| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 8.458447874814972, | |
| "eval_steps": 1000, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010573059843518714, | |
| "grad_norm": 58.068878173828125, | |
| "learning_rate": 4.4e-07, | |
| "loss": 3.1142, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02114611968703743, | |
| "grad_norm": 37.93722152709961, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 2.5594, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03171917953055614, | |
| "grad_norm": 22.363622665405273, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.6703, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04229223937407486, | |
| "grad_norm": 15.931524276733398, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.0597, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05286529921759357, | |
| "grad_norm": 13.918318748474121, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.765, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06343835906111228, | |
| "grad_norm": 14.767135620117188, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.6955, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.074011418904631, | |
| "grad_norm": 11.596781730651855, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.6118, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.08458447874814971, | |
| "grad_norm": 10.992063522338867, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.6096, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09515753859166842, | |
| "grad_norm": 12.467724800109863, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.5579, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.10573059843518715, | |
| "grad_norm": 11.876873016357422, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.5435, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.11630365827870585, | |
| "grad_norm": 12.889082908630371, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.5693, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.12687671812222456, | |
| "grad_norm": 10.841497421264648, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.5494, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.13744977796574329, | |
| "grad_norm": 10.560270309448242, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.5065, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.148022837809262, | |
| "grad_norm": 13.699323654174805, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.4911, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1585958976527807, | |
| "grad_norm": 9.73985481262207, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.5293, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.16916895749629943, | |
| "grad_norm": 11.41083812713623, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.4365, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.17974201733981815, | |
| "grad_norm": 10.309052467346191, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.5207, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.19031507718333684, | |
| "grad_norm": 10.460412979125977, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.4537, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.20088813702685557, | |
| "grad_norm": 10.022929191589355, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.4395, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.2114611968703743, | |
| "grad_norm": 10.307358741760254, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.4439, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.222034256713893, | |
| "grad_norm": 8.121826171875, | |
| "learning_rate": 9.98871794871795e-06, | |
| "loss": 0.4624, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.2326073165574117, | |
| "grad_norm": 9.346823692321777, | |
| "learning_rate": 9.975897435897436e-06, | |
| "loss": 0.4533, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.24318037640093043, | |
| "grad_norm": 15.096284866333008, | |
| "learning_rate": 9.963076923076925e-06, | |
| "loss": 0.4372, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.2537534362444491, | |
| "grad_norm": 9.635919570922852, | |
| "learning_rate": 9.950256410256412e-06, | |
| "loss": 0.4149, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2643264960879679, | |
| "grad_norm": 12.7789945602417, | |
| "learning_rate": 9.937435897435898e-06, | |
| "loss": 0.4219, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.27489955593148657, | |
| "grad_norm": 10.612743377685547, | |
| "learning_rate": 9.924615384615385e-06, | |
| "loss": 0.4103, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.28547261577500527, | |
| "grad_norm": 9.644041061401367, | |
| "learning_rate": 9.911794871794874e-06, | |
| "loss": 0.3747, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.296045675618524, | |
| "grad_norm": 12.278302192687988, | |
| "learning_rate": 9.89897435897436e-06, | |
| "loss": 0.3818, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.3066187354620427, | |
| "grad_norm": 11.676104545593262, | |
| "learning_rate": 9.886153846153846e-06, | |
| "loss": 0.3964, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.3171917953055614, | |
| "grad_norm": 9.235854148864746, | |
| "learning_rate": 9.873333333333334e-06, | |
| "loss": 0.3954, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.32776485514908016, | |
| "grad_norm": 10.16644287109375, | |
| "learning_rate": 9.860512820512821e-06, | |
| "loss": 0.4021, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.33833791499259885, | |
| "grad_norm": 11.959391593933105, | |
| "learning_rate": 9.847692307692308e-06, | |
| "loss": 0.4094, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.34891097483611755, | |
| "grad_norm": 9.894736289978027, | |
| "learning_rate": 9.834871794871795e-06, | |
| "loss": 0.386, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.3594840346796363, | |
| "grad_norm": 10.717971801757812, | |
| "learning_rate": 9.822051282051283e-06, | |
| "loss": 0.3786, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.370057094523155, | |
| "grad_norm": 9.43459701538086, | |
| "learning_rate": 9.80923076923077e-06, | |
| "loss": 0.3558, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.3806301543666737, | |
| "grad_norm": 8.999226570129395, | |
| "learning_rate": 9.796410256410257e-06, | |
| "loss": 0.4059, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.39120321421019244, | |
| "grad_norm": 9.803775787353516, | |
| "learning_rate": 9.783589743589744e-06, | |
| "loss": 0.3665, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.40177627405371114, | |
| "grad_norm": 6.630070209503174, | |
| "learning_rate": 9.770769230769232e-06, | |
| "loss": 0.3913, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.4123493338972299, | |
| "grad_norm": 8.892091751098633, | |
| "learning_rate": 9.757948717948719e-06, | |
| "loss": 0.4345, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.4229223937407486, | |
| "grad_norm": 11.313458442687988, | |
| "learning_rate": 9.745128205128206e-06, | |
| "loss": 0.3523, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4229223937407486, | |
| "eval_loss": 0.3129430413246155, | |
| "eval_runtime": 466.842, | |
| "eval_samples_per_second": 7.799, | |
| "eval_steps_per_second": 0.977, | |
| "eval_wer": 0.23648429161446008, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.4334954535842673, | |
| "grad_norm": 10.663793563842773, | |
| "learning_rate": 9.732307692307693e-06, | |
| "loss": 0.3711, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.444068513427786, | |
| "grad_norm": 8.810164451599121, | |
| "learning_rate": 9.71948717948718e-06, | |
| "loss": 0.355, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.4546415732713047, | |
| "grad_norm": 17.50876235961914, | |
| "learning_rate": 9.706666666666668e-06, | |
| "loss": 0.3652, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.4652146331148234, | |
| "grad_norm": 6.720116138458252, | |
| "learning_rate": 9.693846153846155e-06, | |
| "loss": 0.3415, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.47578769295834217, | |
| "grad_norm": 6.864238739013672, | |
| "learning_rate": 9.681025641025642e-06, | |
| "loss": 0.3203, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.48636075280186086, | |
| "grad_norm": 10.059639930725098, | |
| "learning_rate": 9.668205128205129e-06, | |
| "loss": 0.4017, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.49693381264537956, | |
| "grad_norm": 8.964330673217773, | |
| "learning_rate": 9.655384615384617e-06, | |
| "loss": 0.3089, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.5075068724888983, | |
| "grad_norm": 11.160856246948242, | |
| "learning_rate": 9.642564102564104e-06, | |
| "loss": 0.357, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.518079932332417, | |
| "grad_norm": 8.594799041748047, | |
| "learning_rate": 9.62974358974359e-06, | |
| "loss": 0.3391, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.5286529921759358, | |
| "grad_norm": 7.494054794311523, | |
| "learning_rate": 9.616923076923077e-06, | |
| "loss": 0.3315, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.5392260520194544, | |
| "grad_norm": 7.793934345245361, | |
| "learning_rate": 9.604102564102566e-06, | |
| "loss": 0.3226, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.5497991118629731, | |
| "grad_norm": 7.225709438323975, | |
| "learning_rate": 9.591282051282053e-06, | |
| "loss": 0.3317, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.5603721717064919, | |
| "grad_norm": 6.296034812927246, | |
| "learning_rate": 9.578461538461538e-06, | |
| "loss": 0.3302, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.5709452315500105, | |
| "grad_norm": 11.198543548583984, | |
| "learning_rate": 9.565641025641026e-06, | |
| "loss": 0.3038, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.5815182913935293, | |
| "grad_norm": 6.4023542404174805, | |
| "learning_rate": 9.552820512820513e-06, | |
| "loss": 0.3511, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.592091351237048, | |
| "grad_norm": 8.326896667480469, | |
| "learning_rate": 9.54e-06, | |
| "loss": 0.3257, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.6026644110805667, | |
| "grad_norm": 7.079158306121826, | |
| "learning_rate": 9.527179487179487e-06, | |
| "loss": 0.335, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.6132374709240854, | |
| "grad_norm": 8.422740936279297, | |
| "learning_rate": 9.514358974358975e-06, | |
| "loss": 0.3212, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.6238105307676042, | |
| "grad_norm": 8.642356872558594, | |
| "learning_rate": 9.501538461538462e-06, | |
| "loss": 0.314, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.6343835906111228, | |
| "grad_norm": 8.129634857177734, | |
| "learning_rate": 9.488717948717949e-06, | |
| "loss": 0.2889, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.6449566504546416, | |
| "grad_norm": 7.115053653717041, | |
| "learning_rate": 9.475897435897436e-06, | |
| "loss": 0.3073, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.6555297102981603, | |
| "grad_norm": 8.397727966308594, | |
| "learning_rate": 9.463076923076924e-06, | |
| "loss": 0.3212, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.666102770141679, | |
| "grad_norm": 9.354975700378418, | |
| "learning_rate": 9.450256410256411e-06, | |
| "loss": 0.3285, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.6766758299851977, | |
| "grad_norm": 7.693446159362793, | |
| "learning_rate": 9.437435897435898e-06, | |
| "loss": 0.3382, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.6872488898287165, | |
| "grad_norm": 10.583995819091797, | |
| "learning_rate": 9.424615384615385e-06, | |
| "loss": 0.3198, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.6978219496722351, | |
| "grad_norm": 9.70487117767334, | |
| "learning_rate": 9.411794871794872e-06, | |
| "loss": 0.2968, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.7083950095157538, | |
| "grad_norm": 8.256582260131836, | |
| "learning_rate": 9.39897435897436e-06, | |
| "loss": 0.2994, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.7189680693592726, | |
| "grad_norm": 8.840137481689453, | |
| "learning_rate": 9.386153846153847e-06, | |
| "loss": 0.347, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.7295411292027912, | |
| "grad_norm": 6.581663608551025, | |
| "learning_rate": 9.373333333333334e-06, | |
| "loss": 0.3436, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.74011418904631, | |
| "grad_norm": 10.300860404968262, | |
| "learning_rate": 9.36051282051282e-06, | |
| "loss": 0.2989, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.7506872488898287, | |
| "grad_norm": 8.777045249938965, | |
| "learning_rate": 9.34769230769231e-06, | |
| "loss": 0.3212, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.7612603087333474, | |
| "grad_norm": 8.94032096862793, | |
| "learning_rate": 9.334871794871796e-06, | |
| "loss": 0.2951, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.7718333685768661, | |
| "grad_norm": 6.139760971069336, | |
| "learning_rate": 9.322051282051283e-06, | |
| "loss": 0.2962, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.7824064284203849, | |
| "grad_norm": 6.843733310699463, | |
| "learning_rate": 9.30923076923077e-06, | |
| "loss": 0.3324, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.7929794882639035, | |
| "grad_norm": 8.275738716125488, | |
| "learning_rate": 9.296410256410258e-06, | |
| "loss": 0.289, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.8035525481074223, | |
| "grad_norm": 7.651088237762451, | |
| "learning_rate": 9.283589743589745e-06, | |
| "loss": 0.3563, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.814125607950941, | |
| "grad_norm": 9.584896087646484, | |
| "learning_rate": 9.270769230769232e-06, | |
| "loss": 0.3009, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.8246986677944598, | |
| "grad_norm": 8.920114517211914, | |
| "learning_rate": 9.257948717948719e-06, | |
| "loss": 0.3229, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.8352717276379784, | |
| "grad_norm": 8.181923866271973, | |
| "learning_rate": 9.245128205128206e-06, | |
| "loss": 0.2707, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.8458447874814972, | |
| "grad_norm": 6.475239276885986, | |
| "learning_rate": 9.232307692307692e-06, | |
| "loss": 0.3002, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8458447874814972, | |
| "eval_loss": 0.2391187697649002, | |
| "eval_runtime": 464.6331, | |
| "eval_samples_per_second": 7.836, | |
| "eval_steps_per_second": 0.981, | |
| "eval_wer": 0.19638962364842916, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8564178473250159, | |
| "grad_norm": 6.2346930503845215, | |
| "learning_rate": 9.21948717948718e-06, | |
| "loss": 0.3355, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.8669909071685346, | |
| "grad_norm": 8.41283130645752, | |
| "learning_rate": 9.207179487179488e-06, | |
| "loss": 0.2924, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.8775639670120533, | |
| "grad_norm": 10.027973175048828, | |
| "learning_rate": 9.194358974358975e-06, | |
| "loss": 0.2711, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.888137026855572, | |
| "grad_norm": 6.886898040771484, | |
| "learning_rate": 9.181538461538464e-06, | |
| "loss": 0.2593, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.8987100866990907, | |
| "grad_norm": 6.60000467300415, | |
| "learning_rate": 9.168717948717949e-06, | |
| "loss": 0.2818, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.9092831465426094, | |
| "grad_norm": 8.06698989868164, | |
| "learning_rate": 9.155897435897436e-06, | |
| "loss": 0.2812, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.9198562063861282, | |
| "grad_norm": 11.074329376220703, | |
| "learning_rate": 9.143076923076924e-06, | |
| "loss": 0.2717, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.9304292662296468, | |
| "grad_norm": 7.394618511199951, | |
| "learning_rate": 9.130256410256411e-06, | |
| "loss": 0.2752, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.9410023260731656, | |
| "grad_norm": 9.914189338684082, | |
| "learning_rate": 9.117435897435898e-06, | |
| "loss": 0.2842, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.9515753859166843, | |
| "grad_norm": 13.050373077392578, | |
| "learning_rate": 9.104615384615385e-06, | |
| "loss": 0.2681, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.962148445760203, | |
| "grad_norm": 9.03541088104248, | |
| "learning_rate": 9.091794871794873e-06, | |
| "loss": 0.2969, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.9727215056037217, | |
| "grad_norm": 9.255524635314941, | |
| "learning_rate": 9.07897435897436e-06, | |
| "loss": 0.2558, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.9832945654472405, | |
| "grad_norm": 9.215723037719727, | |
| "learning_rate": 9.066153846153847e-06, | |
| "loss": 0.2648, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.9938676252907591, | |
| "grad_norm": 10.28739070892334, | |
| "learning_rate": 9.053333333333334e-06, | |
| "loss": 0.2809, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.0044406851342778, | |
| "grad_norm": 5.855769157409668, | |
| "learning_rate": 9.04051282051282e-06, | |
| "loss": 0.2384, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.0150137449777965, | |
| "grad_norm": 5.34947395324707, | |
| "learning_rate": 9.027692307692309e-06, | |
| "loss": 0.2049, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.0255868048213153, | |
| "grad_norm": 6.5480170249938965, | |
| "learning_rate": 9.014871794871796e-06, | |
| "loss": 0.1907, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.036159864664834, | |
| "grad_norm": 8.17578125, | |
| "learning_rate": 9.002051282051283e-06, | |
| "loss": 0.1948, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.0467329245083528, | |
| "grad_norm": 6.4552083015441895, | |
| "learning_rate": 8.98923076923077e-06, | |
| "loss": 0.1703, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.0573059843518715, | |
| "grad_norm": 6.099579811096191, | |
| "learning_rate": 8.976410256410258e-06, | |
| "loss": 0.1765, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.06787904419539, | |
| "grad_norm": 7.8965301513671875, | |
| "learning_rate": 8.963589743589745e-06, | |
| "loss": 0.1849, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.0784521040389088, | |
| "grad_norm": 6.395140171051025, | |
| "learning_rate": 8.950769230769232e-06, | |
| "loss": 0.2055, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.0890251638824275, | |
| "grad_norm": 5.077881813049316, | |
| "learning_rate": 8.937948717948718e-06, | |
| "loss": 0.1857, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.0995982237259463, | |
| "grad_norm": 6.063033103942871, | |
| "learning_rate": 8.925128205128207e-06, | |
| "loss": 0.1629, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.110171283569465, | |
| "grad_norm": 5.514612197875977, | |
| "learning_rate": 8.912307692307694e-06, | |
| "loss": 0.1735, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.1207443434129838, | |
| "grad_norm": 6.513278961181641, | |
| "learning_rate": 8.89948717948718e-06, | |
| "loss": 0.174, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.1313174032565025, | |
| "grad_norm": 5.930003643035889, | |
| "learning_rate": 8.886666666666667e-06, | |
| "loss": 0.183, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.141890463100021, | |
| "grad_norm": 6.062052249908447, | |
| "learning_rate": 8.873846153846156e-06, | |
| "loss": 0.2031, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.1524635229435398, | |
| "grad_norm": 6.619091987609863, | |
| "learning_rate": 8.861025641025641e-06, | |
| "loss": 0.1669, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.1630365827870586, | |
| "grad_norm": 6.383360385894775, | |
| "learning_rate": 8.848205128205128e-06, | |
| "loss": 0.1698, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.1736096426305773, | |
| "grad_norm": 5.248838901519775, | |
| "learning_rate": 8.835384615384616e-06, | |
| "loss": 0.1783, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.184182702474096, | |
| "grad_norm": 4.1173319816589355, | |
| "learning_rate": 8.822564102564103e-06, | |
| "loss": 0.1689, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.1947557623176146, | |
| "grad_norm": 4.528750419616699, | |
| "learning_rate": 8.80974358974359e-06, | |
| "loss": 0.1714, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.2053288221611334, | |
| "grad_norm": 7.49066686630249, | |
| "learning_rate": 8.796923076923077e-06, | |
| "loss": 0.1567, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.215901882004652, | |
| "grad_norm": 6.05342960357666, | |
| "learning_rate": 8.784102564102565e-06, | |
| "loss": 0.1675, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.2264749418481709, | |
| "grad_norm": 6.280456066131592, | |
| "learning_rate": 8.771282051282052e-06, | |
| "loss": 0.1885, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.2370480016916896, | |
| "grad_norm": 6.7942657470703125, | |
| "learning_rate": 8.758461538461539e-06, | |
| "loss": 0.1725, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.2476210615352084, | |
| "grad_norm": 5.41685152053833, | |
| "learning_rate": 8.745641025641026e-06, | |
| "loss": 0.178, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.258194121378727, | |
| "grad_norm": 7.254810333251953, | |
| "learning_rate": 8.732820512820513e-06, | |
| "loss": 0.1788, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.2687671812222456, | |
| "grad_norm": 5.75822639465332, | |
| "learning_rate": 8.720000000000001e-06, | |
| "loss": 0.1718, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2687671812222456, | |
| "eval_loss": 0.20488913357257843, | |
| "eval_runtime": 470.0703, | |
| "eval_samples_per_second": 7.746, | |
| "eval_steps_per_second": 0.97, | |
| "eval_wer": 0.16594737574829457, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.2793402410657644, | |
| "grad_norm": 4.381278038024902, | |
| "learning_rate": 8.707179487179488e-06, | |
| "loss": 0.1656, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.2899133009092831, | |
| "grad_norm": 6.3774919509887695, | |
| "learning_rate": 8.694358974358975e-06, | |
| "loss": 0.1987, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.3004863607528019, | |
| "grad_norm": 5.695129871368408, | |
| "learning_rate": 8.681538461538462e-06, | |
| "loss": 0.1602, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.3110594205963206, | |
| "grad_norm": 5.8820013999938965, | |
| "learning_rate": 8.66871794871795e-06, | |
| "loss": 0.1788, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.3216324804398392, | |
| "grad_norm": 6.164638519287109, | |
| "learning_rate": 8.655897435897437e-06, | |
| "loss": 0.2075, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.3322055402833581, | |
| "grad_norm": 6.681603908538818, | |
| "learning_rate": 8.643076923076924e-06, | |
| "loss": 0.1867, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.3427786001268767, | |
| "grad_norm": 4.936852931976318, | |
| "learning_rate": 8.63025641025641e-06, | |
| "loss": 0.1917, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.3533516599703954, | |
| "grad_norm": 5.827350616455078, | |
| "learning_rate": 8.6174358974359e-06, | |
| "loss": 0.1898, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.3639247198139142, | |
| "grad_norm": 6.6198272705078125, | |
| "learning_rate": 8.604615384615386e-06, | |
| "loss": 0.1659, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.374497779657433, | |
| "grad_norm": 8.098386764526367, | |
| "learning_rate": 8.591794871794873e-06, | |
| "loss": 0.1793, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.3850708395009517, | |
| "grad_norm": 7.836790561676025, | |
| "learning_rate": 8.57897435897436e-06, | |
| "loss": 0.1551, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.3956438993444702, | |
| "grad_norm": 4.521047115325928, | |
| "learning_rate": 8.566153846153848e-06, | |
| "loss": 0.156, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.406216959187989, | |
| "grad_norm": 6.938994407653809, | |
| "learning_rate": 8.553333333333333e-06, | |
| "loss": 0.1722, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.4167900190315077, | |
| "grad_norm": 7.659555435180664, | |
| "learning_rate": 8.54051282051282e-06, | |
| "loss": 0.1781, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.4273630788750264, | |
| "grad_norm": 5.197014808654785, | |
| "learning_rate": 8.527692307692309e-06, | |
| "loss": 0.1606, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.4379361387185452, | |
| "grad_norm": 4.832046031951904, | |
| "learning_rate": 8.514871794871795e-06, | |
| "loss": 0.1579, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.4485091985620637, | |
| "grad_norm": 7.802398681640625, | |
| "learning_rate": 8.502051282051282e-06, | |
| "loss": 0.1922, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.4590822584055827, | |
| "grad_norm": 6.803133487701416, | |
| "learning_rate": 8.489230769230769e-06, | |
| "loss": 0.2039, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.4696553182491012, | |
| "grad_norm": 6.156517505645752, | |
| "learning_rate": 8.476410256410258e-06, | |
| "loss": 0.1723, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.48022837809262, | |
| "grad_norm": 6.290517330169678, | |
| "learning_rate": 8.463589743589744e-06, | |
| "loss": 0.1739, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.4908014379361387, | |
| "grad_norm": 7.008253574371338, | |
| "learning_rate": 8.450769230769231e-06, | |
| "loss": 0.1465, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.5013744977796575, | |
| "grad_norm": 5.356746673583984, | |
| "learning_rate": 8.437948717948718e-06, | |
| "loss": 0.185, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.5119475576231762, | |
| "grad_norm": 7.195596694946289, | |
| "learning_rate": 8.425128205128205e-06, | |
| "loss": 0.1968, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.5225206174666948, | |
| "grad_norm": 7.071013927459717, | |
| "learning_rate": 8.412307692307693e-06, | |
| "loss": 0.1722, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.5330936773102137, | |
| "grad_norm": 6.704313278198242, | |
| "learning_rate": 8.39948717948718e-06, | |
| "loss": 0.1763, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.5436667371537323, | |
| "grad_norm": 3.831084966659546, | |
| "learning_rate": 8.386666666666667e-06, | |
| "loss": 0.1795, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.554239796997251, | |
| "grad_norm": 5.931239604949951, | |
| "learning_rate": 8.373846153846154e-06, | |
| "loss": 0.1717, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.5648128568407698, | |
| "grad_norm": 8.750226974487305, | |
| "learning_rate": 8.361025641025642e-06, | |
| "loss": 0.1679, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.5753859166842883, | |
| "grad_norm": 5.434117317199707, | |
| "learning_rate": 8.34820512820513e-06, | |
| "loss": 0.1343, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.5859589765278073, | |
| "grad_norm": 6.517702102661133, | |
| "learning_rate": 8.335384615384616e-06, | |
| "loss": 0.1641, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.5965320363713258, | |
| "grad_norm": 7.22881555557251, | |
| "learning_rate": 8.322564102564103e-06, | |
| "loss": 0.152, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.6071050962148445, | |
| "grad_norm": 9.106505393981934, | |
| "learning_rate": 8.309743589743591e-06, | |
| "loss": 0.168, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.6176781560583633, | |
| "grad_norm": 6.742056369781494, | |
| "learning_rate": 8.296923076923078e-06, | |
| "loss": 0.1624, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.628251215901882, | |
| "grad_norm": 5.521613597869873, | |
| "learning_rate": 8.284102564102565e-06, | |
| "loss": 0.1476, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.6388242757454008, | |
| "grad_norm": 5.299131870269775, | |
| "learning_rate": 8.271282051282052e-06, | |
| "loss": 0.1576, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.6493973355889193, | |
| "grad_norm": 3.890312671661377, | |
| "learning_rate": 8.25846153846154e-06, | |
| "loss": 0.162, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.6599703954324383, | |
| "grad_norm": 5.404117584228516, | |
| "learning_rate": 8.245641025641027e-06, | |
| "loss": 0.1622, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.6705434552759568, | |
| "grad_norm": 8.395030975341797, | |
| "learning_rate": 8.232820512820512e-06, | |
| "loss": 0.1473, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.6811165151194756, | |
| "grad_norm": 7.276181221008301, | |
| "learning_rate": 8.220000000000001e-06, | |
| "loss": 0.1538, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.6916895749629943, | |
| "grad_norm": 8.637175559997559, | |
| "learning_rate": 8.207179487179488e-06, | |
| "loss": 0.1537, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.6916895749629943, | |
| "eval_loss": 0.1817464828491211, | |
| "eval_runtime": 469.7808, | |
| "eval_samples_per_second": 7.75, | |
| "eval_steps_per_second": 0.971, | |
| "eval_wer": 0.15156155738085295, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.7022626348065129, | |
| "grad_norm": 2.9336845874786377, | |
| "learning_rate": 8.194358974358975e-06, | |
| "loss": 0.1789, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.7128356946500318, | |
| "grad_norm": 4.7298688888549805, | |
| "learning_rate": 8.181538461538461e-06, | |
| "loss": 0.1631, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.7234087544935504, | |
| "grad_norm": 6.827828407287598, | |
| "learning_rate": 8.16871794871795e-06, | |
| "loss": 0.1599, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.733981814337069, | |
| "grad_norm": 5.460879325866699, | |
| "learning_rate": 8.155897435897437e-06, | |
| "loss": 0.1836, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.7445548741805879, | |
| "grad_norm": 6.356947898864746, | |
| "learning_rate": 8.143076923076924e-06, | |
| "loss": 0.1533, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.7551279340241066, | |
| "grad_norm": 7.573319435119629, | |
| "learning_rate": 8.13025641025641e-06, | |
| "loss": 0.1437, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.7657009938676254, | |
| "grad_norm": 4.901601791381836, | |
| "learning_rate": 8.117435897435897e-06, | |
| "loss": 0.1371, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.7762740537111439, | |
| "grad_norm": 6.051869869232178, | |
| "learning_rate": 8.104615384615386e-06, | |
| "loss": 0.1883, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.7868471135546629, | |
| "grad_norm": 7.222878932952881, | |
| "learning_rate": 8.091794871794873e-06, | |
| "loss": 0.1614, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.7974201733981814, | |
| "grad_norm": 6.554632663726807, | |
| "learning_rate": 8.07897435897436e-06, | |
| "loss": 0.1629, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.8079932332417001, | |
| "grad_norm": 6.802252292633057, | |
| "learning_rate": 8.066153846153846e-06, | |
| "loss": 0.1526, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.8185662930852189, | |
| "grad_norm": 6.444076061248779, | |
| "learning_rate": 8.053333333333335e-06, | |
| "loss": 0.1549, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.8291393529287374, | |
| "grad_norm": 6.49000358581543, | |
| "learning_rate": 8.040512820512822e-06, | |
| "loss": 0.1469, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.8397124127722564, | |
| "grad_norm": 5.989566802978516, | |
| "learning_rate": 8.027692307692308e-06, | |
| "loss": 0.1587, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.850285472615775, | |
| "grad_norm": 4.621600151062012, | |
| "learning_rate": 8.014871794871795e-06, | |
| "loss": 0.1394, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.8608585324592937, | |
| "grad_norm": 7.434966087341309, | |
| "learning_rate": 8.002051282051284e-06, | |
| "loss": 0.1484, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.8714315923028124, | |
| "grad_norm": 4.955178737640381, | |
| "learning_rate": 7.98923076923077e-06, | |
| "loss": 0.1304, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.8820046521463312, | |
| "grad_norm": 6.121462821960449, | |
| "learning_rate": 7.976410256410257e-06, | |
| "loss": 0.1676, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.89257771198985, | |
| "grad_norm": 5.908482551574707, | |
| "learning_rate": 7.963589743589744e-06, | |
| "loss": 0.1463, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.9031507718333684, | |
| "grad_norm": 4.962184906005859, | |
| "learning_rate": 7.950769230769233e-06, | |
| "loss": 0.1494, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.9137238316768874, | |
| "grad_norm": 6.131477355957031, | |
| "learning_rate": 7.93794871794872e-06, | |
| "loss": 0.146, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.924296891520406, | |
| "grad_norm": 7.316920280456543, | |
| "learning_rate": 7.925128205128205e-06, | |
| "loss": 0.1451, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.9348699513639247, | |
| "grad_norm": 6.914496898651123, | |
| "learning_rate": 7.912307692307693e-06, | |
| "loss": 0.1263, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.9454430112074435, | |
| "grad_norm": 6.739365100860596, | |
| "learning_rate": 7.89948717948718e-06, | |
| "loss": 0.1467, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.9560160710509622, | |
| "grad_norm": 7.861861705780029, | |
| "learning_rate": 7.886666666666667e-06, | |
| "loss": 0.1582, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.966589130894481, | |
| "grad_norm": 8.007691383361816, | |
| "learning_rate": 7.873846153846154e-06, | |
| "loss": 0.1569, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.9771621907379995, | |
| "grad_norm": 9.34388542175293, | |
| "learning_rate": 7.861025641025642e-06, | |
| "loss": 0.1419, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.9877352505815185, | |
| "grad_norm": 7.265613555908203, | |
| "learning_rate": 7.848205128205129e-06, | |
| "loss": 0.1436, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.998308310425037, | |
| "grad_norm": 5.50709867477417, | |
| "learning_rate": 7.835384615384616e-06, | |
| "loss": 0.1467, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.0088813702685555, | |
| "grad_norm": 4.5442962646484375, | |
| "learning_rate": 7.822564102564103e-06, | |
| "loss": 0.0813, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.0194544301120745, | |
| "grad_norm": 6.023507595062256, | |
| "learning_rate": 7.80974358974359e-06, | |
| "loss": 0.0902, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.030027489955593, | |
| "grad_norm": 2.5194263458251953, | |
| "learning_rate": 7.796923076923078e-06, | |
| "loss": 0.092, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.040600549799112, | |
| "grad_norm": 5.59611701965332, | |
| "learning_rate": 7.784102564102565e-06, | |
| "loss": 0.0888, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.0511736096426305, | |
| "grad_norm": 5.128935813903809, | |
| "learning_rate": 7.771282051282052e-06, | |
| "loss": 0.0886, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.0617466694861495, | |
| "grad_norm": 4.068778038024902, | |
| "learning_rate": 7.758461538461538e-06, | |
| "loss": 0.0754, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.072319729329668, | |
| "grad_norm": 3.985748052597046, | |
| "learning_rate": 7.745641025641027e-06, | |
| "loss": 0.0882, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.0828927891731865, | |
| "grad_norm": 3.7527148723602295, | |
| "learning_rate": 7.732820512820514e-06, | |
| "loss": 0.0752, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.0934658490167055, | |
| "grad_norm": 6.429442405700684, | |
| "learning_rate": 7.72e-06, | |
| "loss": 0.0999, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.104038908860224, | |
| "grad_norm": 3.7136788368225098, | |
| "learning_rate": 7.707179487179487e-06, | |
| "loss": 0.0745, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.114611968703743, | |
| "grad_norm": 3.5466103553771973, | |
| "learning_rate": 7.694358974358976e-06, | |
| "loss": 0.0807, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.114611968703743, | |
| "eval_loss": 0.16425499320030212, | |
| "eval_runtime": 474.1617, | |
| "eval_samples_per_second": 7.679, | |
| "eval_steps_per_second": 0.962, | |
| "eval_wer": 0.14989094621560164, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.1251850285472615, | |
| "grad_norm": 4.999958515167236, | |
| "learning_rate": 7.681538461538463e-06, | |
| "loss": 0.0804, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 2.13575808839078, | |
| "grad_norm": 5.525317668914795, | |
| "learning_rate": 7.66871794871795e-06, | |
| "loss": 0.1002, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 2.146331148234299, | |
| "grad_norm": 4.599583148956299, | |
| "learning_rate": 7.655897435897436e-06, | |
| "loss": 0.0868, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 2.1569042080778176, | |
| "grad_norm": 5.12149715423584, | |
| "learning_rate": 7.643076923076925e-06, | |
| "loss": 0.0937, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 2.1674772679213365, | |
| "grad_norm": 3.3281843662261963, | |
| "learning_rate": 7.630256410256412e-06, | |
| "loss": 0.0748, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 2.178050327764855, | |
| "grad_norm": 5.142365455627441, | |
| "learning_rate": 7.617435897435898e-06, | |
| "loss": 0.0931, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.188623387608374, | |
| "grad_norm": 3.4713194370269775, | |
| "learning_rate": 7.604615384615385e-06, | |
| "loss": 0.0747, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 2.1991964474518926, | |
| "grad_norm": 3.9539084434509277, | |
| "learning_rate": 7.591794871794872e-06, | |
| "loss": 0.0917, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.209769507295411, | |
| "grad_norm": 6.873681545257568, | |
| "learning_rate": 7.578974358974359e-06, | |
| "loss": 0.0842, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 2.22034256713893, | |
| "grad_norm": 3.6090524196624756, | |
| "learning_rate": 7.566153846153847e-06, | |
| "loss": 0.0779, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.2309156269824486, | |
| "grad_norm": 4.458488464355469, | |
| "learning_rate": 7.553333333333334e-06, | |
| "loss": 0.0854, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 2.2414886868259676, | |
| "grad_norm": 4.1330718994140625, | |
| "learning_rate": 7.540512820512821e-06, | |
| "loss": 0.0674, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.252061746669486, | |
| "grad_norm": 3.256821393966675, | |
| "learning_rate": 7.527692307692308e-06, | |
| "loss": 0.0787, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 2.262634806513005, | |
| "grad_norm": 3.738051176071167, | |
| "learning_rate": 7.514871794871795e-06, | |
| "loss": 0.0844, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.2732078663565236, | |
| "grad_norm": 6.081026554107666, | |
| "learning_rate": 7.5020512820512826e-06, | |
| "loss": 0.0794, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 2.283780926200042, | |
| "grad_norm": 4.990882396697998, | |
| "learning_rate": 7.489230769230769e-06, | |
| "loss": 0.0806, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.294353986043561, | |
| "grad_norm": 3.432616710662842, | |
| "learning_rate": 7.476410256410257e-06, | |
| "loss": 0.0961, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 2.3049270458870796, | |
| "grad_norm": 9.63456916809082, | |
| "learning_rate": 7.463589743589744e-06, | |
| "loss": 0.0801, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.3155001057305986, | |
| "grad_norm": 4.464268207550049, | |
| "learning_rate": 7.4507692307692316e-06, | |
| "loss": 0.0796, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 2.326073165574117, | |
| "grad_norm": 3.0448217391967773, | |
| "learning_rate": 7.437948717948718e-06, | |
| "loss": 0.0718, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.3366462254176357, | |
| "grad_norm": 6.034780979156494, | |
| "learning_rate": 7.425128205128206e-06, | |
| "loss": 0.1072, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 2.3472192852611546, | |
| "grad_norm": 5.285013675689697, | |
| "learning_rate": 7.412307692307693e-06, | |
| "loss": 0.094, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.357792345104673, | |
| "grad_norm": 4.443994998931885, | |
| "learning_rate": 7.3994871794871806e-06, | |
| "loss": 0.0897, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 2.368365404948192, | |
| "grad_norm": 6.323433876037598, | |
| "learning_rate": 7.386666666666667e-06, | |
| "loss": 0.0774, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.3789384647917107, | |
| "grad_norm": 7.71209192276001, | |
| "learning_rate": 7.373846153846155e-06, | |
| "loss": 0.097, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 2.389511524635229, | |
| "grad_norm": 2.249687910079956, | |
| "learning_rate": 7.361025641025642e-06, | |
| "loss": 0.1082, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.400084584478748, | |
| "grad_norm": 4.210668087005615, | |
| "learning_rate": 7.3482051282051295e-06, | |
| "loss": 0.0649, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 2.4106576443222667, | |
| "grad_norm": 5.31463098526001, | |
| "learning_rate": 7.335384615384616e-06, | |
| "loss": 0.0924, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.4212307041657857, | |
| "grad_norm": 5.068974494934082, | |
| "learning_rate": 7.322564102564104e-06, | |
| "loss": 0.0783, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 2.431803764009304, | |
| "grad_norm": 5.510648727416992, | |
| "learning_rate": 7.309743589743591e-06, | |
| "loss": 0.0791, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.442376823852823, | |
| "grad_norm": 5.710975646972656, | |
| "learning_rate": 7.296923076923077e-06, | |
| "loss": 0.0842, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 2.4529498836963417, | |
| "grad_norm": 4.2290215492248535, | |
| "learning_rate": 7.2841025641025645e-06, | |
| "loss": 0.0827, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.4635229435398607, | |
| "grad_norm": 5.621830940246582, | |
| "learning_rate": 7.271282051282051e-06, | |
| "loss": 0.0809, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 2.474096003383379, | |
| "grad_norm": 2.6628761291503906, | |
| "learning_rate": 7.258461538461539e-06, | |
| "loss": 0.0784, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.4846690632268977, | |
| "grad_norm": 5.889407634735107, | |
| "learning_rate": 7.245641025641026e-06, | |
| "loss": 0.0908, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.4952421230704167, | |
| "grad_norm": 4.617676258087158, | |
| "learning_rate": 7.2328205128205135e-06, | |
| "loss": 0.0883, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.5058151829139352, | |
| "grad_norm": 4.988363265991211, | |
| "learning_rate": 7.22e-06, | |
| "loss": 0.1109, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 2.516388242757454, | |
| "grad_norm": 5.262870788574219, | |
| "learning_rate": 7.207179487179487e-06, | |
| "loss": 0.1085, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.5269613026009727, | |
| "grad_norm": 19.423404693603516, | |
| "learning_rate": 7.194358974358975e-06, | |
| "loss": 0.0868, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 2.5375343624444913, | |
| "grad_norm": 6.528380870819092, | |
| "learning_rate": 7.181538461538462e-06, | |
| "loss": 0.089, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.5375343624444913, | |
| "eval_loss": 0.15615449845790863, | |
| "eval_runtime": 479.6175, | |
| "eval_samples_per_second": 7.591, | |
| "eval_steps_per_second": 0.951, | |
| "eval_wer": 0.13476263399693722, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.5481074222880102, | |
| "grad_norm": 4.123988628387451, | |
| "learning_rate": 7.168717948717949e-06, | |
| "loss": 0.085, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 2.5586804821315288, | |
| "grad_norm": 3.1818017959594727, | |
| "learning_rate": 7.155897435897436e-06, | |
| "loss": 0.0906, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.5692535419750477, | |
| "grad_norm": 5.67425537109375, | |
| "learning_rate": 7.143076923076924e-06, | |
| "loss": 0.0928, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 2.5798266018185663, | |
| "grad_norm": 4.55906867980957, | |
| "learning_rate": 7.130256410256411e-06, | |
| "loss": 0.0732, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.590399661662085, | |
| "grad_norm": 3.958922863006592, | |
| "learning_rate": 7.117435897435898e-06, | |
| "loss": 0.0867, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 2.6009727215056038, | |
| "grad_norm": 4.0345306396484375, | |
| "learning_rate": 7.104615384615385e-06, | |
| "loss": 0.0636, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.6115457813491223, | |
| "grad_norm": 6.317999839782715, | |
| "learning_rate": 7.091794871794873e-06, | |
| "loss": 0.0816, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 2.6221188411926413, | |
| "grad_norm": 7.474951267242432, | |
| "learning_rate": 7.07897435897436e-06, | |
| "loss": 0.0953, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.63269190103616, | |
| "grad_norm": 4.363856315612793, | |
| "learning_rate": 7.066153846153847e-06, | |
| "loss": 0.0931, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 2.6432649608796783, | |
| "grad_norm": 6.427609920501709, | |
| "learning_rate": 7.053333333333334e-06, | |
| "loss": 0.0794, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.6538380207231973, | |
| "grad_norm": 5.55384635925293, | |
| "learning_rate": 7.040512820512822e-06, | |
| "loss": 0.0892, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 2.6644110805667163, | |
| "grad_norm": 5.477039813995361, | |
| "learning_rate": 7.027692307692309e-06, | |
| "loss": 0.0681, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.674984140410235, | |
| "grad_norm": 4.115767955780029, | |
| "learning_rate": 7.014871794871796e-06, | |
| "loss": 0.0673, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 2.6855572002537533, | |
| "grad_norm": 3.343461751937866, | |
| "learning_rate": 7.002051282051283e-06, | |
| "loss": 0.0651, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.6961302600972723, | |
| "grad_norm": 5.022345542907715, | |
| "learning_rate": 6.989230769230769e-06, | |
| "loss": 0.0727, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 2.706703319940791, | |
| "grad_norm": 5.0488128662109375, | |
| "learning_rate": 6.976410256410257e-06, | |
| "loss": 0.0808, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.71727637978431, | |
| "grad_norm": 5.218719005584717, | |
| "learning_rate": 6.963589743589744e-06, | |
| "loss": 0.0955, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 2.7278494396278283, | |
| "grad_norm": 4.660473346710205, | |
| "learning_rate": 6.950769230769231e-06, | |
| "loss": 0.0855, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.738422499471347, | |
| "grad_norm": 5.673701286315918, | |
| "learning_rate": 6.937948717948718e-06, | |
| "loss": 0.0899, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 2.748995559314866, | |
| "grad_norm": 4.193291187286377, | |
| "learning_rate": 6.925128205128206e-06, | |
| "loss": 0.0986, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.7595686191583844, | |
| "grad_norm": 3.0398550033569336, | |
| "learning_rate": 6.912307692307693e-06, | |
| "loss": 0.0747, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 2.7701416790019033, | |
| "grad_norm": 4.971275806427002, | |
| "learning_rate": 6.899487179487179e-06, | |
| "loss": 0.0829, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.780714738845422, | |
| "grad_norm": 5.15726375579834, | |
| "learning_rate": 6.886666666666667e-06, | |
| "loss": 0.074, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 2.7912877986889404, | |
| "grad_norm": 4.134614944458008, | |
| "learning_rate": 6.873846153846154e-06, | |
| "loss": 0.0764, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.8018608585324594, | |
| "grad_norm": 3.367145538330078, | |
| "learning_rate": 6.861025641025642e-06, | |
| "loss": 0.085, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 2.812433918375978, | |
| "grad_norm": 5.638070106506348, | |
| "learning_rate": 6.848205128205128e-06, | |
| "loss": 0.0687, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.823006978219497, | |
| "grad_norm": 5.024541854858398, | |
| "learning_rate": 6.835384615384616e-06, | |
| "loss": 0.0778, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 2.8335800380630154, | |
| "grad_norm": 2.271406412124634, | |
| "learning_rate": 6.822564102564103e-06, | |
| "loss": 0.0755, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.844153097906534, | |
| "grad_norm": 3.323329210281372, | |
| "learning_rate": 6.8097435897435906e-06, | |
| "loss": 0.0762, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 2.854726157750053, | |
| "grad_norm": 3.916624069213867, | |
| "learning_rate": 6.796923076923077e-06, | |
| "loss": 0.0985, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.8652992175935714, | |
| "grad_norm": 4.36145544052124, | |
| "learning_rate": 6.784102564102565e-06, | |
| "loss": 0.0815, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 2.8758722774370904, | |
| "grad_norm": 3.7215347290039062, | |
| "learning_rate": 6.771282051282052e-06, | |
| "loss": 0.0721, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.886445337280609, | |
| "grad_norm": 3.406437397003174, | |
| "learning_rate": 6.7584615384615396e-06, | |
| "loss": 0.0794, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 2.8970183971241275, | |
| "grad_norm": 4.781268119812012, | |
| "learning_rate": 6.745641025641026e-06, | |
| "loss": 0.0831, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.9075914569676464, | |
| "grad_norm": 7.577853202819824, | |
| "learning_rate": 6.732820512820514e-06, | |
| "loss": 0.0837, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 2.9181645168111654, | |
| "grad_norm": 4.966701984405518, | |
| "learning_rate": 6.720000000000001e-06, | |
| "loss": 0.0882, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.928737576654684, | |
| "grad_norm": 4.185604572296143, | |
| "learning_rate": 6.7071794871794886e-06, | |
| "loss": 0.0894, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 2.9393106364982025, | |
| "grad_norm": 6.426883697509766, | |
| "learning_rate": 6.694358974358975e-06, | |
| "loss": 0.0762, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.9498836963417214, | |
| "grad_norm": 5.418060779571533, | |
| "learning_rate": 6.681538461538463e-06, | |
| "loss": 0.0837, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 2.96045675618524, | |
| "grad_norm": 7.900200366973877, | |
| "learning_rate": 6.668717948717949e-06, | |
| "loss": 0.0883, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.96045675618524, | |
| "eval_loss": 0.1452471762895584, | |
| "eval_runtime": 481.3751, | |
| "eval_samples_per_second": 7.564, | |
| "eval_steps_per_second": 0.947, | |
| "eval_wer": 0.12678082509629218, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.971029816028759, | |
| "grad_norm": 7.546687126159668, | |
| "learning_rate": 6.655897435897436e-06, | |
| "loss": 0.083, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 2.9816028758722775, | |
| "grad_norm": 4.659261703491211, | |
| "learning_rate": 6.6430769230769235e-06, | |
| "loss": 0.0789, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.992175935715796, | |
| "grad_norm": 5.171137809753418, | |
| "learning_rate": 6.63025641025641e-06, | |
| "loss": 0.0843, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 3.002748995559315, | |
| "grad_norm": 2.0631213188171387, | |
| "learning_rate": 6.617435897435898e-06, | |
| "loss": 0.0597, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 3.0133220554028335, | |
| "grad_norm": 1.6943330764770508, | |
| "learning_rate": 6.604615384615385e-06, | |
| "loss": 0.0441, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 3.0238951152463525, | |
| "grad_norm": 2.18405818939209, | |
| "learning_rate": 6.5917948717948725e-06, | |
| "loss": 0.0399, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 3.034468175089871, | |
| "grad_norm": 4.707269668579102, | |
| "learning_rate": 6.578974358974359e-06, | |
| "loss": 0.0497, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 3.0450412349333895, | |
| "grad_norm": 2.237076759338379, | |
| "learning_rate": 6.566153846153846e-06, | |
| "loss": 0.0381, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 3.0556142947769085, | |
| "grad_norm": 2.0544979572296143, | |
| "learning_rate": 6.553333333333334e-06, | |
| "loss": 0.0458, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 3.066187354620427, | |
| "grad_norm": 3.0701844692230225, | |
| "learning_rate": 6.540512820512821e-06, | |
| "loss": 0.0641, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 3.076760414463946, | |
| "grad_norm": 2.446314573287964, | |
| "learning_rate": 6.527692307692308e-06, | |
| "loss": 0.0398, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 3.0873334743074645, | |
| "grad_norm": 1.1977494955062866, | |
| "learning_rate": 6.514871794871795e-06, | |
| "loss": 0.0347, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 3.0979065341509835, | |
| "grad_norm": 3.8696768283843994, | |
| "learning_rate": 6.5025641025641026e-06, | |
| "loss": 0.0454, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 3.108479593994502, | |
| "grad_norm": 4.348515033721924, | |
| "learning_rate": 6.48974358974359e-06, | |
| "loss": 0.0358, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 3.1190526538380206, | |
| "grad_norm": 3.3741064071655273, | |
| "learning_rate": 6.476923076923077e-06, | |
| "loss": 0.0477, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 3.1296257136815395, | |
| "grad_norm": 1.3603962659835815, | |
| "learning_rate": 6.464102564102565e-06, | |
| "loss": 0.0424, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 3.140198773525058, | |
| "grad_norm": 2.087542772293091, | |
| "learning_rate": 6.4512820512820516e-06, | |
| "loss": 0.034, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 3.150771833368577, | |
| "grad_norm": 3.6920878887176514, | |
| "learning_rate": 6.438461538461539e-06, | |
| "loss": 0.0441, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 3.1613448932120956, | |
| "grad_norm": 2.65889573097229, | |
| "learning_rate": 6.425641025641026e-06, | |
| "loss": 0.0399, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 3.1719179530556145, | |
| "grad_norm": 6.821660995483398, | |
| "learning_rate": 6.412820512820514e-06, | |
| "loss": 0.0331, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.182491012899133, | |
| "grad_norm": 3.360375165939331, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.0424, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 3.1930640727426516, | |
| "grad_norm": 2.295417547225952, | |
| "learning_rate": 6.387179487179488e-06, | |
| "loss": 0.0406, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 3.2036371325861706, | |
| "grad_norm": 2.809596300125122, | |
| "learning_rate": 6.374358974358975e-06, | |
| "loss": 0.0536, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 3.214210192429689, | |
| "grad_norm": 2.86702823638916, | |
| "learning_rate": 6.361538461538463e-06, | |
| "loss": 0.0401, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 3.224783252273208, | |
| "grad_norm": 3.1539106369018555, | |
| "learning_rate": 6.3487179487179495e-06, | |
| "loss": 0.038, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 3.2353563121167266, | |
| "grad_norm": 5.205074787139893, | |
| "learning_rate": 6.335897435897436e-06, | |
| "loss": 0.0402, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 3.245929371960245, | |
| "grad_norm": 4.798672199249268, | |
| "learning_rate": 6.323076923076924e-06, | |
| "loss": 0.0412, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 3.256502431803764, | |
| "grad_norm": 2.3215603828430176, | |
| "learning_rate": 6.310256410256411e-06, | |
| "loss": 0.0389, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 3.2670754916472826, | |
| "grad_norm": 2.9327590465545654, | |
| "learning_rate": 6.2974358974358985e-06, | |
| "loss": 0.0327, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 3.2776485514908016, | |
| "grad_norm": 2.3523976802825928, | |
| "learning_rate": 6.284615384615385e-06, | |
| "loss": 0.0415, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 3.28822161133432, | |
| "grad_norm": 2.7430717945098877, | |
| "learning_rate": 6.271794871794872e-06, | |
| "loss": 0.038, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 3.2987946711778386, | |
| "grad_norm": 3.922346830368042, | |
| "learning_rate": 6.258974358974359e-06, | |
| "loss": 0.0505, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 3.3093677310213576, | |
| "grad_norm": 3.7691900730133057, | |
| "learning_rate": 6.246153846153846e-06, | |
| "loss": 0.037, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 3.319940790864876, | |
| "grad_norm": 5.420505523681641, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.0403, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 3.330513850708395, | |
| "grad_norm": 2.5405962467193604, | |
| "learning_rate": 6.22051282051282e-06, | |
| "loss": 0.0481, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 3.3410869105519136, | |
| "grad_norm": 5.486624717712402, | |
| "learning_rate": 6.207692307692308e-06, | |
| "loss": 0.0499, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 3.3516599703954326, | |
| "grad_norm": 7.166432857513428, | |
| "learning_rate": 6.194871794871795e-06, | |
| "loss": 0.0455, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 3.362233030238951, | |
| "grad_norm": 2.30519437789917, | |
| "learning_rate": 6.1820512820512825e-06, | |
| "loss": 0.0526, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 3.3728060900824697, | |
| "grad_norm": 1.9276854991912842, | |
| "learning_rate": 6.169230769230769e-06, | |
| "loss": 0.0496, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 3.3833791499259886, | |
| "grad_norm": 2.095248222351074, | |
| "learning_rate": 6.156410256410257e-06, | |
| "loss": 0.0368, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.3833791499259886, | |
| "eval_loss": 0.14458830654621124, | |
| "eval_runtime": 482.4832, | |
| "eval_samples_per_second": 7.546, | |
| "eval_steps_per_second": 0.945, | |
| "eval_wer": 0.13239593484616455, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.393952209769507, | |
| "grad_norm": 1.7885074615478516, | |
| "learning_rate": 6.143589743589744e-06, | |
| "loss": 0.0361, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 3.404525269613026, | |
| "grad_norm": 4.022132873535156, | |
| "learning_rate": 6.1307692307692315e-06, | |
| "loss": 0.0463, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 3.4150983294565447, | |
| "grad_norm": 2.1497411727905273, | |
| "learning_rate": 6.117948717948718e-06, | |
| "loss": 0.0427, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 3.4256713893000637, | |
| "grad_norm": 1.6746113300323486, | |
| "learning_rate": 6.105128205128206e-06, | |
| "loss": 0.0344, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 3.436244449143582, | |
| "grad_norm": 3.6092941761016846, | |
| "learning_rate": 6.092307692307693e-06, | |
| "loss": 0.0469, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 3.4468175089871007, | |
| "grad_norm": 4.466283798217773, | |
| "learning_rate": 6.0794871794871805e-06, | |
| "loss": 0.0423, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 3.4573905688306197, | |
| "grad_norm": 2.550199508666992, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 0.0542, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 3.467963628674138, | |
| "grad_norm": 2.781538963317871, | |
| "learning_rate": 6.053846153846155e-06, | |
| "loss": 0.0423, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 3.478536688517657, | |
| "grad_norm": 1.9282548427581787, | |
| "learning_rate": 6.041025641025642e-06, | |
| "loss": 0.0386, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 3.4891097483611757, | |
| "grad_norm": 2.867671012878418, | |
| "learning_rate": 6.028205128205129e-06, | |
| "loss": 0.0364, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 3.4996828082046942, | |
| "grad_norm": 2.200422763824463, | |
| "learning_rate": 6.015384615384616e-06, | |
| "loss": 0.0396, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 3.510255868048213, | |
| "grad_norm": 2.84183931350708, | |
| "learning_rate": 6.002564102564103e-06, | |
| "loss": 0.0432, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 3.5208289278917317, | |
| "grad_norm": 3.7579634189605713, | |
| "learning_rate": 5.989743589743591e-06, | |
| "loss": 0.0503, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 3.5314019877352507, | |
| "grad_norm": 5.463958740234375, | |
| "learning_rate": 5.976923076923078e-06, | |
| "loss": 0.0367, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 3.5419750475787692, | |
| "grad_norm": 2.921443462371826, | |
| "learning_rate": 5.9641025641025644e-06, | |
| "loss": 0.0474, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 3.5525481074222878, | |
| "grad_norm": 2.1998775005340576, | |
| "learning_rate": 5.951282051282051e-06, | |
| "loss": 0.0393, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 3.5631211672658067, | |
| "grad_norm": 2.0558321475982666, | |
| "learning_rate": 5.938461538461538e-06, | |
| "loss": 0.0434, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 3.5736942271093257, | |
| "grad_norm": 4.614722728729248, | |
| "learning_rate": 5.925641025641026e-06, | |
| "loss": 0.0419, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 3.5842672869528442, | |
| "grad_norm": 3.2002508640289307, | |
| "learning_rate": 5.912820512820513e-06, | |
| "loss": 0.0579, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 3.5948403467963628, | |
| "grad_norm": 2.0797934532165527, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.047, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 3.6054134066398817, | |
| "grad_norm": 1.3923156261444092, | |
| "learning_rate": 5.887179487179487e-06, | |
| "loss": 0.035, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 3.6159864664834003, | |
| "grad_norm": 2.350816011428833, | |
| "learning_rate": 5.874358974358975e-06, | |
| "loss": 0.0678, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 3.6265595263269192, | |
| "grad_norm": 3.1680469512939453, | |
| "learning_rate": 5.861538461538462e-06, | |
| "loss": 0.055, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 3.6371325861704378, | |
| "grad_norm": 2.174107551574707, | |
| "learning_rate": 5.848717948717949e-06, | |
| "loss": 0.0475, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 3.6477056460139563, | |
| "grad_norm": 3.112668037414551, | |
| "learning_rate": 5.835897435897436e-06, | |
| "loss": 0.0447, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 3.6582787058574753, | |
| "grad_norm": 4.258768558502197, | |
| "learning_rate": 5.823076923076924e-06, | |
| "loss": 0.0322, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 3.668851765700994, | |
| "grad_norm": 3.305208683013916, | |
| "learning_rate": 5.8102564102564106e-06, | |
| "loss": 0.0454, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 3.679424825544513, | |
| "grad_norm": 2.314253568649292, | |
| "learning_rate": 5.797435897435898e-06, | |
| "loss": 0.0376, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 3.6899978853880313, | |
| "grad_norm": 2.856414318084717, | |
| "learning_rate": 5.784615384615385e-06, | |
| "loss": 0.0396, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 3.70057094523155, | |
| "grad_norm": 1.2516982555389404, | |
| "learning_rate": 5.771794871794873e-06, | |
| "loss": 0.0323, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 3.711144005075069, | |
| "grad_norm": 3.374429225921631, | |
| "learning_rate": 5.7589743589743596e-06, | |
| "loss": 0.0487, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 3.7217170649185873, | |
| "grad_norm": 3.7010583877563477, | |
| "learning_rate": 5.746153846153847e-06, | |
| "loss": 0.0489, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 3.7322901247621063, | |
| "grad_norm": 1.8617634773254395, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 0.0376, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 3.742863184605625, | |
| "grad_norm": 1.5360052585601807, | |
| "learning_rate": 5.720512820512821e-06, | |
| "loss": 0.0392, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 3.7534362444491434, | |
| "grad_norm": 3.569289207458496, | |
| "learning_rate": 5.7076923076923086e-06, | |
| "loss": 0.0506, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 3.7640093042926623, | |
| "grad_norm": 4.679437160491943, | |
| "learning_rate": 5.694871794871795e-06, | |
| "loss": 0.0397, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 3.774582364136181, | |
| "grad_norm": 1.4421815872192383, | |
| "learning_rate": 5.682051282051283e-06, | |
| "loss": 0.0507, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 3.7851554239797, | |
| "grad_norm": 4.020344257354736, | |
| "learning_rate": 5.66923076923077e-06, | |
| "loss": 0.0426, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 3.7957284838232184, | |
| "grad_norm": 1.8350247144699097, | |
| "learning_rate": 5.6564102564102575e-06, | |
| "loss": 0.0577, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 3.806301543666737, | |
| "grad_norm": 4.344027519226074, | |
| "learning_rate": 5.6435897435897435e-06, | |
| "loss": 0.0463, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.806301543666737, | |
| "eval_loss": 0.1400885134935379, | |
| "eval_runtime": 478.4392, | |
| "eval_samples_per_second": 7.61, | |
| "eval_steps_per_second": 0.953, | |
| "eval_wer": 0.12863705972434916, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.816874603510256, | |
| "grad_norm": 1.8994994163513184, | |
| "learning_rate": 5.63076923076923e-06, | |
| "loss": 0.0396, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 3.827447663353775, | |
| "grad_norm": 2.9287285804748535, | |
| "learning_rate": 5.617948717948718e-06, | |
| "loss": 0.0353, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 3.8380207231972934, | |
| "grad_norm": 5.816341876983643, | |
| "learning_rate": 5.605128205128205e-06, | |
| "loss": 0.0416, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 3.848593783040812, | |
| "grad_norm": 1.939314603805542, | |
| "learning_rate": 5.5923076923076925e-06, | |
| "loss": 0.0356, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 3.859166842884331, | |
| "grad_norm": 2.6920735836029053, | |
| "learning_rate": 5.579487179487179e-06, | |
| "loss": 0.0361, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 3.8697399027278494, | |
| "grad_norm": 3.10347056388855, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 0.0465, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 3.8803129625713684, | |
| "grad_norm": 3.6406362056732178, | |
| "learning_rate": 5.553846153846154e-06, | |
| "loss": 0.0451, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 3.890886022414887, | |
| "grad_norm": 1.8818650245666504, | |
| "learning_rate": 5.5410256410256415e-06, | |
| "loss": 0.0399, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 3.9014590822584054, | |
| "grad_norm": 2.670872688293457, | |
| "learning_rate": 5.528205128205128e-06, | |
| "loss": 0.0394, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 3.9120321421019244, | |
| "grad_norm": 6.323610782623291, | |
| "learning_rate": 5.515384615384616e-06, | |
| "loss": 0.0469, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 3.922605201945443, | |
| "grad_norm": 3.8592729568481445, | |
| "learning_rate": 5.502564102564103e-06, | |
| "loss": 0.0432, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 3.933178261788962, | |
| "grad_norm": 1.5597033500671387, | |
| "learning_rate": 5.4897435897435905e-06, | |
| "loss": 0.0326, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 3.9437513216324804, | |
| "grad_norm": 1.413038969039917, | |
| "learning_rate": 5.476923076923077e-06, | |
| "loss": 0.0439, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 3.954324381475999, | |
| "grad_norm": 1.5831608772277832, | |
| "learning_rate": 5.464102564102565e-06, | |
| "loss": 0.0406, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 3.964897441319518, | |
| "grad_norm": 4.640722274780273, | |
| "learning_rate": 5.451282051282052e-06, | |
| "loss": 0.046, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 3.9754705011630365, | |
| "grad_norm": 2.2848503589630127, | |
| "learning_rate": 5.4384615384615395e-06, | |
| "loss": 0.039, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 3.9860435610065554, | |
| "grad_norm": 3.464656352996826, | |
| "learning_rate": 5.425641025641026e-06, | |
| "loss": 0.0482, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 3.996616620850074, | |
| "grad_norm": 1.6633790731430054, | |
| "learning_rate": 5.412820512820514e-06, | |
| "loss": 0.0458, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 4.0071896806935925, | |
| "grad_norm": 3.9742796421051025, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.0301, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 4.017762740537111, | |
| "grad_norm": 0.8984088897705078, | |
| "learning_rate": 5.387179487179488e-06, | |
| "loss": 0.0171, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.02833580038063, | |
| "grad_norm": 1.1975101232528687, | |
| "learning_rate": 5.374358974358975e-06, | |
| "loss": 0.0346, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 4.038908860224149, | |
| "grad_norm": 1.3089258670806885, | |
| "learning_rate": 5.361538461538462e-06, | |
| "loss": 0.0261, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 4.0494819200676675, | |
| "grad_norm": 1.7242599725723267, | |
| "learning_rate": 5.34871794871795e-06, | |
| "loss": 0.024, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 4.060054979911186, | |
| "grad_norm": 3.1359024047851562, | |
| "learning_rate": 5.335897435897436e-06, | |
| "loss": 0.019, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 4.070628039754705, | |
| "grad_norm": 1.996058702468872, | |
| "learning_rate": 5.323076923076923e-06, | |
| "loss": 0.0165, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 4.081201099598224, | |
| "grad_norm": 1.165366530418396, | |
| "learning_rate": 5.31025641025641e-06, | |
| "loss": 0.0241, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 4.0917741594417425, | |
| "grad_norm": 1.7959648370742798, | |
| "learning_rate": 5.297435897435897e-06, | |
| "loss": 0.0193, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 4.102347219285261, | |
| "grad_norm": 3.0931942462921143, | |
| "learning_rate": 5.284615384615385e-06, | |
| "loss": 0.0188, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 4.1129202791287796, | |
| "grad_norm": 1.2411391735076904, | |
| "learning_rate": 5.271794871794872e-06, | |
| "loss": 0.0315, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 4.123493338972299, | |
| "grad_norm": 4.090996265411377, | |
| "learning_rate": 5.258974358974359e-06, | |
| "loss": 0.0216, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 4.1340663988158175, | |
| "grad_norm": 1.1179065704345703, | |
| "learning_rate": 5.246153846153846e-06, | |
| "loss": 0.0234, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 4.144639458659336, | |
| "grad_norm": 1.4874383211135864, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.0187, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 4.155212518502855, | |
| "grad_norm": 0.8263124823570251, | |
| "learning_rate": 5.220512820512821e-06, | |
| "loss": 0.0253, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 4.165785578346373, | |
| "grad_norm": 2.70497727394104, | |
| "learning_rate": 5.207692307692308e-06, | |
| "loss": 0.0228, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 4.1763586381898925, | |
| "grad_norm": 4.427598476409912, | |
| "learning_rate": 5.194871794871795e-06, | |
| "loss": 0.0248, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 4.186931698033411, | |
| "grad_norm": 1.1242289543151855, | |
| "learning_rate": 5.182051282051283e-06, | |
| "loss": 0.0205, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 4.19750475787693, | |
| "grad_norm": 0.6889589428901672, | |
| "learning_rate": 5.16923076923077e-06, | |
| "loss": 0.0152, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 4.208077817720448, | |
| "grad_norm": 1.5763376951217651, | |
| "learning_rate": 5.156410256410257e-06, | |
| "loss": 0.0164, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 4.218650877563967, | |
| "grad_norm": 3.120882034301758, | |
| "learning_rate": 5.143589743589744e-06, | |
| "loss": 0.0249, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 4.229223937407486, | |
| "grad_norm": 1.5591440200805664, | |
| "learning_rate": 5.130769230769232e-06, | |
| "loss": 0.0278, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.229223937407486, | |
| "eval_loss": 0.14359265565872192, | |
| "eval_runtime": 472.3801, | |
| "eval_samples_per_second": 7.708, | |
| "eval_steps_per_second": 0.965, | |
| "eval_wer": 0.11810292821012576, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.239796997251005, | |
| "grad_norm": 1.2260849475860596, | |
| "learning_rate": 5.1179487179487186e-06, | |
| "loss": 0.0169, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 4.250370057094523, | |
| "grad_norm": 0.49284470081329346, | |
| "learning_rate": 5.105128205128206e-06, | |
| "loss": 0.018, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 4.260943116938042, | |
| "grad_norm": 2.0094289779663086, | |
| "learning_rate": 5.092307692307693e-06, | |
| "loss": 0.0218, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 4.27151617678156, | |
| "grad_norm": 1.3108640909194946, | |
| "learning_rate": 5.07948717948718e-06, | |
| "loss": 0.025, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 4.28208923662508, | |
| "grad_norm": 1.7406195402145386, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 0.0235, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 4.292662296468598, | |
| "grad_norm": 0.876416027545929, | |
| "learning_rate": 5.053846153846154e-06, | |
| "loss": 0.0195, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 4.303235356312117, | |
| "grad_norm": 1.6580686569213867, | |
| "learning_rate": 5.041025641025642e-06, | |
| "loss": 0.0167, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 4.313808416155635, | |
| "grad_norm": 0.5891424417495728, | |
| "learning_rate": 5.028205128205128e-06, | |
| "loss": 0.0262, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 4.324381475999155, | |
| "grad_norm": 2.2991206645965576, | |
| "learning_rate": 5.015384615384616e-06, | |
| "loss": 0.0253, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 4.334954535842673, | |
| "grad_norm": 1.3528612852096558, | |
| "learning_rate": 5.0025641025641025e-06, | |
| "loss": 0.0269, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 4.345527595686192, | |
| "grad_norm": 0.8127634525299072, | |
| "learning_rate": 4.98974358974359e-06, | |
| "loss": 0.0225, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 4.35610065552971, | |
| "grad_norm": 0.9224340915679932, | |
| "learning_rate": 4.976923076923078e-06, | |
| "loss": 0.0219, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 4.366673715373229, | |
| "grad_norm": 1.4799253940582275, | |
| "learning_rate": 4.964102564102565e-06, | |
| "loss": 0.0176, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 4.377246775216748, | |
| "grad_norm": 3.0369462966918945, | |
| "learning_rate": 4.9512820512820515e-06, | |
| "loss": 0.0207, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 4.387819835060267, | |
| "grad_norm": 0.7996551990509033, | |
| "learning_rate": 4.938461538461538e-06, | |
| "loss": 0.0236, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 4.398392894903785, | |
| "grad_norm": 0.9955740571022034, | |
| "learning_rate": 4.925641025641026e-06, | |
| "loss": 0.0274, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 4.408965954747304, | |
| "grad_norm": 1.9452871084213257, | |
| "learning_rate": 4.912820512820513e-06, | |
| "loss": 0.0164, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 4.419539014590822, | |
| "grad_norm": 10.20065975189209, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.0215, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 4.430112074434342, | |
| "grad_norm": 0.7727457880973816, | |
| "learning_rate": 4.887179487179487e-06, | |
| "loss": 0.0154, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 4.44068513427786, | |
| "grad_norm": 2.406764268875122, | |
| "learning_rate": 4.874358974358975e-06, | |
| "loss": 0.0197, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 4.451258194121379, | |
| "grad_norm": 1.1302127838134766, | |
| "learning_rate": 4.861538461538462e-06, | |
| "loss": 0.0213, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 4.461831253964897, | |
| "grad_norm": 1.5602991580963135, | |
| "learning_rate": 4.8487179487179495e-06, | |
| "loss": 0.0222, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 4.472404313808416, | |
| "grad_norm": 2.2636196613311768, | |
| "learning_rate": 4.835897435897436e-06, | |
| "loss": 0.0208, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 4.482977373651935, | |
| "grad_norm": 1.560081124305725, | |
| "learning_rate": 4.823076923076924e-06, | |
| "loss": 0.0212, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 4.493550433495454, | |
| "grad_norm": 1.5913350582122803, | |
| "learning_rate": 4.810256410256411e-06, | |
| "loss": 0.0202, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 4.504123493338972, | |
| "grad_norm": 2.0870747566223145, | |
| "learning_rate": 4.7974358974358985e-06, | |
| "loss": 0.0197, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 4.514696553182491, | |
| "grad_norm": 0.832817554473877, | |
| "learning_rate": 4.7846153846153845e-06, | |
| "loss": 0.0229, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 4.52526961302601, | |
| "grad_norm": 0.8193099498748779, | |
| "learning_rate": 4.771794871794872e-06, | |
| "loss": 0.0325, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 4.535842672869529, | |
| "grad_norm": 2.3790667057037354, | |
| "learning_rate": 4.758974358974359e-06, | |
| "loss": 0.0226, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 4.546415732713047, | |
| "grad_norm": 4.055891990661621, | |
| "learning_rate": 4.746153846153847e-06, | |
| "loss": 0.0339, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 4.556988792556566, | |
| "grad_norm": 1.3302873373031616, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.0205, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 4.567561852400084, | |
| "grad_norm": 1.9055285453796387, | |
| "learning_rate": 4.720512820512821e-06, | |
| "loss": 0.0246, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 4.578134912243604, | |
| "grad_norm": 0.8478215336799622, | |
| "learning_rate": 4.707692307692308e-06, | |
| "loss": 0.0283, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 4.588707972087122, | |
| "grad_norm": 1.2983384132385254, | |
| "learning_rate": 4.694871794871796e-06, | |
| "loss": 0.0214, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 4.599281031930641, | |
| "grad_norm": 0.70320063829422, | |
| "learning_rate": 4.6820512820512825e-06, | |
| "loss": 0.0344, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 4.609854091774159, | |
| "grad_norm": 1.7480090856552124, | |
| "learning_rate": 4.66923076923077e-06, | |
| "loss": 0.0198, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 4.620427151617678, | |
| "grad_norm": 1.8584214448928833, | |
| "learning_rate": 4.656410256410257e-06, | |
| "loss": 0.0198, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 4.631000211461197, | |
| "grad_norm": 3.884225368499756, | |
| "learning_rate": 4.643589743589745e-06, | |
| "loss": 0.0197, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 4.641573271304716, | |
| "grad_norm": 0.8960859775543213, | |
| "learning_rate": 4.630769230769231e-06, | |
| "loss": 0.0212, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 4.652146331148234, | |
| "grad_norm": 0.8120248317718506, | |
| "learning_rate": 4.617948717948718e-06, | |
| "loss": 0.0157, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 4.652146331148234, | |
| "eval_loss": 0.14061123132705688, | |
| "eval_runtime": 474.4412, | |
| "eval_samples_per_second": 7.674, | |
| "eval_steps_per_second": 0.961, | |
| "eval_wer": 0.11253422432595481, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 4.662719390991753, | |
| "grad_norm": 0.5586819648742676, | |
| "learning_rate": 4.605128205128205e-06, | |
| "loss": 0.0201, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 4.673292450835271, | |
| "grad_norm": 2.2997443675994873, | |
| "learning_rate": 4.592307692307693e-06, | |
| "loss": 0.0207, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 4.683865510678791, | |
| "grad_norm": 3.0606119632720947, | |
| "learning_rate": 4.57948717948718e-06, | |
| "loss": 0.0261, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 4.694438570522309, | |
| "grad_norm": 1.2574974298477173, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.0229, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 4.705011630365828, | |
| "grad_norm": 0.962881326675415, | |
| "learning_rate": 4.553846153846154e-06, | |
| "loss": 0.0198, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 4.715584690209346, | |
| "grad_norm": 1.8664745092391968, | |
| "learning_rate": 4.5415384615384615e-06, | |
| "loss": 0.0185, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 4.726157750052865, | |
| "grad_norm": 2.5618274211883545, | |
| "learning_rate": 4.528717948717949e-06, | |
| "loss": 0.0306, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 4.736730809896384, | |
| "grad_norm": 0.9622751474380493, | |
| "learning_rate": 4.515897435897436e-06, | |
| "loss": 0.037, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 4.747303869739903, | |
| "grad_norm": 0.5940545201301575, | |
| "learning_rate": 4.503076923076924e-06, | |
| "loss": 0.0178, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 4.757876929583421, | |
| "grad_norm": 2.6667637825012207, | |
| "learning_rate": 4.4902564102564105e-06, | |
| "loss": 0.015, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 4.76844998942694, | |
| "grad_norm": 0.793354868888855, | |
| "learning_rate": 4.477435897435898e-06, | |
| "loss": 0.0299, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 4.779023049270458, | |
| "grad_norm": 4.2273030281066895, | |
| "learning_rate": 4.464615384615385e-06, | |
| "loss": 0.0165, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 4.789596109113978, | |
| "grad_norm": 2.6417057514190674, | |
| "learning_rate": 4.451794871794872e-06, | |
| "loss": 0.0232, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 4.800169168957496, | |
| "grad_norm": 2.1996712684631348, | |
| "learning_rate": 4.4389743589743595e-06, | |
| "loss": 0.015, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 4.810742228801015, | |
| "grad_norm": 5.317453384399414, | |
| "learning_rate": 4.426153846153846e-06, | |
| "loss": 0.0199, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 4.821315288644533, | |
| "grad_norm": 0.8886387944221497, | |
| "learning_rate": 4.413333333333334e-06, | |
| "loss": 0.0219, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 4.831888348488053, | |
| "grad_norm": 1.3750237226486206, | |
| "learning_rate": 4.400512820512821e-06, | |
| "loss": 0.0248, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 4.842461408331571, | |
| "grad_norm": 1.9163587093353271, | |
| "learning_rate": 4.387692307692308e-06, | |
| "loss": 0.0299, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 4.85303446817509, | |
| "grad_norm": 1.0380934476852417, | |
| "learning_rate": 4.374871794871795e-06, | |
| "loss": 0.016, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 4.863607528018608, | |
| "grad_norm": 2.2600953578948975, | |
| "learning_rate": 4.362051282051282e-06, | |
| "loss": 0.0201, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 4.874180587862127, | |
| "grad_norm": 2.316375255584717, | |
| "learning_rate": 4.34923076923077e-06, | |
| "loss": 0.0244, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 4.884753647705646, | |
| "grad_norm": 2.1449790000915527, | |
| "learning_rate": 4.336410256410257e-06, | |
| "loss": 0.0167, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 4.895326707549165, | |
| "grad_norm": 2.5895638465881348, | |
| "learning_rate": 4.323589743589744e-06, | |
| "loss": 0.0162, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 4.905899767392683, | |
| "grad_norm": 2.5529627799987793, | |
| "learning_rate": 4.310769230769231e-06, | |
| "loss": 0.0227, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 4.916472827236202, | |
| "grad_norm": 1.019690752029419, | |
| "learning_rate": 4.297948717948718e-06, | |
| "loss": 0.0307, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 4.927045887079721, | |
| "grad_norm": 1.7469873428344727, | |
| "learning_rate": 4.285128205128206e-06, | |
| "loss": 0.0137, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 4.93761894692324, | |
| "grad_norm": 1.8508704900741577, | |
| "learning_rate": 4.2723076923076925e-06, | |
| "loss": 0.0228, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 4.948192006766758, | |
| "grad_norm": 0.8989993333816528, | |
| "learning_rate": 4.25948717948718e-06, | |
| "loss": 0.0165, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 4.958765066610277, | |
| "grad_norm": 1.5571892261505127, | |
| "learning_rate": 4.246666666666667e-06, | |
| "loss": 0.0234, | |
| "step": 11725 | |
| }, | |
| { | |
| "epoch": 4.9693381264537955, | |
| "grad_norm": 0.6282488703727722, | |
| "learning_rate": 4.233846153846154e-06, | |
| "loss": 0.0194, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 4.979911186297315, | |
| "grad_norm": 0.898673415184021, | |
| "learning_rate": 4.2210256410256414e-06, | |
| "loss": 0.0177, | |
| "step": 11775 | |
| }, | |
| { | |
| "epoch": 4.990484246140833, | |
| "grad_norm": 0.9171255826950073, | |
| "learning_rate": 4.208205128205128e-06, | |
| "loss": 0.0163, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 5.001057305984352, | |
| "grad_norm": 0.9866194128990173, | |
| "learning_rate": 4.195384615384616e-06, | |
| "loss": 0.0175, | |
| "step": 11825 | |
| }, | |
| { | |
| "epoch": 5.0116303658278705, | |
| "grad_norm": 0.3489556312561035, | |
| "learning_rate": 4.182564102564103e-06, | |
| "loss": 0.0107, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 5.022203425671389, | |
| "grad_norm": 3.7504043579101562, | |
| "learning_rate": 4.1697435897435904e-06, | |
| "loss": 0.0104, | |
| "step": 11875 | |
| }, | |
| { | |
| "epoch": 5.032776485514908, | |
| "grad_norm": 1.138054609298706, | |
| "learning_rate": 4.156923076923077e-06, | |
| "loss": 0.0078, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 5.043349545358427, | |
| "grad_norm": 2.7337746620178223, | |
| "learning_rate": 4.144102564102564e-06, | |
| "loss": 0.0079, | |
| "step": 11925 | |
| }, | |
| { | |
| "epoch": 5.0539226052019455, | |
| "grad_norm": 0.4507981836795807, | |
| "learning_rate": 4.131282051282052e-06, | |
| "loss": 0.0127, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 5.064495665045464, | |
| "grad_norm": 0.9632282853126526, | |
| "learning_rate": 4.118461538461539e-06, | |
| "loss": 0.0107, | |
| "step": 11975 | |
| }, | |
| { | |
| "epoch": 5.0750687248889825, | |
| "grad_norm": 0.3650486171245575, | |
| "learning_rate": 4.105641025641026e-06, | |
| "loss": 0.0201, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.0750687248889825, | |
| "eval_loss": 0.1392413079738617, | |
| "eval_runtime": 476.9371, | |
| "eval_samples_per_second": 7.634, | |
| "eval_steps_per_second": 0.956, | |
| "eval_wer": 0.11443686481971321, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.085641784732502, | |
| "grad_norm": 0.6850853562355042, | |
| "learning_rate": 4.092820512820513e-06, | |
| "loss": 0.0168, | |
| "step": 12025 | |
| }, | |
| { | |
| "epoch": 5.0962148445760205, | |
| "grad_norm": 0.9726247787475586, | |
| "learning_rate": 4.08e-06, | |
| "loss": 0.0156, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 5.106787904419539, | |
| "grad_norm": 0.6085502505302429, | |
| "learning_rate": 4.0671794871794876e-06, | |
| "loss": 0.0092, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 5.1173609642630575, | |
| "grad_norm": 0.3284889757633209, | |
| "learning_rate": 4.054358974358974e-06, | |
| "loss": 0.0128, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 5.127934024106576, | |
| "grad_norm": 0.4705301523208618, | |
| "learning_rate": 4.041538461538462e-06, | |
| "loss": 0.0104, | |
| "step": 12125 | |
| }, | |
| { | |
| "epoch": 5.1385070839500955, | |
| "grad_norm": 1.9244158267974854, | |
| "learning_rate": 4.028717948717949e-06, | |
| "loss": 0.0081, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 5.149080143793614, | |
| "grad_norm": 2.1168432235717773, | |
| "learning_rate": 4.0158974358974366e-06, | |
| "loss": 0.0087, | |
| "step": 12175 | |
| }, | |
| { | |
| "epoch": 5.1596532036371325, | |
| "grad_norm": 1.652687430381775, | |
| "learning_rate": 4.003076923076923e-06, | |
| "loss": 0.0113, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 5.170226263480651, | |
| "grad_norm": 2.1567976474761963, | |
| "learning_rate": 3.990256410256411e-06, | |
| "loss": 0.0088, | |
| "step": 12225 | |
| }, | |
| { | |
| "epoch": 5.18079932332417, | |
| "grad_norm": 0.7141952514648438, | |
| "learning_rate": 3.977435897435898e-06, | |
| "loss": 0.0108, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 5.191372383167689, | |
| "grad_norm": 1.3821399211883545, | |
| "learning_rate": 3.964615384615385e-06, | |
| "loss": 0.0111, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 5.2019454430112075, | |
| "grad_norm": 0.7689725160598755, | |
| "learning_rate": 3.951794871794872e-06, | |
| "loss": 0.0084, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 5.212518502854726, | |
| "grad_norm": 0.7550517320632935, | |
| "learning_rate": 3.938974358974359e-06, | |
| "loss": 0.0143, | |
| "step": 12325 | |
| }, | |
| { | |
| "epoch": 5.223091562698245, | |
| "grad_norm": 0.4243580102920532, | |
| "learning_rate": 3.926153846153846e-06, | |
| "loss": 0.008, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 5.233664622541764, | |
| "grad_norm": 0.6995705366134644, | |
| "learning_rate": 3.913333333333334e-06, | |
| "loss": 0.0078, | |
| "step": 12375 | |
| }, | |
| { | |
| "epoch": 5.2442376823852825, | |
| "grad_norm": 0.3228248357772827, | |
| "learning_rate": 3.9005128205128205e-06, | |
| "loss": 0.0089, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 5.254810742228801, | |
| "grad_norm": 0.5981309413909912, | |
| "learning_rate": 3.887692307692308e-06, | |
| "loss": 0.0116, | |
| "step": 12425 | |
| }, | |
| { | |
| "epoch": 5.26538380207232, | |
| "grad_norm": 1.0151786804199219, | |
| "learning_rate": 3.874871794871795e-06, | |
| "loss": 0.0109, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 5.275956861915838, | |
| "grad_norm": 1.7748947143554688, | |
| "learning_rate": 3.862051282051283e-06, | |
| "loss": 0.0165, | |
| "step": 12475 | |
| }, | |
| { | |
| "epoch": 5.2865299217593575, | |
| "grad_norm": 3.3617355823516846, | |
| "learning_rate": 3.8492307692307695e-06, | |
| "loss": 0.0113, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 5.297102981602876, | |
| "grad_norm": 0.9399589896202087, | |
| "learning_rate": 3.836410256410257e-06, | |
| "loss": 0.0145, | |
| "step": 12525 | |
| }, | |
| { | |
| "epoch": 5.307676041446395, | |
| "grad_norm": 0.4181835651397705, | |
| "learning_rate": 3.823589743589744e-06, | |
| "loss": 0.0132, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 5.318249101289913, | |
| "grad_norm": 2.0555076599121094, | |
| "learning_rate": 3.8107692307692313e-06, | |
| "loss": 0.0081, | |
| "step": 12575 | |
| }, | |
| { | |
| "epoch": 5.328822161133432, | |
| "grad_norm": 1.8507764339447021, | |
| "learning_rate": 3.7979487179487185e-06, | |
| "loss": 0.0108, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 5.339395220976951, | |
| "grad_norm": 3.1068594455718994, | |
| "learning_rate": 3.7851282051282058e-06, | |
| "loss": 0.0146, | |
| "step": 12625 | |
| }, | |
| { | |
| "epoch": 5.34996828082047, | |
| "grad_norm": 1.1594772338867188, | |
| "learning_rate": 3.772307692307693e-06, | |
| "loss": 0.0138, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 5.360541340663988, | |
| "grad_norm": 0.45777344703674316, | |
| "learning_rate": 3.7594871794871794e-06, | |
| "loss": 0.0175, | |
| "step": 12675 | |
| }, | |
| { | |
| "epoch": 5.371114400507507, | |
| "grad_norm": 0.6128593683242798, | |
| "learning_rate": 3.7466666666666667e-06, | |
| "loss": 0.0084, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 5.381687460351025, | |
| "grad_norm": 0.644332230091095, | |
| "learning_rate": 3.733846153846154e-06, | |
| "loss": 0.0111, | |
| "step": 12725 | |
| }, | |
| { | |
| "epoch": 5.392260520194545, | |
| "grad_norm": 0.603568971157074, | |
| "learning_rate": 3.721025641025641e-06, | |
| "loss": 0.0111, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 5.402833580038063, | |
| "grad_norm": 0.3822285532951355, | |
| "learning_rate": 3.7082051282051284e-06, | |
| "loss": 0.0095, | |
| "step": 12775 | |
| }, | |
| { | |
| "epoch": 5.413406639881582, | |
| "grad_norm": 0.493023157119751, | |
| "learning_rate": 3.6953846153846156e-06, | |
| "loss": 0.0157, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 5.4239796997251, | |
| "grad_norm": 1.2487260103225708, | |
| "learning_rate": 3.682564102564103e-06, | |
| "loss": 0.0097, | |
| "step": 12825 | |
| }, | |
| { | |
| "epoch": 5.43455275956862, | |
| "grad_norm": 0.8562780618667603, | |
| "learning_rate": 3.66974358974359e-06, | |
| "loss": 0.0139, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 5.445125819412138, | |
| "grad_norm": 0.750851571559906, | |
| "learning_rate": 3.6569230769230774e-06, | |
| "loss": 0.0123, | |
| "step": 12875 | |
| }, | |
| { | |
| "epoch": 5.455698879255657, | |
| "grad_norm": 0.41030699014663696, | |
| "learning_rate": 3.6441025641025646e-06, | |
| "loss": 0.0116, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 5.466271939099175, | |
| "grad_norm": 0.5965930223464966, | |
| "learning_rate": 3.631282051282052e-06, | |
| "loss": 0.0107, | |
| "step": 12925 | |
| }, | |
| { | |
| "epoch": 5.476844998942694, | |
| "grad_norm": 0.4181739389896393, | |
| "learning_rate": 3.618461538461539e-06, | |
| "loss": 0.0077, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 5.487418058786213, | |
| "grad_norm": 3.0473992824554443, | |
| "learning_rate": 3.6056410256410255e-06, | |
| "loss": 0.0144, | |
| "step": 12975 | |
| }, | |
| { | |
| "epoch": 5.497991118629732, | |
| "grad_norm": 6.770436763763428, | |
| "learning_rate": 3.5928205128205128e-06, | |
| "loss": 0.0121, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 5.497991118629732, | |
| "eval_loss": 0.14054465293884277, | |
| "eval_runtime": 479.854, | |
| "eval_samples_per_second": 7.588, | |
| "eval_steps_per_second": 0.95, | |
| "eval_wer": 0.11290547125156619, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 5.50856417847325, | |
| "grad_norm": 1.682664155960083, | |
| "learning_rate": 3.58e-06, | |
| "loss": 0.0098, | |
| "step": 13025 | |
| }, | |
| { | |
| "epoch": 5.519137238316769, | |
| "grad_norm": 1.461300253868103, | |
| "learning_rate": 3.5671794871794873e-06, | |
| "loss": 0.0099, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 5.529710298160287, | |
| "grad_norm": 0.3437669575214386, | |
| "learning_rate": 3.5543589743589745e-06, | |
| "loss": 0.014, | |
| "step": 13075 | |
| }, | |
| { | |
| "epoch": 5.540283358003807, | |
| "grad_norm": 1.4875051975250244, | |
| "learning_rate": 3.5415384615384618e-06, | |
| "loss": 0.013, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 5.550856417847325, | |
| "grad_norm": 3.039585590362549, | |
| "learning_rate": 3.528717948717949e-06, | |
| "loss": 0.0131, | |
| "step": 13125 | |
| }, | |
| { | |
| "epoch": 5.561429477690844, | |
| "grad_norm": 0.7710385918617249, | |
| "learning_rate": 3.5158974358974363e-06, | |
| "loss": 0.011, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 5.572002537534362, | |
| "grad_norm": 4.138525485992432, | |
| "learning_rate": 3.5030769230769235e-06, | |
| "loss": 0.0098, | |
| "step": 13175 | |
| }, | |
| { | |
| "epoch": 5.582575597377881, | |
| "grad_norm": 4.096315860748291, | |
| "learning_rate": 3.4902564102564108e-06, | |
| "loss": 0.0143, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 5.5931486572214, | |
| "grad_norm": 1.198435664176941, | |
| "learning_rate": 3.477435897435898e-06, | |
| "loss": 0.0127, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 5.603721717064919, | |
| "grad_norm": 0.3887302875518799, | |
| "learning_rate": 3.4646153846153853e-06, | |
| "loss": 0.0138, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 5.614294776908437, | |
| "grad_norm": 1.2115799188613892, | |
| "learning_rate": 3.4517948717948717e-06, | |
| "loss": 0.0172, | |
| "step": 13275 | |
| }, | |
| { | |
| "epoch": 5.624867836751956, | |
| "grad_norm": 0.7135342955589294, | |
| "learning_rate": 3.438974358974359e-06, | |
| "loss": 0.0148, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 5.635440896595474, | |
| "grad_norm": 2.1200053691864014, | |
| "learning_rate": 3.426153846153846e-06, | |
| "loss": 0.0084, | |
| "step": 13325 | |
| }, | |
| { | |
| "epoch": 5.646013956438994, | |
| "grad_norm": 1.8395202159881592, | |
| "learning_rate": 3.4133333333333334e-06, | |
| "loss": 0.01, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 5.656587016282512, | |
| "grad_norm": 4.734637260437012, | |
| "learning_rate": 3.4005128205128207e-06, | |
| "loss": 0.0141, | |
| "step": 13375 | |
| }, | |
| { | |
| "epoch": 5.667160076126031, | |
| "grad_norm": 2.1322085857391357, | |
| "learning_rate": 3.387692307692308e-06, | |
| "loss": 0.0154, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 5.677733135969549, | |
| "grad_norm": 0.8213221430778503, | |
| "learning_rate": 3.374871794871795e-06, | |
| "loss": 0.0065, | |
| "step": 13425 | |
| }, | |
| { | |
| "epoch": 5.688306195813068, | |
| "grad_norm": 0.4765898585319519, | |
| "learning_rate": 3.3620512820512824e-06, | |
| "loss": 0.008, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 5.698879255656587, | |
| "grad_norm": 0.424668550491333, | |
| "learning_rate": 3.3492307692307696e-06, | |
| "loss": 0.0104, | |
| "step": 13475 | |
| }, | |
| { | |
| "epoch": 5.709452315500106, | |
| "grad_norm": 0.7294663786888123, | |
| "learning_rate": 3.336410256410257e-06, | |
| "loss": 0.0114, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 5.720025375343624, | |
| "grad_norm": 0.4553682208061218, | |
| "learning_rate": 3.323589743589744e-06, | |
| "loss": 0.0133, | |
| "step": 13525 | |
| }, | |
| { | |
| "epoch": 5.730598435187143, | |
| "grad_norm": 4.255309104919434, | |
| "learning_rate": 3.3107692307692314e-06, | |
| "loss": 0.0269, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 5.741171495030661, | |
| "grad_norm": 0.8475791811943054, | |
| "learning_rate": 3.297948717948718e-06, | |
| "loss": 0.0086, | |
| "step": 13575 | |
| }, | |
| { | |
| "epoch": 5.751744554874181, | |
| "grad_norm": 4.737633228302002, | |
| "learning_rate": 3.285128205128205e-06, | |
| "loss": 0.0193, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 5.762317614717699, | |
| "grad_norm": 1.0523945093154907, | |
| "learning_rate": 3.2723076923076923e-06, | |
| "loss": 0.0124, | |
| "step": 13625 | |
| }, | |
| { | |
| "epoch": 5.772890674561218, | |
| "grad_norm": 0.8394791483879089, | |
| "learning_rate": 3.2594871794871795e-06, | |
| "loss": 0.0109, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 5.783463734404736, | |
| "grad_norm": 2.474153518676758, | |
| "learning_rate": 3.2466666666666668e-06, | |
| "loss": 0.0123, | |
| "step": 13675 | |
| }, | |
| { | |
| "epoch": 5.794036794248256, | |
| "grad_norm": 0.8185378909111023, | |
| "learning_rate": 3.233846153846154e-06, | |
| "loss": 0.0084, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 5.804609854091774, | |
| "grad_norm": 4.937212944030762, | |
| "learning_rate": 3.2210256410256413e-06, | |
| "loss": 0.0134, | |
| "step": 13725 | |
| }, | |
| { | |
| "epoch": 5.815182913935293, | |
| "grad_norm": 0.6099960207939148, | |
| "learning_rate": 3.2082051282051285e-06, | |
| "loss": 0.0091, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 5.825755973778811, | |
| "grad_norm": 0.43338268995285034, | |
| "learning_rate": 3.1953846153846158e-06, | |
| "loss": 0.0118, | |
| "step": 13775 | |
| }, | |
| { | |
| "epoch": 5.836329033622331, | |
| "grad_norm": 0.47640514373779297, | |
| "learning_rate": 3.182564102564103e-06, | |
| "loss": 0.0072, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 5.846902093465849, | |
| "grad_norm": 2.1803908348083496, | |
| "learning_rate": 3.1697435897435903e-06, | |
| "loss": 0.0119, | |
| "step": 13825 | |
| }, | |
| { | |
| "epoch": 5.857475153309368, | |
| "grad_norm": 1.2694458961486816, | |
| "learning_rate": 3.1569230769230775e-06, | |
| "loss": 0.0102, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 5.868048213152886, | |
| "grad_norm": 0.539225697517395, | |
| "learning_rate": 3.144102564102564e-06, | |
| "loss": 0.0107, | |
| "step": 13875 | |
| }, | |
| { | |
| "epoch": 5.878621272996405, | |
| "grad_norm": 2.2293202877044678, | |
| "learning_rate": 3.131282051282051e-06, | |
| "loss": 0.0093, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 5.889194332839924, | |
| "grad_norm": 0.3595990538597107, | |
| "learning_rate": 3.1184615384615384e-06, | |
| "loss": 0.0075, | |
| "step": 13925 | |
| }, | |
| { | |
| "epoch": 5.899767392683443, | |
| "grad_norm": 1.155290961265564, | |
| "learning_rate": 3.1056410256410257e-06, | |
| "loss": 0.011, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 5.910340452526961, | |
| "grad_norm": 6.233605861663818, | |
| "learning_rate": 3.092820512820513e-06, | |
| "loss": 0.0071, | |
| "step": 13975 | |
| }, | |
| { | |
| "epoch": 5.92091351237048, | |
| "grad_norm": 0.6025995016098022, | |
| "learning_rate": 3.08e-06, | |
| "loss": 0.0074, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 5.92091351237048, | |
| "eval_loss": 0.13853086531162262, | |
| "eval_runtime": 478.5411, | |
| "eval_samples_per_second": 7.609, | |
| "eval_steps_per_second": 0.953, | |
| "eval_wer": 0.11954151004686993, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 5.931486572213998, | |
| "grad_norm": 0.934971034526825, | |
| "learning_rate": 3.0671794871794874e-06, | |
| "loss": 0.0149, | |
| "step": 14025 | |
| }, | |
| { | |
| "epoch": 5.942059632057518, | |
| "grad_norm": 2.112450122833252, | |
| "learning_rate": 3.0543589743589747e-06, | |
| "loss": 0.0254, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 5.952632691901036, | |
| "grad_norm": 1.504460334777832, | |
| "learning_rate": 3.041538461538462e-06, | |
| "loss": 0.0079, | |
| "step": 14075 | |
| }, | |
| { | |
| "epoch": 5.963205751744555, | |
| "grad_norm": 3.3018078804016113, | |
| "learning_rate": 3.028717948717949e-06, | |
| "loss": 0.0084, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 5.9737788115880734, | |
| "grad_norm": 0.2936100363731384, | |
| "learning_rate": 3.0158974358974364e-06, | |
| "loss": 0.0077, | |
| "step": 14125 | |
| }, | |
| { | |
| "epoch": 5.984351871431592, | |
| "grad_norm": 2.9025211334228516, | |
| "learning_rate": 3.0030769230769236e-06, | |
| "loss": 0.0107, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 5.994924931275111, | |
| "grad_norm": 0.5673441886901855, | |
| "learning_rate": 2.99025641025641e-06, | |
| "loss": 0.0101, | |
| "step": 14175 | |
| }, | |
| { | |
| "epoch": 6.00549799111863, | |
| "grad_norm": 0.29159656167030334, | |
| "learning_rate": 2.9774358974358973e-06, | |
| "loss": 0.009, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 6.0160710509621484, | |
| "grad_norm": 0.2572082579135895, | |
| "learning_rate": 2.9646153846153845e-06, | |
| "loss": 0.007, | |
| "step": 14225 | |
| }, | |
| { | |
| "epoch": 6.026644110805667, | |
| "grad_norm": 0.2045765519142151, | |
| "learning_rate": 2.951794871794872e-06, | |
| "loss": 0.0058, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 6.0372171706491855, | |
| "grad_norm": 0.34850168228149414, | |
| "learning_rate": 2.938974358974359e-06, | |
| "loss": 0.0067, | |
| "step": 14275 | |
| }, | |
| { | |
| "epoch": 6.047790230492705, | |
| "grad_norm": 5.987231254577637, | |
| "learning_rate": 2.9261538461538463e-06, | |
| "loss": 0.0052, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 6.0583632903362235, | |
| "grad_norm": 0.34981194138526917, | |
| "learning_rate": 2.9133333333333335e-06, | |
| "loss": 0.0059, | |
| "step": 14325 | |
| }, | |
| { | |
| "epoch": 6.068936350179742, | |
| "grad_norm": 0.2722649574279785, | |
| "learning_rate": 2.9005128205128208e-06, | |
| "loss": 0.0105, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 6.0795094100232605, | |
| "grad_norm": 0.33031409978866577, | |
| "learning_rate": 2.887692307692308e-06, | |
| "loss": 0.0144, | |
| "step": 14375 | |
| }, | |
| { | |
| "epoch": 6.090082469866779, | |
| "grad_norm": 0.26611846685409546, | |
| "learning_rate": 2.8748717948717953e-06, | |
| "loss": 0.0053, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 6.1006555297102985, | |
| "grad_norm": 0.29690757393836975, | |
| "learning_rate": 2.8620512820512825e-06, | |
| "loss": 0.0042, | |
| "step": 14425 | |
| }, | |
| { | |
| "epoch": 6.111228589553817, | |
| "grad_norm": 0.2454768419265747, | |
| "learning_rate": 2.8492307692307698e-06, | |
| "loss": 0.0065, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 6.1218016493973355, | |
| "grad_norm": 0.3441830575466156, | |
| "learning_rate": 2.836410256410257e-06, | |
| "loss": 0.0088, | |
| "step": 14475 | |
| }, | |
| { | |
| "epoch": 6.132374709240854, | |
| "grad_norm": 0.4179341793060303, | |
| "learning_rate": 2.8235897435897434e-06, | |
| "loss": 0.0137, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 6.142947769084373, | |
| "grad_norm": 0.3850744664669037, | |
| "learning_rate": 2.8107692307692307e-06, | |
| "loss": 0.0033, | |
| "step": 14525 | |
| }, | |
| { | |
| "epoch": 6.153520828927892, | |
| "grad_norm": 0.2846491038799286, | |
| "learning_rate": 2.797948717948718e-06, | |
| "loss": 0.0061, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 6.1640938887714105, | |
| "grad_norm": 0.3616473376750946, | |
| "learning_rate": 2.785128205128205e-06, | |
| "loss": 0.0052, | |
| "step": 14575 | |
| }, | |
| { | |
| "epoch": 6.174666948614929, | |
| "grad_norm": 0.38456159830093384, | |
| "learning_rate": 2.7723076923076924e-06, | |
| "loss": 0.0046, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 6.185240008458448, | |
| "grad_norm": 0.267269492149353, | |
| "learning_rate": 2.7594871794871797e-06, | |
| "loss": 0.0069, | |
| "step": 14625 | |
| }, | |
| { | |
| "epoch": 6.195813068301967, | |
| "grad_norm": 0.9138497114181519, | |
| "learning_rate": 2.746666666666667e-06, | |
| "loss": 0.0068, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 6.2063861281454855, | |
| "grad_norm": 0.2973681390285492, | |
| "learning_rate": 2.733846153846154e-06, | |
| "loss": 0.0113, | |
| "step": 14675 | |
| }, | |
| { | |
| "epoch": 6.216959187989004, | |
| "grad_norm": 0.3064761757850647, | |
| "learning_rate": 2.7210256410256414e-06, | |
| "loss": 0.0076, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 6.227532247832523, | |
| "grad_norm": 0.2878686785697937, | |
| "learning_rate": 2.7082051282051287e-06, | |
| "loss": 0.0044, | |
| "step": 14725 | |
| }, | |
| { | |
| "epoch": 6.238105307676041, | |
| "grad_norm": 0.2550056576728821, | |
| "learning_rate": 2.695384615384616e-06, | |
| "loss": 0.0064, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 6.2486783675195605, | |
| "grad_norm": 0.6837669610977173, | |
| "learning_rate": 2.682564102564103e-06, | |
| "loss": 0.0046, | |
| "step": 14775 | |
| }, | |
| { | |
| "epoch": 6.259251427363079, | |
| "grad_norm": 0.244142547249794, | |
| "learning_rate": 2.6697435897435896e-06, | |
| "loss": 0.0033, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 6.269824487206598, | |
| "grad_norm": 0.15427115559577942, | |
| "learning_rate": 2.656923076923077e-06, | |
| "loss": 0.0055, | |
| "step": 14825 | |
| }, | |
| { | |
| "epoch": 6.280397547050116, | |
| "grad_norm": 0.7226958274841309, | |
| "learning_rate": 2.644102564102564e-06, | |
| "loss": 0.0053, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 6.290970606893635, | |
| "grad_norm": 0.2157323956489563, | |
| "learning_rate": 2.6312820512820513e-06, | |
| "loss": 0.007, | |
| "step": 14875 | |
| }, | |
| { | |
| "epoch": 6.301543666737154, | |
| "grad_norm": 2.2182655334472656, | |
| "learning_rate": 2.6184615384615385e-06, | |
| "loss": 0.0079, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 6.312116726580673, | |
| "grad_norm": 0.17396897077560425, | |
| "learning_rate": 2.605641025641026e-06, | |
| "loss": 0.0059, | |
| "step": 14925 | |
| }, | |
| { | |
| "epoch": 6.322689786424191, | |
| "grad_norm": 0.8146637082099915, | |
| "learning_rate": 2.592820512820513e-06, | |
| "loss": 0.0048, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 6.33326284626771, | |
| "grad_norm": 0.2695443332195282, | |
| "learning_rate": 2.5800000000000003e-06, | |
| "loss": 0.0049, | |
| "step": 14975 | |
| }, | |
| { | |
| "epoch": 6.343835906111229, | |
| "grad_norm": 0.7036349177360535, | |
| "learning_rate": 2.5671794871794875e-06, | |
| "loss": 0.0064, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 6.343835906111229, | |
| "eval_loss": 0.1409808248281479, | |
| "eval_runtime": 474.6061, | |
| "eval_samples_per_second": 7.672, | |
| "eval_steps_per_second": 0.961, | |
| "eval_wer": 0.11146688941482204, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 6.354408965954748, | |
| "grad_norm": 0.4002957344055176, | |
| "learning_rate": 2.5543589743589748e-06, | |
| "loss": 0.0055, | |
| "step": 15025 | |
| }, | |
| { | |
| "epoch": 6.364982025798266, | |
| "grad_norm": 2.1062963008880615, | |
| "learning_rate": 2.541538461538462e-06, | |
| "loss": 0.0038, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 6.375555085641785, | |
| "grad_norm": 0.4128209948539734, | |
| "learning_rate": 2.5287179487179493e-06, | |
| "loss": 0.0062, | |
| "step": 15075 | |
| }, | |
| { | |
| "epoch": 6.386128145485303, | |
| "grad_norm": 0.24224917590618134, | |
| "learning_rate": 2.5158974358974357e-06, | |
| "loss": 0.0076, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 6.396701205328823, | |
| "grad_norm": 0.31282123923301697, | |
| "learning_rate": 2.503076923076923e-06, | |
| "loss": 0.0066, | |
| "step": 15125 | |
| }, | |
| { | |
| "epoch": 6.407274265172341, | |
| "grad_norm": 0.28958702087402344, | |
| "learning_rate": 2.4902564102564106e-06, | |
| "loss": 0.0054, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 6.41784732501586, | |
| "grad_norm": 0.23962663114070892, | |
| "learning_rate": 2.4774358974358974e-06, | |
| "loss": 0.0043, | |
| "step": 15175 | |
| }, | |
| { | |
| "epoch": 6.428420384859378, | |
| "grad_norm": 0.22258694469928741, | |
| "learning_rate": 2.4646153846153847e-06, | |
| "loss": 0.0098, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 6.438993444702897, | |
| "grad_norm": 0.219313845038414, | |
| "learning_rate": 2.451794871794872e-06, | |
| "loss": 0.0064, | |
| "step": 15225 | |
| }, | |
| { | |
| "epoch": 6.449566504546416, | |
| "grad_norm": 0.24351318180561066, | |
| "learning_rate": 2.438974358974359e-06, | |
| "loss": 0.0053, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 6.460139564389935, | |
| "grad_norm": 1.7564576864242554, | |
| "learning_rate": 2.4261538461538464e-06, | |
| "loss": 0.0106, | |
| "step": 15275 | |
| }, | |
| { | |
| "epoch": 6.470712624233453, | |
| "grad_norm": 1.4861177206039429, | |
| "learning_rate": 2.4133333333333337e-06, | |
| "loss": 0.0063, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 6.481285684076972, | |
| "grad_norm": 0.41736364364624023, | |
| "learning_rate": 2.4005128205128205e-06, | |
| "loss": 0.0052, | |
| "step": 15325 | |
| }, | |
| { | |
| "epoch": 6.49185874392049, | |
| "grad_norm": 0.21913078427314758, | |
| "learning_rate": 2.3876923076923077e-06, | |
| "loss": 0.0063, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 6.50243180376401, | |
| "grad_norm": 0.18712382018566132, | |
| "learning_rate": 2.374871794871795e-06, | |
| "loss": 0.01, | |
| "step": 15375 | |
| }, | |
| { | |
| "epoch": 6.513004863607528, | |
| "grad_norm": 0.7857060432434082, | |
| "learning_rate": 2.3620512820512822e-06, | |
| "loss": 0.0081, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 6.523577923451047, | |
| "grad_norm": 0.23228535056114197, | |
| "learning_rate": 2.3492307692307695e-06, | |
| "loss": 0.0055, | |
| "step": 15425 | |
| }, | |
| { | |
| "epoch": 6.534150983294565, | |
| "grad_norm": 0.2794516682624817, | |
| "learning_rate": 2.3364102564102567e-06, | |
| "loss": 0.0118, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 6.544724043138084, | |
| "grad_norm": 0.3594117760658264, | |
| "learning_rate": 2.3235897435897436e-06, | |
| "loss": 0.0049, | |
| "step": 15475 | |
| }, | |
| { | |
| "epoch": 6.555297102981603, | |
| "grad_norm": 0.38121312856674194, | |
| "learning_rate": 2.310769230769231e-06, | |
| "loss": 0.0098, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 6.565870162825122, | |
| "grad_norm": 0.48879972100257874, | |
| "learning_rate": 2.2984615384615386e-06, | |
| "loss": 0.0055, | |
| "step": 15525 | |
| }, | |
| { | |
| "epoch": 6.57644322266864, | |
| "grad_norm": 0.3019927442073822, | |
| "learning_rate": 2.285641025641026e-06, | |
| "loss": 0.0066, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 6.587016282512159, | |
| "grad_norm": 2.143726348876953, | |
| "learning_rate": 2.272820512820513e-06, | |
| "loss": 0.0078, | |
| "step": 15575 | |
| }, | |
| { | |
| "epoch": 6.597589342355677, | |
| "grad_norm": 0.1708030104637146, | |
| "learning_rate": 2.2600000000000004e-06, | |
| "loss": 0.004, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 6.608162402199197, | |
| "grad_norm": 1.7582453489303589, | |
| "learning_rate": 2.2471794871794876e-06, | |
| "loss": 0.0092, | |
| "step": 15625 | |
| }, | |
| { | |
| "epoch": 6.618735462042715, | |
| "grad_norm": 0.3129083514213562, | |
| "learning_rate": 2.2343589743589745e-06, | |
| "loss": 0.0073, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 6.629308521886234, | |
| "grad_norm": 0.3408079743385315, | |
| "learning_rate": 2.2215384615384617e-06, | |
| "loss": 0.0069, | |
| "step": 15675 | |
| }, | |
| { | |
| "epoch": 6.639881581729752, | |
| "grad_norm": 11.103797912597656, | |
| "learning_rate": 2.208717948717949e-06, | |
| "loss": 0.0105, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 6.650454641573271, | |
| "grad_norm": 0.554865300655365, | |
| "learning_rate": 2.195897435897436e-06, | |
| "loss": 0.0082, | |
| "step": 15725 | |
| }, | |
| { | |
| "epoch": 6.66102770141679, | |
| "grad_norm": 0.32903775572776794, | |
| "learning_rate": 2.1830769230769234e-06, | |
| "loss": 0.0033, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 6.671600761260309, | |
| "grad_norm": 0.2150058150291443, | |
| "learning_rate": 2.1702564102564107e-06, | |
| "loss": 0.0119, | |
| "step": 15775 | |
| }, | |
| { | |
| "epoch": 6.682173821103827, | |
| "grad_norm": 0.253409743309021, | |
| "learning_rate": 2.1574358974358975e-06, | |
| "loss": 0.0046, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 6.692746880947346, | |
| "grad_norm": 2.6021835803985596, | |
| "learning_rate": 2.1446153846153848e-06, | |
| "loss": 0.0077, | |
| "step": 15825 | |
| }, | |
| { | |
| "epoch": 6.703319940790865, | |
| "grad_norm": 1.3991787433624268, | |
| "learning_rate": 2.131794871794872e-06, | |
| "loss": 0.0072, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 6.713893000634384, | |
| "grad_norm": 0.40495139360427856, | |
| "learning_rate": 2.1189743589743593e-06, | |
| "loss": 0.013, | |
| "step": 15875 | |
| }, | |
| { | |
| "epoch": 6.724466060477902, | |
| "grad_norm": 0.28818759322166443, | |
| "learning_rate": 2.1061538461538465e-06, | |
| "loss": 0.0065, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 6.735039120321421, | |
| "grad_norm": 0.6223933696746826, | |
| "learning_rate": 2.0933333333333338e-06, | |
| "loss": 0.0057, | |
| "step": 15925 | |
| }, | |
| { | |
| "epoch": 6.745612180164939, | |
| "grad_norm": 0.24611905217170715, | |
| "learning_rate": 2.0805128205128206e-06, | |
| "loss": 0.0049, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 6.756185240008459, | |
| "grad_norm": 0.6771820783615112, | |
| "learning_rate": 2.067692307692308e-06, | |
| "loss": 0.0048, | |
| "step": 15975 | |
| }, | |
| { | |
| "epoch": 6.766758299851977, | |
| "grad_norm": 14.729320526123047, | |
| "learning_rate": 2.054871794871795e-06, | |
| "loss": 0.0066, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 6.766758299851977, | |
| "eval_loss": 0.14149095118045807, | |
| "eval_runtime": 479.8155, | |
| "eval_samples_per_second": 7.588, | |
| "eval_steps_per_second": 0.95, | |
| "eval_wer": 0.11842776927003573, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 6.777331359695496, | |
| "grad_norm": 0.3452744781970978, | |
| "learning_rate": 2.0420512820512823e-06, | |
| "loss": 0.0052, | |
| "step": 16025 | |
| }, | |
| { | |
| "epoch": 6.787904419539014, | |
| "grad_norm": 0.376597136259079, | |
| "learning_rate": 2.0292307692307696e-06, | |
| "loss": 0.0044, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 6.798477479382534, | |
| "grad_norm": 0.2520189583301544, | |
| "learning_rate": 2.016410256410257e-06, | |
| "loss": 0.0088, | |
| "step": 16075 | |
| }, | |
| { | |
| "epoch": 6.809050539226052, | |
| "grad_norm": 0.19255167245864868, | |
| "learning_rate": 2.0035897435897436e-06, | |
| "loss": 0.0046, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 6.819623599069571, | |
| "grad_norm": 0.38311922550201416, | |
| "learning_rate": 1.990769230769231e-06, | |
| "loss": 0.0028, | |
| "step": 16125 | |
| }, | |
| { | |
| "epoch": 6.830196658913089, | |
| "grad_norm": 0.2503233850002289, | |
| "learning_rate": 1.977948717948718e-06, | |
| "loss": 0.0038, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 6.840769718756608, | |
| "grad_norm": 0.8586848378181458, | |
| "learning_rate": 1.9651282051282054e-06, | |
| "loss": 0.0052, | |
| "step": 16175 | |
| }, | |
| { | |
| "epoch": 6.851342778600127, | |
| "grad_norm": 0.256672203540802, | |
| "learning_rate": 1.9523076923076926e-06, | |
| "loss": 0.0031, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 6.861915838443646, | |
| "grad_norm": 0.3382306396961212, | |
| "learning_rate": 1.93948717948718e-06, | |
| "loss": 0.0072, | |
| "step": 16225 | |
| }, | |
| { | |
| "epoch": 6.872488898287164, | |
| "grad_norm": 1.6218035221099854, | |
| "learning_rate": 1.926666666666667e-06, | |
| "loss": 0.0071, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 6.883061958130683, | |
| "grad_norm": 0.33794450759887695, | |
| "learning_rate": 1.913846153846154e-06, | |
| "loss": 0.0062, | |
| "step": 16275 | |
| }, | |
| { | |
| "epoch": 6.893635017974201, | |
| "grad_norm": 0.20873773097991943, | |
| "learning_rate": 1.9010256410256412e-06, | |
| "loss": 0.0047, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 6.904208077817721, | |
| "grad_norm": 0.2936864495277405, | |
| "learning_rate": 1.8882051282051285e-06, | |
| "loss": 0.0034, | |
| "step": 16325 | |
| }, | |
| { | |
| "epoch": 6.914781137661239, | |
| "grad_norm": 0.5449936389923096, | |
| "learning_rate": 1.8753846153846155e-06, | |
| "loss": 0.0065, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 6.925354197504758, | |
| "grad_norm": 0.29695820808410645, | |
| "learning_rate": 1.8625641025641027e-06, | |
| "loss": 0.0125, | |
| "step": 16375 | |
| }, | |
| { | |
| "epoch": 6.935927257348276, | |
| "grad_norm": 0.4382512867450714, | |
| "learning_rate": 1.84974358974359e-06, | |
| "loss": 0.0058, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 6.946500317191795, | |
| "grad_norm": 0.5371158719062805, | |
| "learning_rate": 1.836923076923077e-06, | |
| "loss": 0.0036, | |
| "step": 16425 | |
| }, | |
| { | |
| "epoch": 6.957073377035314, | |
| "grad_norm": 0.21529348194599152, | |
| "learning_rate": 1.8241025641025643e-06, | |
| "loss": 0.0042, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 6.967646436878833, | |
| "grad_norm": 0.25407838821411133, | |
| "learning_rate": 1.8112820512820515e-06, | |
| "loss": 0.0048, | |
| "step": 16475 | |
| }, | |
| { | |
| "epoch": 6.978219496722351, | |
| "grad_norm": 0.41775551438331604, | |
| "learning_rate": 1.7984615384615386e-06, | |
| "loss": 0.0042, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 6.98879255656587, | |
| "grad_norm": 0.13575445115566254, | |
| "learning_rate": 1.7856410256410258e-06, | |
| "loss": 0.0057, | |
| "step": 16525 | |
| }, | |
| { | |
| "epoch": 6.9993656164093885, | |
| "grad_norm": 0.21849684417247772, | |
| "learning_rate": 1.772820512820513e-06, | |
| "loss": 0.0065, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 7.009938676252908, | |
| "grad_norm": 0.17400676012039185, | |
| "learning_rate": 1.76e-06, | |
| "loss": 0.0069, | |
| "step": 16575 | |
| }, | |
| { | |
| "epoch": 7.020511736096426, | |
| "grad_norm": 0.2058832198381424, | |
| "learning_rate": 1.7471794871794873e-06, | |
| "loss": 0.005, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 7.031084795939945, | |
| "grad_norm": 0.14491191506385803, | |
| "learning_rate": 1.7343589743589746e-06, | |
| "loss": 0.0026, | |
| "step": 16625 | |
| }, | |
| { | |
| "epoch": 7.0416578557834635, | |
| "grad_norm": 2.8785014152526855, | |
| "learning_rate": 1.7215384615384616e-06, | |
| "loss": 0.0053, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 7.052230915626982, | |
| "grad_norm": 0.17464618384838104, | |
| "learning_rate": 1.7087179487179489e-06, | |
| "loss": 0.0024, | |
| "step": 16675 | |
| }, | |
| { | |
| "epoch": 7.062803975470501, | |
| "grad_norm": 0.17902244627475739, | |
| "learning_rate": 1.6958974358974361e-06, | |
| "loss": 0.004, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 7.07337703531402, | |
| "grad_norm": 1.190285563468933, | |
| "learning_rate": 1.6830769230769232e-06, | |
| "loss": 0.0023, | |
| "step": 16725 | |
| }, | |
| { | |
| "epoch": 7.0839500951575385, | |
| "grad_norm": 0.24982137978076935, | |
| "learning_rate": 1.6702564102564104e-06, | |
| "loss": 0.0029, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 7.094523155001057, | |
| "grad_norm": 0.21240869164466858, | |
| "learning_rate": 1.6574358974358976e-06, | |
| "loss": 0.0034, | |
| "step": 16775 | |
| }, | |
| { | |
| "epoch": 7.105096214844576, | |
| "grad_norm": 0.22856955230236053, | |
| "learning_rate": 1.6446153846153847e-06, | |
| "loss": 0.0028, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 7.115669274688095, | |
| "grad_norm": 0.2120029181241989, | |
| "learning_rate": 1.631794871794872e-06, | |
| "loss": 0.0026, | |
| "step": 16825 | |
| }, | |
| { | |
| "epoch": 7.1262423345316135, | |
| "grad_norm": 0.10486655682325363, | |
| "learning_rate": 1.6189743589743592e-06, | |
| "loss": 0.0063, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 7.136815394375132, | |
| "grad_norm": 0.1727411448955536, | |
| "learning_rate": 1.6061538461538462e-06, | |
| "loss": 0.0058, | |
| "step": 16875 | |
| }, | |
| { | |
| "epoch": 7.1473884542186505, | |
| "grad_norm": 0.2189004272222519, | |
| "learning_rate": 1.5933333333333335e-06, | |
| "loss": 0.0043, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 7.15796151406217, | |
| "grad_norm": 0.224544957280159, | |
| "learning_rate": 1.5805128205128207e-06, | |
| "loss": 0.005, | |
| "step": 16925 | |
| }, | |
| { | |
| "epoch": 7.1685345739056885, | |
| "grad_norm": 0.17426727712154388, | |
| "learning_rate": 1.5676923076923078e-06, | |
| "loss": 0.0039, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 7.179107633749207, | |
| "grad_norm": 0.1636071801185608, | |
| "learning_rate": 1.554871794871795e-06, | |
| "loss": 0.0027, | |
| "step": 16975 | |
| }, | |
| { | |
| "epoch": 7.1896806935927255, | |
| "grad_norm": 0.13778026401996613, | |
| "learning_rate": 1.5420512820512822e-06, | |
| "loss": 0.0029, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 7.1896806935927255, | |
| "eval_loss": 0.14258068799972534, | |
| "eval_runtime": 481.0853, | |
| "eval_samples_per_second": 7.568, | |
| "eval_steps_per_second": 0.948, | |
| "eval_wer": 0.11903104552415425, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 7.200253753436244, | |
| "grad_norm": 0.1793927550315857, | |
| "learning_rate": 1.5292307692307693e-06, | |
| "loss": 0.0054, | |
| "step": 17025 | |
| }, | |
| { | |
| "epoch": 7.2108268132797635, | |
| "grad_norm": 0.17130379378795624, | |
| "learning_rate": 1.5164102564102565e-06, | |
| "loss": 0.0036, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 7.221399873123282, | |
| "grad_norm": 0.20788300037384033, | |
| "learning_rate": 1.5035897435897438e-06, | |
| "loss": 0.0023, | |
| "step": 17075 | |
| }, | |
| { | |
| "epoch": 7.2319729329668005, | |
| "grad_norm": 0.28084486722946167, | |
| "learning_rate": 1.4907692307692308e-06, | |
| "loss": 0.0027, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 7.242545992810319, | |
| "grad_norm": 0.1289544254541397, | |
| "learning_rate": 1.477948717948718e-06, | |
| "loss": 0.0074, | |
| "step": 17125 | |
| }, | |
| { | |
| "epoch": 7.253119052653838, | |
| "grad_norm": 0.14958246052265167, | |
| "learning_rate": 1.4651282051282053e-06, | |
| "loss": 0.0063, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 7.263692112497357, | |
| "grad_norm": 0.7993505001068115, | |
| "learning_rate": 1.4523076923076923e-06, | |
| "loss": 0.0069, | |
| "step": 17175 | |
| }, | |
| { | |
| "epoch": 7.2742651723408756, | |
| "grad_norm": 0.30847251415252686, | |
| "learning_rate": 1.4394871794871796e-06, | |
| "loss": 0.0072, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 7.284838232184394, | |
| "grad_norm": 0.17055633664131165, | |
| "learning_rate": 1.4266666666666668e-06, | |
| "loss": 0.0024, | |
| "step": 17225 | |
| }, | |
| { | |
| "epoch": 7.295411292027913, | |
| "grad_norm": 0.13648909330368042, | |
| "learning_rate": 1.4138461538461539e-06, | |
| "loss": 0.0022, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 7.305984351871432, | |
| "grad_norm": 0.17881402373313904, | |
| "learning_rate": 1.4010256410256411e-06, | |
| "loss": 0.0033, | |
| "step": 17275 | |
| }, | |
| { | |
| "epoch": 7.3165574117149506, | |
| "grad_norm": 0.11867067217826843, | |
| "learning_rate": 1.3882051282051284e-06, | |
| "loss": 0.0034, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 7.327130471558469, | |
| "grad_norm": 0.1998029202222824, | |
| "learning_rate": 1.3753846153846154e-06, | |
| "loss": 0.0026, | |
| "step": 17325 | |
| }, | |
| { | |
| "epoch": 7.337703531401988, | |
| "grad_norm": 0.19812311232089996, | |
| "learning_rate": 1.3625641025641027e-06, | |
| "loss": 0.0022, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 7.348276591245506, | |
| "grad_norm": 0.20121921598911285, | |
| "learning_rate": 1.34974358974359e-06, | |
| "loss": 0.0025, | |
| "step": 17375 | |
| }, | |
| { | |
| "epoch": 7.358849651089026, | |
| "grad_norm": 0.11519061774015427, | |
| "learning_rate": 1.336923076923077e-06, | |
| "loss": 0.0038, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 7.369422710932544, | |
| "grad_norm": 0.2508073151111603, | |
| "learning_rate": 1.3241025641025642e-06, | |
| "loss": 0.0033, | |
| "step": 17425 | |
| }, | |
| { | |
| "epoch": 7.379995770776063, | |
| "grad_norm": 0.1964157670736313, | |
| "learning_rate": 1.3112820512820514e-06, | |
| "loss": 0.003, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 7.390568830619581, | |
| "grad_norm": 0.1781347244977951, | |
| "learning_rate": 1.2984615384615385e-06, | |
| "loss": 0.0111, | |
| "step": 17475 | |
| }, | |
| { | |
| "epoch": 7.4011418904631, | |
| "grad_norm": 2.567892551422119, | |
| "learning_rate": 1.2856410256410257e-06, | |
| "loss": 0.0045, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 7.411714950306619, | |
| "grad_norm": 4.157364368438721, | |
| "learning_rate": 1.272820512820513e-06, | |
| "loss": 0.0084, | |
| "step": 17525 | |
| }, | |
| { | |
| "epoch": 7.422288010150138, | |
| "grad_norm": 0.1634252667427063, | |
| "learning_rate": 1.26e-06, | |
| "loss": 0.0045, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 7.432861069993656, | |
| "grad_norm": 0.5479081273078918, | |
| "learning_rate": 1.2471794871794873e-06, | |
| "loss": 0.0075, | |
| "step": 17575 | |
| }, | |
| { | |
| "epoch": 7.443434129837175, | |
| "grad_norm": 0.1536071002483368, | |
| "learning_rate": 1.2343589743589745e-06, | |
| "loss": 0.0038, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 7.454007189680693, | |
| "grad_norm": 2.710284948348999, | |
| "learning_rate": 1.2215384615384618e-06, | |
| "loss": 0.0055, | |
| "step": 17625 | |
| }, | |
| { | |
| "epoch": 7.464580249524213, | |
| "grad_norm": 0.19599756598472595, | |
| "learning_rate": 1.2087179487179488e-06, | |
| "loss": 0.0043, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 7.475153309367731, | |
| "grad_norm": 0.1853959560394287, | |
| "learning_rate": 1.195897435897436e-06, | |
| "loss": 0.0038, | |
| "step": 17675 | |
| }, | |
| { | |
| "epoch": 7.48572636921125, | |
| "grad_norm": 0.223322793841362, | |
| "learning_rate": 1.1830769230769233e-06, | |
| "loss": 0.0073, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 7.496299429054768, | |
| "grad_norm": 0.33646872639656067, | |
| "learning_rate": 1.1702564102564103e-06, | |
| "loss": 0.0044, | |
| "step": 17725 | |
| }, | |
| { | |
| "epoch": 7.506872488898287, | |
| "grad_norm": 0.24336710572242737, | |
| "learning_rate": 1.1574358974358976e-06, | |
| "loss": 0.0034, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 7.517445548741806, | |
| "grad_norm": 0.16662658751010895, | |
| "learning_rate": 1.1446153846153848e-06, | |
| "loss": 0.0041, | |
| "step": 17775 | |
| }, | |
| { | |
| "epoch": 7.528018608585325, | |
| "grad_norm": 0.1463777869939804, | |
| "learning_rate": 1.1317948717948719e-06, | |
| "loss": 0.0128, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 7.538591668428843, | |
| "grad_norm": 0.26682642102241516, | |
| "learning_rate": 1.118974358974359e-06, | |
| "loss": 0.0035, | |
| "step": 17825 | |
| }, | |
| { | |
| "epoch": 7.549164728272362, | |
| "grad_norm": 0.2606651782989502, | |
| "learning_rate": 1.1061538461538463e-06, | |
| "loss": 0.0038, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 7.55973778811588, | |
| "grad_norm": 0.24456697702407837, | |
| "learning_rate": 1.0933333333333334e-06, | |
| "loss": 0.0069, | |
| "step": 17875 | |
| }, | |
| { | |
| "epoch": 7.5703108479594, | |
| "grad_norm": 0.9612045884132385, | |
| "learning_rate": 1.0805128205128206e-06, | |
| "loss": 0.0047, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 7.580883907802918, | |
| "grad_norm": 0.20409923791885376, | |
| "learning_rate": 1.0676923076923079e-06, | |
| "loss": 0.0019, | |
| "step": 17925 | |
| }, | |
| { | |
| "epoch": 7.591456967646437, | |
| "grad_norm": 0.14264066517353058, | |
| "learning_rate": 1.054871794871795e-06, | |
| "loss": 0.0033, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 7.602030027489955, | |
| "grad_norm": 0.43361735343933105, | |
| "learning_rate": 1.0420512820512822e-06, | |
| "loss": 0.0022, | |
| "step": 17975 | |
| }, | |
| { | |
| "epoch": 7.612603087333475, | |
| "grad_norm": 0.25520941615104675, | |
| "learning_rate": 1.0292307692307694e-06, | |
| "loss": 0.0024, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 7.612603087333475, | |
| "eval_loss": 0.14294278621673584, | |
| "eval_runtime": 482.1709, | |
| "eval_samples_per_second": 7.551, | |
| "eval_steps_per_second": 0.946, | |
| "eval_wer": 0.11782449301591721, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 7.623176147176993, | |
| "grad_norm": 0.2945314943790436, | |
| "learning_rate": 1.0164102564102564e-06, | |
| "loss": 0.0025, | |
| "step": 18025 | |
| }, | |
| { | |
| "epoch": 7.633749207020512, | |
| "grad_norm": 0.1885748952627182, | |
| "learning_rate": 1.0035897435897437e-06, | |
| "loss": 0.0031, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 7.64432226686403, | |
| "grad_norm": 0.16705763339996338, | |
| "learning_rate": 9.90769230769231e-07, | |
| "loss": 0.0081, | |
| "step": 18075 | |
| }, | |
| { | |
| "epoch": 7.654895326707549, | |
| "grad_norm": 0.1523977816104889, | |
| "learning_rate": 9.77948717948718e-07, | |
| "loss": 0.0028, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 7.665468386551068, | |
| "grad_norm": 0.2194071263074875, | |
| "learning_rate": 9.651282051282052e-07, | |
| "loss": 0.0027, | |
| "step": 18125 | |
| }, | |
| { | |
| "epoch": 7.676041446394587, | |
| "grad_norm": 0.23766735196113586, | |
| "learning_rate": 9.523076923076924e-07, | |
| "loss": 0.0031, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 7.686614506238105, | |
| "grad_norm": 0.43536919355392456, | |
| "learning_rate": 9.394871794871796e-07, | |
| "loss": 0.0059, | |
| "step": 18175 | |
| }, | |
| { | |
| "epoch": 7.697187566081624, | |
| "grad_norm": 0.20056919753551483, | |
| "learning_rate": 9.266666666666667e-07, | |
| "loss": 0.0027, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 7.707760625925143, | |
| "grad_norm": 0.267135888338089, | |
| "learning_rate": 9.138461538461539e-07, | |
| "loss": 0.0046, | |
| "step": 18225 | |
| }, | |
| { | |
| "epoch": 7.718333685768662, | |
| "grad_norm": 0.12573710083961487, | |
| "learning_rate": 9.010256410256411e-07, | |
| "loss": 0.0043, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 7.72890674561218, | |
| "grad_norm": 0.12320298701524734, | |
| "learning_rate": 8.882051282051282e-07, | |
| "loss": 0.0047, | |
| "step": 18275 | |
| }, | |
| { | |
| "epoch": 7.739479805455699, | |
| "grad_norm": 0.18382315337657928, | |
| "learning_rate": 8.753846153846154e-07, | |
| "loss": 0.0029, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 7.750052865299217, | |
| "grad_norm": 1.264244794845581, | |
| "learning_rate": 8.625641025641027e-07, | |
| "loss": 0.0029, | |
| "step": 18325 | |
| }, | |
| { | |
| "epoch": 7.760625925142737, | |
| "grad_norm": 0.31755584478378296, | |
| "learning_rate": 8.497435897435897e-07, | |
| "loss": 0.0023, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 7.771198984986255, | |
| "grad_norm": 0.14317964017391205, | |
| "learning_rate": 8.36923076923077e-07, | |
| "loss": 0.0037, | |
| "step": 18375 | |
| }, | |
| { | |
| "epoch": 7.781772044829774, | |
| "grad_norm": 0.1884177327156067, | |
| "learning_rate": 8.241025641025642e-07, | |
| "loss": 0.0022, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 7.792345104673292, | |
| "grad_norm": 0.24375373125076294, | |
| "learning_rate": 8.112820512820512e-07, | |
| "loss": 0.0048, | |
| "step": 18425 | |
| }, | |
| { | |
| "epoch": 7.802918164516811, | |
| "grad_norm": 0.2112288922071457, | |
| "learning_rate": 7.984615384615385e-07, | |
| "loss": 0.0038, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 7.81349122436033, | |
| "grad_norm": 0.21161673963069916, | |
| "learning_rate": 7.856410256410257e-07, | |
| "loss": 0.0036, | |
| "step": 18475 | |
| }, | |
| { | |
| "epoch": 7.824064284203849, | |
| "grad_norm": 0.17189516127109528, | |
| "learning_rate": 7.728205128205128e-07, | |
| "loss": 0.0029, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 7.834637344047367, | |
| "grad_norm": 0.28677284717559814, | |
| "learning_rate": 7.6e-07, | |
| "loss": 0.0069, | |
| "step": 18525 | |
| }, | |
| { | |
| "epoch": 7.845210403890886, | |
| "grad_norm": 0.5086202025413513, | |
| "learning_rate": 7.471794871794873e-07, | |
| "loss": 0.0024, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 7.855783463734404, | |
| "grad_norm": 0.20552664995193481, | |
| "learning_rate": 7.343589743589743e-07, | |
| "loss": 0.0064, | |
| "step": 18575 | |
| }, | |
| { | |
| "epoch": 7.866356523577924, | |
| "grad_norm": 0.17549338936805725, | |
| "learning_rate": 7.215384615384616e-07, | |
| "loss": 0.0026, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 7.876929583421442, | |
| "grad_norm": 0.28019997477531433, | |
| "learning_rate": 7.087179487179488e-07, | |
| "loss": 0.0037, | |
| "step": 18625 | |
| }, | |
| { | |
| "epoch": 7.887502643264961, | |
| "grad_norm": 0.1256396323442459, | |
| "learning_rate": 6.958974358974358e-07, | |
| "loss": 0.0045, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 7.898075703108479, | |
| "grad_norm": 0.3035779893398285, | |
| "learning_rate": 6.830769230769231e-07, | |
| "loss": 0.0047, | |
| "step": 18675 | |
| }, | |
| { | |
| "epoch": 7.908648762951998, | |
| "grad_norm": 0.2889888882637024, | |
| "learning_rate": 6.702564102564103e-07, | |
| "loss": 0.0045, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 7.919221822795517, | |
| "grad_norm": 0.17718684673309326, | |
| "learning_rate": 6.574358974358976e-07, | |
| "loss": 0.0028, | |
| "step": 18725 | |
| }, | |
| { | |
| "epoch": 7.929794882639036, | |
| "grad_norm": 0.16636326909065247, | |
| "learning_rate": 6.446153846153846e-07, | |
| "loss": 0.0023, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 7.940367942482554, | |
| "grad_norm": 0.21917343139648438, | |
| "learning_rate": 6.317948717948719e-07, | |
| "loss": 0.0066, | |
| "step": 18775 | |
| }, | |
| { | |
| "epoch": 7.950941002326073, | |
| "grad_norm": 0.1804770529270172, | |
| "learning_rate": 6.18974358974359e-07, | |
| "loss": 0.0022, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 7.9615140621695915, | |
| "grad_norm": 0.15297465026378632, | |
| "learning_rate": 6.061538461538462e-07, | |
| "loss": 0.0025, | |
| "step": 18825 | |
| }, | |
| { | |
| "epoch": 7.972087122013111, | |
| "grad_norm": 2.2187414169311523, | |
| "learning_rate": 5.933333333333334e-07, | |
| "loss": 0.0036, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 7.982660181856629, | |
| "grad_norm": 0.14911529421806335, | |
| "learning_rate": 5.805128205128205e-07, | |
| "loss": 0.0035, | |
| "step": 18875 | |
| }, | |
| { | |
| "epoch": 7.993233241700148, | |
| "grad_norm": 0.3315187692642212, | |
| "learning_rate": 5.676923076923077e-07, | |
| "loss": 0.0031, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 8.003806301543667, | |
| "grad_norm": 0.2590886950492859, | |
| "learning_rate": 5.548717948717949e-07, | |
| "loss": 0.0037, | |
| "step": 18925 | |
| }, | |
| { | |
| "epoch": 8.014379361387185, | |
| "grad_norm": 0.15489330887794495, | |
| "learning_rate": 5.420512820512821e-07, | |
| "loss": 0.0018, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 8.024952421230704, | |
| "grad_norm": 0.18797287344932556, | |
| "learning_rate": 5.292307692307692e-07, | |
| "loss": 0.0018, | |
| "step": 18975 | |
| }, | |
| { | |
| "epoch": 8.035525481074222, | |
| "grad_norm": 0.1324182152748108, | |
| "learning_rate": 5.164102564102565e-07, | |
| "loss": 0.0021, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 8.035525481074222, | |
| "eval_loss": 0.1434488594532013, | |
| "eval_runtime": 479.0168, | |
| "eval_samples_per_second": 7.601, | |
| "eval_steps_per_second": 0.952, | |
| "eval_wer": 0.1180101164787229, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 8.046098540917741, | |
| "grad_norm": 0.15593498945236206, | |
| "learning_rate": 5.035897435897436e-07, | |
| "loss": 0.0018, | |
| "step": 19025 | |
| }, | |
| { | |
| "epoch": 8.05667160076126, | |
| "grad_norm": 0.14166687428951263, | |
| "learning_rate": 4.907692307692308e-07, | |
| "loss": 0.0025, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 8.067244660604779, | |
| "grad_norm": 0.06104918569326401, | |
| "learning_rate": 4.77948717948718e-07, | |
| "loss": 0.0018, | |
| "step": 19075 | |
| }, | |
| { | |
| "epoch": 8.077817720448298, | |
| "grad_norm": 0.11749964207410812, | |
| "learning_rate": 4.6512820512820514e-07, | |
| "loss": 0.0031, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 8.088390780291816, | |
| "grad_norm": 0.09838801622390747, | |
| "learning_rate": 4.523076923076924e-07, | |
| "loss": 0.0031, | |
| "step": 19125 | |
| }, | |
| { | |
| "epoch": 8.098963840135335, | |
| "grad_norm": 0.1402641385793686, | |
| "learning_rate": 4.3948717948717953e-07, | |
| "loss": 0.0027, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 8.109536899978854, | |
| "grad_norm": 0.16856901347637177, | |
| "learning_rate": 4.266666666666667e-07, | |
| "loss": 0.0023, | |
| "step": 19175 | |
| }, | |
| { | |
| "epoch": 8.120109959822372, | |
| "grad_norm": 0.12997014820575714, | |
| "learning_rate": 4.138461538461539e-07, | |
| "loss": 0.0058, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 8.130683019665891, | |
| "grad_norm": 0.14169169962406158, | |
| "learning_rate": 4.0102564102564107e-07, | |
| "loss": 0.0036, | |
| "step": 19225 | |
| }, | |
| { | |
| "epoch": 8.14125607950941, | |
| "grad_norm": 0.09970966726541519, | |
| "learning_rate": 3.882051282051282e-07, | |
| "loss": 0.0022, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 8.151829139352929, | |
| "grad_norm": 1.0204663276672363, | |
| "learning_rate": 3.7538461538461546e-07, | |
| "loss": 0.0023, | |
| "step": 19275 | |
| }, | |
| { | |
| "epoch": 8.162402199196448, | |
| "grad_norm": 0.24392949044704437, | |
| "learning_rate": 3.625641025641026e-07, | |
| "loss": 0.0029, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 8.172975259039966, | |
| "grad_norm": 0.12695328891277313, | |
| "learning_rate": 3.4974358974358974e-07, | |
| "loss": 0.0035, | |
| "step": 19325 | |
| }, | |
| { | |
| "epoch": 8.183548318883485, | |
| "grad_norm": 0.15973243117332458, | |
| "learning_rate": 3.36923076923077e-07, | |
| "loss": 0.0019, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 8.194121378727004, | |
| "grad_norm": 0.2357715517282486, | |
| "learning_rate": 3.2410256410256413e-07, | |
| "loss": 0.0019, | |
| "step": 19375 | |
| }, | |
| { | |
| "epoch": 8.204694438570522, | |
| "grad_norm": 0.15530213713645935, | |
| "learning_rate": 3.112820512820513e-07, | |
| "loss": 0.0019, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 8.215267498414041, | |
| "grad_norm": 3.935415267944336, | |
| "learning_rate": 2.9846153846153847e-07, | |
| "loss": 0.0065, | |
| "step": 19425 | |
| }, | |
| { | |
| "epoch": 8.225840558257559, | |
| "grad_norm": 3.079641580581665, | |
| "learning_rate": 2.8564102564102566e-07, | |
| "loss": 0.004, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 8.236413618101079, | |
| "grad_norm": 0.1831885129213333, | |
| "learning_rate": 2.7282051282051286e-07, | |
| "loss": 0.0018, | |
| "step": 19475 | |
| }, | |
| { | |
| "epoch": 8.246986677944598, | |
| "grad_norm": 0.11785794049501419, | |
| "learning_rate": 2.6e-07, | |
| "loss": 0.0018, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 8.257559737788116, | |
| "grad_norm": 0.3607770800590515, | |
| "learning_rate": 2.471794871794872e-07, | |
| "loss": 0.0032, | |
| "step": 19525 | |
| }, | |
| { | |
| "epoch": 8.268132797631635, | |
| "grad_norm": 0.2167283445596695, | |
| "learning_rate": 2.343589743589744e-07, | |
| "loss": 0.0019, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 8.278705857475153, | |
| "grad_norm": 0.1869659721851349, | |
| "learning_rate": 2.2153846153846153e-07, | |
| "loss": 0.0021, | |
| "step": 19575 | |
| }, | |
| { | |
| "epoch": 8.289278917318672, | |
| "grad_norm": 0.16754354536533356, | |
| "learning_rate": 2.0871794871794873e-07, | |
| "loss": 0.002, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 8.299851977162191, | |
| "grad_norm": 0.8075239658355713, | |
| "learning_rate": 1.9589743589743592e-07, | |
| "loss": 0.003, | |
| "step": 19625 | |
| }, | |
| { | |
| "epoch": 8.31042503700571, | |
| "grad_norm": 2.266139507293701, | |
| "learning_rate": 1.8307692307692306e-07, | |
| "loss": 0.0065, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 8.320998096849229, | |
| "grad_norm": 0.1474863439798355, | |
| "learning_rate": 1.7025641025641026e-07, | |
| "loss": 0.0056, | |
| "step": 19675 | |
| }, | |
| { | |
| "epoch": 8.331571156692746, | |
| "grad_norm": 0.1871362030506134, | |
| "learning_rate": 1.5743589743589745e-07, | |
| "loss": 0.0047, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 8.342144216536266, | |
| "grad_norm": 0.14866340160369873, | |
| "learning_rate": 1.4461538461538462e-07, | |
| "loss": 0.0037, | |
| "step": 19725 | |
| }, | |
| { | |
| "epoch": 8.352717276379785, | |
| "grad_norm": 0.15600642561912537, | |
| "learning_rate": 1.317948717948718e-07, | |
| "loss": 0.0038, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 8.363290336223303, | |
| "grad_norm": 0.18436697125434875, | |
| "learning_rate": 1.1897435897435898e-07, | |
| "loss": 0.0024, | |
| "step": 19775 | |
| }, | |
| { | |
| "epoch": 8.373863396066822, | |
| "grad_norm": 0.12687981128692627, | |
| "learning_rate": 1.0615384615384615e-07, | |
| "loss": 0.0018, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 8.38443645591034, | |
| "grad_norm": 0.11940109729766846, | |
| "learning_rate": 9.333333333333335e-08, | |
| "loss": 0.0025, | |
| "step": 19825 | |
| }, | |
| { | |
| "epoch": 8.39500951575386, | |
| "grad_norm": 0.16202621161937714, | |
| "learning_rate": 8.051282051282052e-08, | |
| "loss": 0.0028, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 8.405582575597379, | |
| "grad_norm": 0.09489905834197998, | |
| "learning_rate": 6.76923076923077e-08, | |
| "loss": 0.0053, | |
| "step": 19875 | |
| }, | |
| { | |
| "epoch": 8.416155635440896, | |
| "grad_norm": 1.8258126974105835, | |
| "learning_rate": 5.4871794871794874e-08, | |
| "loss": 0.0029, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 8.426728695284416, | |
| "grad_norm": 0.18347540497779846, | |
| "learning_rate": 4.2051282051282056e-08, | |
| "loss": 0.002, | |
| "step": 19925 | |
| }, | |
| { | |
| "epoch": 8.437301755127933, | |
| "grad_norm": 0.19024056196212769, | |
| "learning_rate": 2.9230769230769234e-08, | |
| "loss": 0.0029, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 8.447874814971453, | |
| "grad_norm": 0.21589778363704681, | |
| "learning_rate": 1.641025641025641e-08, | |
| "loss": 0.0031, | |
| "step": 19975 | |
| }, | |
| { | |
| "epoch": 8.458447874814972, | |
| "grad_norm": 0.12975341081619263, | |
| "learning_rate": 3.5897435897435903e-09, | |
| "loss": 0.0018, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 8.458447874814972, | |
| "eval_loss": 0.1440751701593399, | |
| "eval_runtime": 481.615, | |
| "eval_samples_per_second": 7.56, | |
| "eval_steps_per_second": 0.947, | |
| "eval_wer": 0.1183813634043343, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 8.458447874814972, | |
| "step": 20000, | |
| "total_flos": 2.07526043713536e+19, | |
| "train_loss": 0.09297830064073205, | |
| "train_runtime": 50820.1517, | |
| "train_samples_per_second": 6.297, | |
| "train_steps_per_second": 0.394 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 20000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9, | |
| "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": 2.07526043713536e+19, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |