rate": 1.46e-06, | |
| "loss": 1.4442, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07134121483897268, | |
| "grad_norm": 7.53511905670166, | |
| "learning_rate": 1.7100000000000004e-06, | |
| "loss": 1.244, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.08153281695882593, | |
| "grad_norm": 7.6938557624816895, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 1.0535, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09172441907867916, | |
| "grad_norm": 7.2217912673950195, | |
| "learning_rate": 2.21e-06, | |
| "loss": 0.9467, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.10191602119853241, | |
| "grad_norm": 6.709696292877197, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.8746, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.11210762331838565, | |
| "grad_norm": 6.30186653137207, | |
| "learning_rate": 2.7100000000000003e-06, | |
| "loss": 0.7813, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.12229922543823889, | |
| "grad_norm": 7.001236915588379, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.7312, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.13249082755809213, | |
| "grad_norm": 6.0768232345581055, | |
| "learning_rate": 3.21e-06, | |
| "loss": 0.6602, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.14268242967794537, | |
| "grad_norm": 6.0217790603637695, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.6294, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.15287403179779863, | |
| "grad_norm": 6.146008491516113, | |
| "learning_rate": 3.7100000000000005e-06, | |
| "loss": 0.6184, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.16306563391765186, | |
| "grad_norm": 5.988298416137695, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.5581, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1732572360375051, | |
| "grad_norm": 6.24209451675415, | |
| "learning_rate": 4.21e-06, | |
| "loss": 0.526, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.18344883815735832, | |
| "grad_norm": 5.728849411010742, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.4973, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.19364044027721158, | |
| "grad_norm": 5.146097660064697, | |
| "learning_rate": 4.71e-06, | |
| "loss": 0.4877, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.20383204239706482, | |
| "grad_norm": 5.125807285308838, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.4718, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.21402364451691805, | |
| "grad_norm": 5.3624653816223145, | |
| "learning_rate": 5.210000000000001e-06, | |
| "loss": 0.455, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.2242152466367713, | |
| "grad_norm": 5.1377854347229, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.4426, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.23440684875662454, | |
| "grad_norm": 5.11069393157959, | |
| "learning_rate": 5.71e-06, | |
| "loss": 0.4176, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.24459845087647777, | |
| "grad_norm": 5.100331783294678, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.4044, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.254790052996331, | |
| "grad_norm": 5.771068572998047, | |
| "learning_rate": 6.210000000000001e-06, | |
| "loss": 0.3994, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.26498165511618427, | |
| "grad_norm": 5.4431986808776855, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.366, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2751732572360375, | |
| "grad_norm": 5.490501880645752, | |
| "learning_rate": 6.710000000000001e-06, | |
| "loss": 0.3742, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.28536485935589073, | |
| "grad_norm": 5.428119659423828, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.3577, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.295556461475744, | |
| "grad_norm": 4.746588230133057, | |
| "learning_rate": 7.2100000000000004e-06, | |
| "loss": 0.3465, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.30574806359559725, | |
| "grad_norm": 4.743890762329102, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.3226, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.31593966571545046, | |
| "grad_norm": 5.242619037628174, | |
| "learning_rate": 7.71e-06, | |
| "loss": 0.3185, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.3261312678353037, | |
| "grad_norm": 4.178338050842285, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.32, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.336322869955157, | |
| "grad_norm": 4.850783824920654, | |
| "learning_rate": 8.210000000000001e-06, | |
| "loss": 0.327, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.3465144720750102, | |
| "grad_norm": 5.147852897644043, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.3334, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.35670607419486344, | |
| "grad_norm": 4.720470905303955, | |
| "learning_rate": 8.710000000000001e-06, | |
| "loss": 0.2821, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.36689767631471665, | |
| "grad_norm": 4.410367965698242, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.2894, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3770892784345699, | |
| "grad_norm": 4.329741477966309, | |
| "learning_rate": 9.210000000000002e-06, | |
| "loss": 0.2792, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.38728088055442317, | |
| "grad_norm": 4.4180803298950195, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.2842, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.3974724826742764, | |
| "grad_norm": 4.213972568511963, | |
| "learning_rate": 9.71e-06, | |
| "loss": 0.2854, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.40766408479412963, | |
| "grad_norm": 3.454677104949951, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.2733, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.40766408479412963, | |
| "eval_loss": 0.2585422396659851, | |
| "eval_runtime": 2430.2946, | |
| "eval_samples_per_second": 4.195, | |
| "eval_steps_per_second": 0.131, | |
| "eval_wer": 32.5923736874671, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4178556869139829, | |
| "grad_norm": 3.999333620071411, | |
| "learning_rate": 9.970000000000001e-06, | |
| "loss": 0.2649, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.4280472890338361, | |
| "grad_norm": 3.658658504486084, | |
| "learning_rate": 9.934285714285715e-06, | |
| "loss": 0.2494, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.43823889115368936, | |
| "grad_norm": 4.042171955108643, | |
| "learning_rate": 9.89857142857143e-06, | |
| "loss": 0.2626, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.4484304932735426, | |
| "grad_norm": 4.251903057098389, | |
| "learning_rate": 9.862857142857144e-06, | |
| "loss": 0.2457, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.4586220953933958, | |
| "grad_norm": 4.49993896484375, | |
| "learning_rate": 9.827142857142859e-06, | |
| "loss": 0.2598, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.4688136975132491, | |
| "grad_norm": 4.043424129486084, | |
| "learning_rate": 9.791428571428571e-06, | |
| "loss": 0.2447, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.47900529963310234, | |
| "grad_norm": 3.787559747695923, | |
| "learning_rate": 9.755714285714286e-06, | |
| "loss": 0.2328, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.48919690175295555, | |
| "grad_norm": 3.875075340270996, | |
| "learning_rate": 9.72e-06, | |
| "loss": 0.2359, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.4993885038728088, | |
| "grad_norm": 4.148637294769287, | |
| "learning_rate": 9.684285714285715e-06, | |
| "loss": 0.2234, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.509580105992662, | |
| "grad_norm": 3.9952499866485596, | |
| "learning_rate": 9.648571428571429e-06, | |
| "loss": 0.225, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.5197717081125153, | |
| "grad_norm": 3.0705387592315674, | |
| "learning_rate": 9.612857142857144e-06, | |
| "loss": 0.2183, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.5299633102323685, | |
| "grad_norm": 3.5826079845428467, | |
| "learning_rate": 9.577142857142858e-06, | |
| "loss": 0.2125, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.5401549123522218, | |
| "grad_norm": 3.478602170944214, | |
| "learning_rate": 9.541428571428572e-06, | |
| "loss": 0.2067, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.550346514472075, | |
| "grad_norm": 3.646801233291626, | |
| "learning_rate": 9.505714285714287e-06, | |
| "loss": 0.2132, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.5605381165919282, | |
| "grad_norm": 4.240411758422852, | |
| "learning_rate": 9.47e-06, | |
| "loss": 0.2077, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.5707297187117815, | |
| "grad_norm": 3.531076669692993, | |
| "learning_rate": 9.434285714285714e-06, | |
| "loss": 0.1978, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.5809213208316347, | |
| "grad_norm": 4.578576564788818, | |
| "learning_rate": 9.39857142857143e-06, | |
| "loss": 0.1976, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.591112922951488, | |
| "grad_norm": 3.237501382827759, | |
| "learning_rate": 9.362857142857143e-06, | |
| "loss": 0.1829, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.6013045250713412, | |
| "grad_norm": 3.9498815536499023, | |
| "learning_rate": 9.327142857142857e-06, | |
| "loss": 0.1926, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.6114961271911945, | |
| "grad_norm": 3.386265277862549, | |
| "learning_rate": 9.291428571428572e-06, | |
| "loss": 0.1829, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6216877293110477, | |
| "grad_norm": 3.7739858627319336, | |
| "learning_rate": 9.255714285714286e-06, | |
| "loss": 0.1807, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.6318793314309009, | |
| "grad_norm": 2.9203953742980957, | |
| "learning_rate": 9.220000000000002e-06, | |
| "loss": 0.1816, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.6420709335507542, | |
| "grad_norm": 3.465407371520996, | |
| "learning_rate": 9.184285714285715e-06, | |
| "loss": 0.1739, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.6522625356706074, | |
| "grad_norm": 3.1187219619750977, | |
| "learning_rate": 9.148571428571429e-06, | |
| "loss": 0.177, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.6624541377904607, | |
| "grad_norm": 3.4814960956573486, | |
| "learning_rate": 9.112857142857142e-06, | |
| "loss": 0.1718, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.672645739910314, | |
| "grad_norm": 3.3516340255737305, | |
| "learning_rate": 9.077142857142858e-06, | |
| "loss": 0.1682, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.6828373420301671, | |
| "grad_norm": 3.460319757461548, | |
| "learning_rate": 9.041428571428572e-06, | |
| "loss": 0.1645, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.6930289441500204, | |
| "grad_norm": 2.6608901023864746, | |
| "learning_rate": 9.005714285714287e-06, | |
| "loss": 0.1721, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.7032205462698736, | |
| "grad_norm": 3.2124173641204834, | |
| "learning_rate": 8.97e-06, | |
| "loss": 0.1647, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.7134121483897269, | |
| "grad_norm": 3.007030725479126, | |
| "learning_rate": 8.934285714285716e-06, | |
| "loss": 0.163, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.7236037505095801, | |
| "grad_norm": 2.7804579734802246, | |
| "learning_rate": 8.89857142857143e-06, | |
| "loss": 0.1732, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.7337953526294333, | |
| "grad_norm": 3.615187644958496, | |
| "learning_rate": 8.862857142857143e-06, | |
| "loss": 0.1578, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.7439869547492866, | |
| "grad_norm": 2.8256614208221436, | |
| "learning_rate": 8.827142857142857e-06, | |
| "loss": 0.1553, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.7541785568691398, | |
| "grad_norm": 2.9422616958618164, | |
| "learning_rate": 8.791428571428572e-06, | |
| "loss": 0.1483, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.7643701589889931, | |
| "grad_norm": 2.9863955974578857, | |
| "learning_rate": 8.755714285714286e-06, | |
| "loss": 0.1446, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.7745617611088463, | |
| "grad_norm": 3.2235851287841797, | |
| "learning_rate": 8.720000000000001e-06, | |
| "loss": 0.1488, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.7847533632286996, | |
| "grad_norm": 3.379194974899292, | |
| "learning_rate": 8.684285714285715e-06, | |
| "loss": 0.1433, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.7949449653485527, | |
| "grad_norm": 3.080751895904541, | |
| "learning_rate": 8.64857142857143e-06, | |
| "loss": 0.1428, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.805136567468406, | |
| "grad_norm": 2.6511690616607666, | |
| "learning_rate": 8.612857142857144e-06, | |
| "loss": 0.1351, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.8153281695882593, | |
| "grad_norm": 3.0043816566467285, | |
| "learning_rate": 8.577142857142858e-06, | |
| "loss": 0.1527, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8153281695882593, | |
| "eval_loss": 0.12457678467035294, | |
| "eval_runtime": 2416.9884, | |
| "eval_samples_per_second": 4.218, | |
| "eval_steps_per_second": 0.132, | |
| "eval_wer": 16.723770120886936, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8255197717081125, | |
| "grad_norm": 2.428893804550171, | |
| "learning_rate": 8.541428571428571e-06, | |
| "loss": 0.1329, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.8357113738279658, | |
| "grad_norm": 2.5849223136901855, | |
| "learning_rate": 8.505714285714287e-06, | |
| "loss": 0.1355, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.845902975947819, | |
| "grad_norm": 2.7230465412139893, | |
| "learning_rate": 8.47e-06, | |
| "loss": 0.1409, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.8560945780676722, | |
| "grad_norm": 2.889012336730957, | |
| "learning_rate": 8.434285714285716e-06, | |
| "loss": 0.1339, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.8662861801875255, | |
| "grad_norm": 2.4034271240234375, | |
| "learning_rate": 8.39857142857143e-06, | |
| "loss": 0.1397, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.8764777823073787, | |
| "grad_norm": 2.2975339889526367, | |
| "learning_rate": 8.362857142857143e-06, | |
| "loss": 0.1254, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.886669384427232, | |
| "grad_norm": 2.7864370346069336, | |
| "learning_rate": 8.327142857142858e-06, | |
| "loss": 0.1298, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.8968609865470852, | |
| "grad_norm": 3.3211889266967773, | |
| "learning_rate": 8.291428571428572e-06, | |
| "loss": 0.1195, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.9070525886669385, | |
| "grad_norm": 2.9740707874298096, | |
| "learning_rate": 8.255714285714287e-06, | |
| "loss": 0.1236, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.9172441907867916, | |
| "grad_norm": 2.8324942588806152, | |
| "learning_rate": 8.220000000000001e-06, | |
| "loss": 0.1256, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.9274357929066449, | |
| "grad_norm": 2.7244651317596436, | |
| "learning_rate": 8.184285714285715e-06, | |
| "loss": 0.1232, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.9376273950264982, | |
| "grad_norm": 2.6477315425872803, | |
| "learning_rate": 8.148571428571428e-06, | |
| "loss": 0.1194, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.9478189971463514, | |
| "grad_norm": 2.737881660461426, | |
| "learning_rate": 8.112857142857144e-06, | |
| "loss": 0.1184, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.9580105992662047, | |
| "grad_norm": 3.096505641937256, | |
| "learning_rate": 8.077142857142857e-06, | |
| "loss": 0.1316, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.9682022013860578, | |
| "grad_norm": 2.7507877349853516, | |
| "learning_rate": 8.041428571428573e-06, | |
| "loss": 0.109, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.9783938035059111, | |
| "grad_norm": 2.7306649684906006, | |
| "learning_rate": 8.005714285714286e-06, | |
| "loss": 0.1263, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.9885854056257644, | |
| "grad_norm": 3.1028060913085938, | |
| "learning_rate": 7.970000000000002e-06, | |
| "loss": 0.1135, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.9987770077456176, | |
| "grad_norm": 2.456820011138916, | |
| "learning_rate": 7.934285714285715e-06, | |
| "loss": 0.1192, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.0089686098654709, | |
| "grad_norm": 2.0729124546051025, | |
| "learning_rate": 7.898571428571429e-06, | |
| "loss": 0.0789, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.019160211985324, | |
| "grad_norm": 2.0284624099731445, | |
| "learning_rate": 7.862857142857143e-06, | |
| "loss": 0.0741, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.0293518141051774, | |
| "grad_norm": 1.9724788665771484, | |
| "learning_rate": 7.827142857142858e-06, | |
| "loss": 0.068, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.0395434162250305, | |
| "grad_norm": 2.1046948432922363, | |
| "learning_rate": 7.791428571428572e-06, | |
| "loss": 0.0785, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.049735018344884, | |
| "grad_norm": 2.5051779747009277, | |
| "learning_rate": 7.755714285714287e-06, | |
| "loss": 0.0788, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.059926620464737, | |
| "grad_norm": 2.253950357437134, | |
| "learning_rate": 7.72e-06, | |
| "loss": 0.0748, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.0701182225845902, | |
| "grad_norm": 2.03273868560791, | |
| "learning_rate": 7.684285714285716e-06, | |
| "loss": 0.0795, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.0803098247044436, | |
| "grad_norm": 1.9357993602752686, | |
| "learning_rate": 7.64857142857143e-06, | |
| "loss": 0.0688, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.0905014268242967, | |
| "grad_norm": 2.5450971126556396, | |
| "learning_rate": 7.612857142857143e-06, | |
| "loss": 0.0807, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.10069302894415, | |
| "grad_norm": 2.2401187419891357, | |
| "learning_rate": 7.577142857142857e-06, | |
| "loss": 0.0741, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.1108846310640033, | |
| "grad_norm": 2.1422953605651855, | |
| "learning_rate": 7.5414285714285715e-06, | |
| "loss": 0.0761, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.1210762331838564, | |
| "grad_norm": 2.204946994781494, | |
| "learning_rate": 7.505714285714286e-06, | |
| "loss": 0.0726, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.1312678353037098, | |
| "grad_norm": 2.2459845542907715, | |
| "learning_rate": 7.4700000000000005e-06, | |
| "loss": 0.0751, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.141459437423563, | |
| "grad_norm": 2.2025928497314453, | |
| "learning_rate": 7.434285714285715e-06, | |
| "loss": 0.074, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.1516510395434163, | |
| "grad_norm": 1.94180166721344, | |
| "learning_rate": 7.3985714285714295e-06, | |
| "loss": 0.0768, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.1618426416632694, | |
| "grad_norm": 2.4948198795318604, | |
| "learning_rate": 7.362857142857144e-06, | |
| "loss": 0.0737, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.1720342437831226, | |
| "grad_norm": 2.136427640914917, | |
| "learning_rate": 7.3271428571428585e-06, | |
| "loss": 0.0711, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.182225845902976, | |
| "grad_norm": 1.7289948463439941, | |
| "learning_rate": 7.291428571428571e-06, | |
| "loss": 0.0713, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.1924174480228291, | |
| "grad_norm": 2.254673957824707, | |
| "learning_rate": 7.255714285714286e-06, | |
| "loss": 0.0696, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.2026090501426825, | |
| "grad_norm": 2.0539391040802, | |
| "learning_rate": 7.22e-06, | |
| "loss": 0.0669, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.2128006522625356, | |
| "grad_norm": 2.1277451515197754, | |
| "learning_rate": 7.184285714285715e-06, | |
| "loss": 0.0682, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.222992254382389, | |
| "grad_norm": 1.7704741954803467, | |
| "learning_rate": 7.148571428571429e-06, | |
| "loss": 0.0655, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.222992254382389, | |
| "eval_loss": 0.07764188200235367, | |
| "eval_runtime": 2406.4856, | |
| "eval_samples_per_second": 4.237, | |
| "eval_steps_per_second": 0.133, | |
| "eval_wer": 10.566755631077823, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2331838565022422, | |
| "grad_norm": 2.0602128505706787, | |
| "learning_rate": 7.112857142857144e-06, | |
| "loss": 0.0667, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.2433754586220953, | |
| "grad_norm": 2.0332977771759033, | |
| "learning_rate": 7.077142857142858e-06, | |
| "loss": 0.0656, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.2535670607419487, | |
| "grad_norm": 2.4050822257995605, | |
| "learning_rate": 7.041428571428572e-06, | |
| "loss": 0.0624, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.2637586628618018, | |
| "grad_norm": 2.1430561542510986, | |
| "learning_rate": 7.0057142857142865e-06, | |
| "loss": 0.0805, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.273950264981655, | |
| "grad_norm": 2.0134341716766357, | |
| "learning_rate": 6.97e-06, | |
| "loss": 0.0664, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.2841418671015083, | |
| "grad_norm": 2.217721462249756, | |
| "learning_rate": 6.934285714285715e-06, | |
| "loss": 0.0745, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.2943334692213617, | |
| "grad_norm": 1.5617090463638306, | |
| "learning_rate": 6.898571428571429e-06, | |
| "loss": 0.0648, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.3045250713412149, | |
| "grad_norm": 1.9269059896469116, | |
| "learning_rate": 6.862857142857144e-06, | |
| "loss": 0.0618, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.314716673461068, | |
| "grad_norm": 2.0952699184417725, | |
| "learning_rate": 6.827142857142857e-06, | |
| "loss": 0.0706, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.3249082755809214, | |
| "grad_norm": 1.764196753501892, | |
| "learning_rate": 6.791428571428572e-06, | |
| "loss": 0.0658, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.3350998777007745, | |
| "grad_norm": 2.4948887825012207, | |
| "learning_rate": 6.755714285714286e-06, | |
| "loss": 0.0573, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.3452914798206277, | |
| "grad_norm": 2.2995638847351074, | |
| "learning_rate": 6.720000000000001e-06, | |
| "loss": 0.0584, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.355483081940481, | |
| "grad_norm": 1.6779873371124268, | |
| "learning_rate": 6.684285714285715e-06, | |
| "loss": 0.0618, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.3656746840603342, | |
| "grad_norm": 2.435180425643921, | |
| "learning_rate": 6.648571428571429e-06, | |
| "loss": 0.0611, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.3758662861801876, | |
| "grad_norm": 1.7600586414337158, | |
| "learning_rate": 6.612857142857143e-06, | |
| "loss": 0.0638, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.3860578883000407, | |
| "grad_norm": 2.179009199142456, | |
| "learning_rate": 6.577142857142857e-06, | |
| "loss": 0.0577, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.396249490419894, | |
| "grad_norm": 1.8130829334259033, | |
| "learning_rate": 6.541428571428572e-06, | |
| "loss": 0.0572, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.4064410925397473, | |
| "grad_norm": 2.2357475757598877, | |
| "learning_rate": 6.505714285714286e-06, | |
| "loss": 0.0617, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.4166326946596004, | |
| "grad_norm": 2.523681163787842, | |
| "learning_rate": 6.470000000000001e-06, | |
| "loss": 0.0592, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.4268242967794538, | |
| "grad_norm": 1.9920995235443115, | |
| "learning_rate": 6.434285714285715e-06, | |
| "loss": 0.0594, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.437015898899307, | |
| "grad_norm": 1.876678705215454, | |
| "learning_rate": 6.39857142857143e-06, | |
| "loss": 0.0588, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.4472075010191603, | |
| "grad_norm": 2.1072115898132324, | |
| "learning_rate": 6.3628571428571426e-06, | |
| "loss": 0.0584, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.4573991031390134, | |
| "grad_norm": 1.9492688179016113, | |
| "learning_rate": 6.327142857142857e-06, | |
| "loss": 0.0619, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.4675907052588668, | |
| "grad_norm": 1.9974353313446045, | |
| "learning_rate": 6.2914285714285716e-06, | |
| "loss": 0.0526, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.47778230737872, | |
| "grad_norm": 2.0630576610565186, | |
| "learning_rate": 6.255714285714286e-06, | |
| "loss": 0.0587, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.487973909498573, | |
| "grad_norm": 2.312988042831421, | |
| "learning_rate": 6.220000000000001e-06, | |
| "loss": 0.0539, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.4981655116184265, | |
| "grad_norm": 2.3259403705596924, | |
| "learning_rate": 6.184285714285715e-06, | |
| "loss": 0.0547, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.5083571137382796, | |
| "grad_norm": 1.7665891647338867, | |
| "learning_rate": 6.14857142857143e-06, | |
| "loss": 0.0546, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.5185487158581328, | |
| "grad_norm": 1.8702497482299805, | |
| "learning_rate": 6.112857142857144e-06, | |
| "loss": 0.059, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.5287403179779862, | |
| "grad_norm": 1.725127935409546, | |
| "learning_rate": 6.077142857142858e-06, | |
| "loss": 0.0593, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.5389319200978395, | |
| "grad_norm": 1.6888962984085083, | |
| "learning_rate": 6.0414285714285714e-06, | |
| "loss": 0.0492, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.5491235222176927, | |
| "grad_norm": 1.8041000366210938, | |
| "learning_rate": 6.005714285714286e-06, | |
| "loss": 0.0524, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.5593151243375458, | |
| "grad_norm": 2.0858354568481445, | |
| "learning_rate": 5.9700000000000004e-06, | |
| "loss": 0.0532, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.5695067264573992, | |
| "grad_norm": 2.3499724864959717, | |
| "learning_rate": 5.934285714285715e-06, | |
| "loss": 0.0509, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.5796983285772523, | |
| "grad_norm": 1.6974126100540161, | |
| "learning_rate": 5.8985714285714295e-06, | |
| "loss": 0.0525, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.5898899306971055, | |
| "grad_norm": 2.386164426803589, | |
| "learning_rate": 5.862857142857143e-06, | |
| "loss": 0.056, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.6000815328169589, | |
| "grad_norm": 2.2181949615478516, | |
| "learning_rate": 5.827142857142858e-06, | |
| "loss": 0.0532, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.610273134936812, | |
| "grad_norm": 2.883762836456299, | |
| "learning_rate": 5.791428571428572e-06, | |
| "loss": 0.0554, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.6204647370566652, | |
| "grad_norm": 2.2483813762664795, | |
| "learning_rate": 5.755714285714287e-06, | |
| "loss": 0.053, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.6306563391765185, | |
| "grad_norm": 1.992173433303833, | |
| "learning_rate": 5.72e-06, | |
| "loss": 0.0455, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.6306563391765185, | |
| "eval_loss": 0.05135625973343849, | |
| "eval_runtime": 2404.8469, | |
| "eval_samples_per_second": 4.24, | |
| "eval_steps_per_second": 0.133, | |
| "eval_wer": 6.767451954600445, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.640847941296372, | |
| "grad_norm": 1.7579491138458252, | |
| "learning_rate": 5.684285714285715e-06, | |
| "loss": 0.0521, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.651039543416225, | |
| "grad_norm": 1.9420897960662842, | |
| "learning_rate": 5.6485714285714285e-06, | |
| "loss": 0.0542, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.6612311455360782, | |
| "grad_norm": 2.2500171661376953, | |
| "learning_rate": 5.612857142857143e-06, | |
| "loss": 0.0516, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.6714227476559316, | |
| "grad_norm": 1.7925150394439697, | |
| "learning_rate": 5.5771428571428575e-06, | |
| "loss": 0.0493, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.6816143497757847, | |
| "grad_norm": 2.5120983123779297, | |
| "learning_rate": 5.541428571428572e-06, | |
| "loss": 0.0496, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.6918059518956379, | |
| "grad_norm": 1.6827411651611328, | |
| "learning_rate": 5.5057142857142865e-06, | |
| "loss": 0.048, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.7019975540154912, | |
| "grad_norm": 1.951802134513855, | |
| "learning_rate": 5.470000000000001e-06, | |
| "loss": 0.0491, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.7121891561353446, | |
| "grad_norm": 1.688637375831604, | |
| "learning_rate": 5.4342857142857155e-06, | |
| "loss": 0.0515, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.7223807582551978, | |
| "grad_norm": 2.034071207046509, | |
| "learning_rate": 5.398571428571428e-06, | |
| "loss": 0.0516, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.732572360375051, | |
| "grad_norm": 1.9581401348114014, | |
| "learning_rate": 5.362857142857143e-06, | |
| "loss": 0.0536, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.7427639624949043, | |
| "grad_norm": 1.8898799419403076, | |
| "learning_rate": 5.327142857142857e-06, | |
| "loss": 0.0491, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.7529555646147574, | |
| "grad_norm": 2.1761457920074463, | |
| "learning_rate": 5.291428571428572e-06, | |
| "loss": 0.0496, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.7631471667346106, | |
| "grad_norm": 1.9841150045394897, | |
| "learning_rate": 5.255714285714286e-06, | |
| "loss": 0.0469, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.773338768854464, | |
| "grad_norm": 1.3026888370513916, | |
| "learning_rate": 5.220000000000001e-06, | |
| "loss": 0.0482, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.7835303709743173, | |
| "grad_norm": 1.3047564029693604, | |
| "learning_rate": 5.184285714285715e-06, | |
| "loss": 0.0493, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.7937219730941703, | |
| "grad_norm": 1.6902995109558105, | |
| "learning_rate": 5.14857142857143e-06, | |
| "loss": 0.0494, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.8039135752140236, | |
| "grad_norm": 2.262953758239746, | |
| "learning_rate": 5.112857142857143e-06, | |
| "loss": 0.0476, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.814105177333877, | |
| "grad_norm": 1.777170181274414, | |
| "learning_rate": 5.077142857142857e-06, | |
| "loss": 0.0471, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.8242967794537301, | |
| "grad_norm": 1.9193710088729858, | |
| "learning_rate": 5.041428571428572e-06, | |
| "loss": 0.0467, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.8344883815735833, | |
| "grad_norm": 1.8790152072906494, | |
| "learning_rate": 5.005714285714286e-06, | |
| "loss": 0.0487, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.8446799836934367, | |
| "grad_norm": 2.3190362453460693, | |
| "learning_rate": 4.970000000000001e-06, | |
| "loss": 0.0466, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.8548715858132898, | |
| "grad_norm": 1.821876049041748, | |
| "learning_rate": 4.934285714285715e-06, | |
| "loss": 0.0456, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.865063187933143, | |
| "grad_norm": 1.3371105194091797, | |
| "learning_rate": 4.898571428571429e-06, | |
| "loss": 0.0414, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.8752547900529963, | |
| "grad_norm": 1.5593416690826416, | |
| "learning_rate": 4.862857142857143e-06, | |
| "loss": 0.0435, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.8854463921728497, | |
| "grad_norm": 1.9036990404129028, | |
| "learning_rate": 4.827142857142858e-06, | |
| "loss": 0.0448, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.8956379942927029, | |
| "grad_norm": 1.8190467357635498, | |
| "learning_rate": 4.7914285714285715e-06, | |
| "loss": 0.0388, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.905829596412556, | |
| "grad_norm": 1.7529155015945435, | |
| "learning_rate": 4.755714285714286e-06, | |
| "loss": 0.0421, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.9160211985324094, | |
| "grad_norm": 2.084097385406494, | |
| "learning_rate": 4.7200000000000005e-06, | |
| "loss": 0.0464, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.9262128006522625, | |
| "grad_norm": 1.569948673248291, | |
| "learning_rate": 4.684285714285714e-06, | |
| "loss": 0.0438, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.9364044027721157, | |
| "grad_norm": 2.0869407653808594, | |
| "learning_rate": 4.648571428571429e-06, | |
| "loss": 0.0421, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.946596004891969, | |
| "grad_norm": 1.8481653928756714, | |
| "learning_rate": 4.612857142857143e-06, | |
| "loss": 0.0429, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.9567876070118224, | |
| "grad_norm": 2.0351133346557617, | |
| "learning_rate": 4.577142857142858e-06, | |
| "loss": 0.0431, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.9669792091316753, | |
| "grad_norm": 1.88765549659729, | |
| "learning_rate": 4.541428571428571e-06, | |
| "loss": 0.0448, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.9771708112515287, | |
| "grad_norm": 1.8413023948669434, | |
| "learning_rate": 4.505714285714286e-06, | |
| "loss": 0.0391, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.987362413371382, | |
| "grad_norm": 2.0287082195281982, | |
| "learning_rate": 4.47e-06, | |
| "loss": 0.0417, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.9975540154912352, | |
| "grad_norm": 2.1331636905670166, | |
| "learning_rate": 4.434285714285715e-06, | |
| "loss": 0.0434, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.0077456176110884, | |
| "grad_norm": 1.133825421333313, | |
| "learning_rate": 4.3985714285714286e-06, | |
| "loss": 0.0246, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.0179372197309418, | |
| "grad_norm": 1.3025981187820435, | |
| "learning_rate": 4.362857142857143e-06, | |
| "loss": 0.017, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.028128821850795, | |
| "grad_norm": 1.008670687675476, | |
| "learning_rate": 4.327142857142858e-06, | |
| "loss": 0.0174, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.038320423970648, | |
| "grad_norm": 0.9607629179954529, | |
| "learning_rate": 4.291428571428572e-06, | |
| "loss": 0.0162, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.038320423970648, | |
| "eval_loss": 0.0353175513446331, | |
| "eval_runtime": 2408.5196, | |
| "eval_samples_per_second": 4.233, | |
| "eval_steps_per_second": 0.132, | |
| "eval_wer": 4.477157079135229, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.0485120260905014, | |
| "grad_norm": 1.6613744497299194, | |
| "learning_rate": 4.255714285714286e-06, | |
| "loss": 0.0213, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 2.058703628210355, | |
| "grad_norm": 0.6007469296455383, | |
| "learning_rate": 4.22e-06, | |
| "loss": 0.015, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.0688952303302077, | |
| "grad_norm": 1.5992757081985474, | |
| "learning_rate": 4.184285714285715e-06, | |
| "loss": 0.0191, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 2.079086832450061, | |
| "grad_norm": 0.913711667060852, | |
| "learning_rate": 4.148571428571429e-06, | |
| "loss": 0.0155, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.0892784345699145, | |
| "grad_norm": 1.1569514274597168, | |
| "learning_rate": 4.112857142857144e-06, | |
| "loss": 0.0164, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 2.099470036689768, | |
| "grad_norm": 1.0935266017913818, | |
| "learning_rate": 4.0771428571428574e-06, | |
| "loss": 0.0146, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.1096616388096208, | |
| "grad_norm": 0.8869621157646179, | |
| "learning_rate": 4.041428571428572e-06, | |
| "loss": 0.0168, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 2.119853240929474, | |
| "grad_norm": 1.3779696226119995, | |
| "learning_rate": 4.0057142857142864e-06, | |
| "loss": 0.0164, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.1300448430493275, | |
| "grad_norm": 1.2132755517959595, | |
| "learning_rate": 3.97e-06, | |
| "loss": 0.0178, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 2.1402364451691804, | |
| "grad_norm": 1.8169926404953003, | |
| "learning_rate": 3.934285714285715e-06, | |
| "loss": 0.0181, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.150428047289034, | |
| "grad_norm": 0.7188290357589722, | |
| "learning_rate": 3.898571428571429e-06, | |
| "loss": 0.0172, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 2.160619649408887, | |
| "grad_norm": 1.0363893508911133, | |
| "learning_rate": 3.862857142857143e-06, | |
| "loss": 0.0151, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.1708112515287405, | |
| "grad_norm": 0.7467776536941528, | |
| "learning_rate": 3.827142857142857e-06, | |
| "loss": 0.0159, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 2.1810028536485935, | |
| "grad_norm": 1.4140534400939941, | |
| "learning_rate": 3.7914285714285722e-06, | |
| "loss": 0.0179, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.191194455768447, | |
| "grad_norm": 0.831071138381958, | |
| "learning_rate": 3.755714285714286e-06, | |
| "loss": 0.018, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 2.2013860578883, | |
| "grad_norm": 0.978151261806488, | |
| "learning_rate": 3.7200000000000004e-06, | |
| "loss": 0.0164, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.211577660008153, | |
| "grad_norm": 1.2061834335327148, | |
| "learning_rate": 3.684285714285715e-06, | |
| "loss": 0.0155, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 2.2217692621280065, | |
| "grad_norm": 1.0183775424957275, | |
| "learning_rate": 3.648571428571429e-06, | |
| "loss": 0.0166, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.23196086424786, | |
| "grad_norm": 0.7747207283973694, | |
| "learning_rate": 3.612857142857143e-06, | |
| "loss": 0.017, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 2.242152466367713, | |
| "grad_norm": 1.3631898164749146, | |
| "learning_rate": 3.5771428571428576e-06, | |
| "loss": 0.0147, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.252344068487566, | |
| "grad_norm": 2.3850202560424805, | |
| "learning_rate": 3.5414285714285716e-06, | |
| "loss": 0.0145, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 2.2625356706074196, | |
| "grad_norm": 1.2588169574737549, | |
| "learning_rate": 3.505714285714286e-06, | |
| "loss": 0.0191, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.2727272727272725, | |
| "grad_norm": 1.3466730117797852, | |
| "learning_rate": 3.4700000000000002e-06, | |
| "loss": 0.0177, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 2.282918874847126, | |
| "grad_norm": 1.3090860843658447, | |
| "learning_rate": 3.4342857142857143e-06, | |
| "loss": 0.014, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.2931104769669792, | |
| "grad_norm": 1.4974130392074585, | |
| "learning_rate": 3.398571428571429e-06, | |
| "loss": 0.0152, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 2.3033020790868326, | |
| "grad_norm": 1.342842936515808, | |
| "learning_rate": 3.3628571428571433e-06, | |
| "loss": 0.0165, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.3134936812066855, | |
| "grad_norm": 0.9690020084381104, | |
| "learning_rate": 3.327142857142858e-06, | |
| "loss": 0.0146, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 2.323685283326539, | |
| "grad_norm": 0.4699741303920746, | |
| "learning_rate": 3.2914285714285715e-06, | |
| "loss": 0.0178, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.3338768854463923, | |
| "grad_norm": 1.166608214378357, | |
| "learning_rate": 3.255714285714286e-06, | |
| "loss": 0.0152, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 2.344068487566245, | |
| "grad_norm": 0.7220166921615601, | |
| "learning_rate": 3.2200000000000005e-06, | |
| "loss": 0.0142, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.3542600896860986, | |
| "grad_norm": 0.8474377989768982, | |
| "learning_rate": 3.1842857142857146e-06, | |
| "loss": 0.0186, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 2.364451691805952, | |
| "grad_norm": 1.246468186378479, | |
| "learning_rate": 3.1485714285714287e-06, | |
| "loss": 0.0143, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.3746432939258053, | |
| "grad_norm": 0.9267088770866394, | |
| "learning_rate": 3.112857142857143e-06, | |
| "loss": 0.0157, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 2.3848348960456582, | |
| "grad_norm": 1.0727224349975586, | |
| "learning_rate": 3.0771428571428573e-06, | |
| "loss": 0.0145, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.3950264981655116, | |
| "grad_norm": 1.3142344951629639, | |
| "learning_rate": 3.0414285714285718e-06, | |
| "loss": 0.0133, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.405218100285365, | |
| "grad_norm": 1.3349863290786743, | |
| "learning_rate": 3.005714285714286e-06, | |
| "loss": 0.0137, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.415409702405218, | |
| "grad_norm": 0.7742011547088623, | |
| "learning_rate": 2.97e-06, | |
| "loss": 0.0149, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 2.4256013045250713, | |
| "grad_norm": 1.1534104347229004, | |
| "learning_rate": 2.9342857142857144e-06, | |
| "loss": 0.0156, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.4357929066449246, | |
| "grad_norm": 1.1034826040267944, | |
| "learning_rate": 2.898571428571429e-06, | |
| "loss": 0.0136, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 2.445984508764778, | |
| "grad_norm": 1.3665175437927246, | |
| "learning_rate": 2.8628571428571435e-06, | |
| "loss": 0.0129, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.445984508764778, | |
| "eval_loss": 0.027366982772946358, | |
| "eval_runtime": 2410.6864, | |
| "eval_samples_per_second": 4.23, | |
| "eval_steps_per_second": 0.132, | |
| "eval_wer": 3.436365819195996, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.456176110884631, | |
| "grad_norm": 0.9552819728851318, | |
| "learning_rate": 2.827142857142857e-06, | |
| "loss": 0.015, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 2.4663677130044843, | |
| "grad_norm": 0.8750997185707092, | |
| "learning_rate": 2.7914285714285716e-06, | |
| "loss": 0.014, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.4765593151243377, | |
| "grad_norm": 1.3504140377044678, | |
| "learning_rate": 2.755714285714286e-06, | |
| "loss": 0.0172, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 2.4867509172441906, | |
| "grad_norm": 1.59817373752594, | |
| "learning_rate": 2.7200000000000002e-06, | |
| "loss": 0.0153, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.496942519364044, | |
| "grad_norm": 1.6574933528900146, | |
| "learning_rate": 2.6842857142857143e-06, | |
| "loss": 0.0167, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 2.5071341214838974, | |
| "grad_norm": 0.7508680820465088, | |
| "learning_rate": 2.648571428571429e-06, | |
| "loss": 0.013, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.5173257236037507, | |
| "grad_norm": 1.0604500770568848, | |
| "learning_rate": 2.612857142857143e-06, | |
| "loss": 0.0162, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 2.5275173257236037, | |
| "grad_norm": 1.3438018560409546, | |
| "learning_rate": 2.5771428571428574e-06, | |
| "loss": 0.0141, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.537708927843457, | |
| "grad_norm": 1.3562772274017334, | |
| "learning_rate": 2.541428571428572e-06, | |
| "loss": 0.0134, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 2.54790052996331, | |
| "grad_norm": 0.681122362613678, | |
| "learning_rate": 2.5057142857142856e-06, | |
| "loss": 0.0123, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.5580921320831633, | |
| "grad_norm": 0.8060537576675415, | |
| "learning_rate": 2.47e-06, | |
| "loss": 0.0143, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 2.5682837342030167, | |
| "grad_norm": 0.9794915318489075, | |
| "learning_rate": 2.4342857142857146e-06, | |
| "loss": 0.0128, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.57847533632287, | |
| "grad_norm": 0.7749600410461426, | |
| "learning_rate": 2.3985714285714287e-06, | |
| "loss": 0.0132, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 2.5886669384427234, | |
| "grad_norm": 0.808312177658081, | |
| "learning_rate": 2.362857142857143e-06, | |
| "loss": 0.0166, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.5988585405625764, | |
| "grad_norm": 1.0200546979904175, | |
| "learning_rate": 2.3271428571428572e-06, | |
| "loss": 0.0138, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 2.6090501426824297, | |
| "grad_norm": 1.4276586771011353, | |
| "learning_rate": 2.2914285714285718e-06, | |
| "loss": 0.0119, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.6192417448022827, | |
| "grad_norm": 1.1479567289352417, | |
| "learning_rate": 2.255714285714286e-06, | |
| "loss": 0.0127, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 2.629433346922136, | |
| "grad_norm": 0.9421886205673218, | |
| "learning_rate": 2.2200000000000003e-06, | |
| "loss": 0.0132, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.6396249490419894, | |
| "grad_norm": 0.573397696018219, | |
| "learning_rate": 2.1842857142857144e-06, | |
| "loss": 0.0131, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 2.649816551161843, | |
| "grad_norm": 1.2187764644622803, | |
| "learning_rate": 2.148571428571429e-06, | |
| "loss": 0.0125, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.660008153281696, | |
| "grad_norm": 0.7359505891799927, | |
| "learning_rate": 2.112857142857143e-06, | |
| "loss": 0.0141, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 2.670199755401549, | |
| "grad_norm": 1.8559486865997314, | |
| "learning_rate": 2.077142857142857e-06, | |
| "loss": 0.0142, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.6803913575214025, | |
| "grad_norm": 1.0747308731079102, | |
| "learning_rate": 2.0414285714285716e-06, | |
| "loss": 0.0105, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 2.6905829596412554, | |
| "grad_norm": 0.8565563559532166, | |
| "learning_rate": 2.0057142857142857e-06, | |
| "loss": 0.0144, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.7007745617611087, | |
| "grad_norm": 1.1864107847213745, | |
| "learning_rate": 1.97e-06, | |
| "loss": 0.013, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 2.710966163880962, | |
| "grad_norm": 0.6643468141555786, | |
| "learning_rate": 1.9342857142857143e-06, | |
| "loss": 0.0149, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.7211577660008155, | |
| "grad_norm": 0.6889024376869202, | |
| "learning_rate": 1.8985714285714288e-06, | |
| "loss": 0.0115, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 2.7313493681206684, | |
| "grad_norm": 1.5234169960021973, | |
| "learning_rate": 1.8628571428571429e-06, | |
| "loss": 0.0139, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.741540970240522, | |
| "grad_norm": 1.095461368560791, | |
| "learning_rate": 1.8271428571428574e-06, | |
| "loss": 0.0132, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 2.751732572360375, | |
| "grad_norm": 1.7404385805130005, | |
| "learning_rate": 1.7914285714285715e-06, | |
| "loss": 0.0135, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.761924174480228, | |
| "grad_norm": 0.8274975419044495, | |
| "learning_rate": 1.755714285714286e-06, | |
| "loss": 0.013, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 2.7721157766000815, | |
| "grad_norm": 0.8858354091644287, | |
| "learning_rate": 1.72e-06, | |
| "loss": 0.0118, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.782307378719935, | |
| "grad_norm": 0.5157521963119507, | |
| "learning_rate": 1.6842857142857143e-06, | |
| "loss": 0.0107, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 2.792498980839788, | |
| "grad_norm": 1.2508779764175415, | |
| "learning_rate": 1.6485714285714289e-06, | |
| "loss": 0.015, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.802690582959641, | |
| "grad_norm": 1.2929726839065552, | |
| "learning_rate": 1.612857142857143e-06, | |
| "loss": 0.0125, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 2.8128821850794945, | |
| "grad_norm": 0.8893064856529236, | |
| "learning_rate": 1.5771428571428574e-06, | |
| "loss": 0.0122, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.823073787199348, | |
| "grad_norm": 1.2606889009475708, | |
| "learning_rate": 1.5414285714285715e-06, | |
| "loss": 0.0137, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 2.833265389319201, | |
| "grad_norm": 1.410904884338379, | |
| "learning_rate": 1.5057142857142858e-06, | |
| "loss": 0.0121, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.843456991439054, | |
| "grad_norm": 0.792271077632904, | |
| "learning_rate": 1.4700000000000001e-06, | |
| "loss": 0.0135, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 2.8536485935589075, | |
| "grad_norm": 1.0926941633224487, | |
| "learning_rate": 1.4342857142857144e-06, | |
| "loss": 0.0117, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.8536485935589075, | |
| "eval_loss": 0.022013096138834953, | |
| "eval_runtime": 2391.195, | |
| "eval_samples_per_second": 4.264, | |
| "eval_steps_per_second": 0.133, | |
| "eval_wer": 2.5110128090281947, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.863840195678761, | |
| "grad_norm": 1.1174569129943848, | |
| "learning_rate": 1.3985714285714285e-06, | |
| "loss": 0.0123, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 2.874031797798614, | |
| "grad_norm": 1.3422982692718506, | |
| "learning_rate": 1.362857142857143e-06, | |
| "loss": 0.0155, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.884223399918467, | |
| "grad_norm": 1.1771271228790283, | |
| "learning_rate": 1.327142857142857e-06, | |
| "loss": 0.014, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 2.8944150020383206, | |
| "grad_norm": 0.8138777017593384, | |
| "learning_rate": 1.2914285714285716e-06, | |
| "loss": 0.0159, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.9046066041581735, | |
| "grad_norm": 0.964419960975647, | |
| "learning_rate": 1.2557142857142859e-06, | |
| "loss": 0.0119, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 2.914798206278027, | |
| "grad_norm": 1.0425853729248047, | |
| "learning_rate": 1.2200000000000002e-06, | |
| "loss": 0.0144, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.9249898083978803, | |
| "grad_norm": 1.2048903703689575, | |
| "learning_rate": 1.1842857142857143e-06, | |
| "loss": 0.0112, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 2.9351814105177336, | |
| "grad_norm": 0.9228710532188416, | |
| "learning_rate": 1.1485714285714286e-06, | |
| "loss": 0.0121, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.9453730126375866, | |
| "grad_norm": 0.35819801688194275, | |
| "learning_rate": 1.1128571428571429e-06, | |
| "loss": 0.0103, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 2.95556461475744, | |
| "grad_norm": 0.8136561512947083, | |
| "learning_rate": 1.0771428571428574e-06, | |
| "loss": 0.0129, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.965756216877293, | |
| "grad_norm": 1.0411508083343506, | |
| "learning_rate": 1.0414285714285717e-06, | |
| "loss": 0.0104, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 2.975947818997146, | |
| "grad_norm": 0.7591568231582642, | |
| "learning_rate": 1.0057142857142857e-06, | |
| "loss": 0.0104, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.9861394211169996, | |
| "grad_norm": 0.9218117594718933, | |
| "learning_rate": 9.7e-07, | |
| "loss": 0.0123, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 2.996331023236853, | |
| "grad_norm": 1.3467174768447876, | |
| "learning_rate": 9.342857142857144e-07, | |
| "loss": 0.01, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.006522625356706, | |
| "grad_norm": 0.14992065727710724, | |
| "learning_rate": 8.985714285714286e-07, | |
| "loss": 0.0068, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 3.0167142274765593, | |
| "grad_norm": 0.44808146357536316, | |
| "learning_rate": 8.628571428571429e-07, | |
| "loss": 0.0044, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.0269058295964126, | |
| "grad_norm": 0.4159485101699829, | |
| "learning_rate": 8.271428571428572e-07, | |
| "loss": 0.0045, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 3.037097431716266, | |
| "grad_norm": 0.5730934143066406, | |
| "learning_rate": 7.914285714285715e-07, | |
| "loss": 0.0041, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 3.047289033836119, | |
| "grad_norm": 0.2851293683052063, | |
| "learning_rate": 7.557142857142857e-07, | |
| "loss": 0.0048, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 3.0574806359559723, | |
| "grad_norm": 0.3623310625553131, | |
| "learning_rate": 7.2e-07, | |
| "loss": 0.0054, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.0676722380758257, | |
| "grad_norm": 0.1989881545305252, | |
| "learning_rate": 6.842857142857143e-07, | |
| "loss": 0.004, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 3.0778638401956786, | |
| "grad_norm": 0.24523906409740448, | |
| "learning_rate": 6.485714285714287e-07, | |
| "loss": 0.0052, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 3.088055442315532, | |
| "grad_norm": 0.09773947298526764, | |
| "learning_rate": 6.128571428571429e-07, | |
| "loss": 0.0037, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 3.0982470444353853, | |
| "grad_norm": 0.8551767468452454, | |
| "learning_rate": 5.771428571428572e-07, | |
| "loss": 0.0041, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.1084386465552383, | |
| "grad_norm": 0.32709449529647827, | |
| "learning_rate": 5.414285714285715e-07, | |
| "loss": 0.0046, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 3.1186302486750916, | |
| "grad_norm": 0.22505348920822144, | |
| "learning_rate": 5.057142857142858e-07, | |
| "loss": 0.0039, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 3.128821850794945, | |
| "grad_norm": 0.6920948028564453, | |
| "learning_rate": 4.7000000000000005e-07, | |
| "loss": 0.005, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 3.1390134529147984, | |
| "grad_norm": 0.37104272842407227, | |
| "learning_rate": 4.342857142857143e-07, | |
| "loss": 0.0041, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.1492050550346513, | |
| "grad_norm": 0.5752152800559998, | |
| "learning_rate": 3.985714285714286e-07, | |
| "loss": 0.0046, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 3.1593966571545047, | |
| "grad_norm": 0.40613701939582825, | |
| "learning_rate": 3.6285714285714283e-07, | |
| "loss": 0.0041, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 3.169588259274358, | |
| "grad_norm": 0.4888840317726135, | |
| "learning_rate": 3.271428571428572e-07, | |
| "loss": 0.0047, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 3.179779861394211, | |
| "grad_norm": 0.22719787061214447, | |
| "learning_rate": 2.914285714285715e-07, | |
| "loss": 0.0039, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 3.1899714635140644, | |
| "grad_norm": 0.3767295479774475, | |
| "learning_rate": 2.557142857142857e-07, | |
| "loss": 0.0047, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 3.2001630656339177, | |
| "grad_norm": 0.6686317324638367, | |
| "learning_rate": 2.2e-07, | |
| "loss": 0.0041, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 3.210354667753771, | |
| "grad_norm": 0.40455150604248047, | |
| "learning_rate": 1.842857142857143e-07, | |
| "loss": 0.0058, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 3.220546269873624, | |
| "grad_norm": 0.2686769962310791, | |
| "learning_rate": 1.4857142857142857e-07, | |
| "loss": 0.0046, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 3.2307378719934774, | |
| "grad_norm": 0.17891818284988403, | |
| "learning_rate": 1.1285714285714287e-07, | |
| "loss": 0.0038, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 3.2409294741133308, | |
| "grad_norm": 0.32351842522621155, | |
| "learning_rate": 7.714285714285715e-08, | |
| "loss": 0.0036, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 3.2511210762331837, | |
| "grad_norm": 0.2572202682495117, | |
| "learning_rate": 4.1428571428571426e-08, | |
| "loss": 0.0036, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 3.261312678353037, | |
| "grad_norm": 0.1762164831161499, | |
| "learning_rate": 5.714285714285715e-09, | |
| "loss": 0.0044, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.261312678353037, | |
| "eval_loss": 0.02037939429283142, | |
| "eval_runtime": 2392.6668, | |
| "eval_samples_per_second": 4.261, | |
| "eval_steps_per_second": 0.133, | |
| "eval_wer": 2.2782892974889872, | |
| "step": 8000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 8000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.6124729999261696e+20, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |