| [ | |
| { | |
| "loss": 1.2604, | |
| "grad_norm": 10.599270820617676, | |
| "learning_rate": 4.000000000000001e-06, | |
| "epoch": 0.004071661237785016, | |
| "step": 10 | |
| }, | |
| { | |
| "loss": 1.1496, | |
| "grad_norm": 15.23047924041748, | |
| "learning_rate": 8.5e-06, | |
| "epoch": 0.008143322475570033, | |
| "step": 20 | |
| }, | |
| { | |
| "loss": 0.9217, | |
| "grad_norm": 13.265022277832031, | |
| "learning_rate": 9.985943775100404e-06, | |
| "epoch": 0.012214983713355049, | |
| "step": 30 | |
| }, | |
| { | |
| "loss": 0.7287, | |
| "grad_norm": 9.713923454284668, | |
| "learning_rate": 9.965863453815262e-06, | |
| "epoch": 0.016286644951140065, | |
| "step": 40 | |
| }, | |
| { | |
| "loss": 0.6314, | |
| "grad_norm": 12.423747062683105, | |
| "learning_rate": 9.945783132530121e-06, | |
| "epoch": 0.02035830618892508, | |
| "step": 50 | |
| }, | |
| { | |
| "loss": 0.6129, | |
| "grad_norm": 17.610557556152344, | |
| "learning_rate": 9.925702811244981e-06, | |
| "epoch": 0.024429967426710098, | |
| "step": 60 | |
| }, | |
| { | |
| "loss": 0.5291, | |
| "grad_norm": 9.495976448059082, | |
| "learning_rate": 9.90562248995984e-06, | |
| "epoch": 0.028501628664495113, | |
| "step": 70 | |
| }, | |
| { | |
| "loss": 0.488, | |
| "grad_norm": 10.973204612731934, | |
| "learning_rate": 9.885542168674699e-06, | |
| "epoch": 0.03257328990228013, | |
| "step": 80 | |
| }, | |
| { | |
| "loss": 0.4437, | |
| "grad_norm": 11.19155216217041, | |
| "learning_rate": 9.865461847389559e-06, | |
| "epoch": 0.03664495114006515, | |
| "step": 90 | |
| }, | |
| { | |
| "loss": 0.4034, | |
| "grad_norm": 8.56933307647705, | |
| "learning_rate": 9.845381526104419e-06, | |
| "epoch": 0.04071661237785016, | |
| "step": 100 | |
| }, | |
| { | |
| "loss": 0.4185, | |
| "grad_norm": 9.449453353881836, | |
| "learning_rate": 9.825301204819277e-06, | |
| "epoch": 0.04478827361563518, | |
| "step": 110 | |
| }, | |
| { | |
| "loss": 0.3604, | |
| "grad_norm": 9.627606391906738, | |
| "learning_rate": 9.805220883534138e-06, | |
| "epoch": 0.048859934853420196, | |
| "step": 120 | |
| }, | |
| { | |
| "loss": 0.3399, | |
| "grad_norm": 10.60472297668457, | |
| "learning_rate": 9.785140562248996e-06, | |
| "epoch": 0.052931596091205214, | |
| "step": 130 | |
| }, | |
| { | |
| "loss": 0.351, | |
| "grad_norm": 7.827078342437744, | |
| "learning_rate": 9.765060240963856e-06, | |
| "epoch": 0.057003257328990226, | |
| "step": 140 | |
| }, | |
| { | |
| "loss": 0.33, | |
| "grad_norm": 7.788185119628906, | |
| "learning_rate": 9.744979919678716e-06, | |
| "epoch": 0.061074918566775244, | |
| "step": 150 | |
| }, | |
| { | |
| "loss": 0.314, | |
| "grad_norm": 8.686625480651855, | |
| "learning_rate": 9.724899598393574e-06, | |
| "epoch": 0.06514657980456026, | |
| "step": 160 | |
| }, | |
| { | |
| "loss": 0.326, | |
| "grad_norm": 8.329667091369629, | |
| "learning_rate": 9.704819277108436e-06, | |
| "epoch": 0.06921824104234528, | |
| "step": 170 | |
| }, | |
| { | |
| "loss": 0.2819, | |
| "grad_norm": 5.676497459411621, | |
| "learning_rate": 9.684738955823294e-06, | |
| "epoch": 0.0732899022801303, | |
| "step": 180 | |
| }, | |
| { | |
| "loss": 0.256, | |
| "grad_norm": 10.663836479187012, | |
| "learning_rate": 9.664658634538154e-06, | |
| "epoch": 0.07736156351791532, | |
| "step": 190 | |
| }, | |
| { | |
| "loss": 0.2586, | |
| "grad_norm": 6.807067394256592, | |
| "learning_rate": 9.644578313253012e-06, | |
| "epoch": 0.08143322475570032, | |
| "step": 200 | |
| }, | |
| { | |
| "loss": 0.2388, | |
| "grad_norm": 6.985593795776367, | |
| "learning_rate": 9.624497991967873e-06, | |
| "epoch": 0.08550488599348534, | |
| "step": 210 | |
| }, | |
| { | |
| "loss": 0.217, | |
| "grad_norm": 8.37208080291748, | |
| "learning_rate": 9.604417670682731e-06, | |
| "epoch": 0.08957654723127036, | |
| "step": 220 | |
| }, | |
| { | |
| "loss": 0.2444, | |
| "grad_norm": 6.593284606933594, | |
| "learning_rate": 9.584337349397591e-06, | |
| "epoch": 0.09364820846905537, | |
| "step": 230 | |
| }, | |
| { | |
| "loss": 0.2287, | |
| "grad_norm": 4.794759750366211, | |
| "learning_rate": 9.564257028112451e-06, | |
| "epoch": 0.09771986970684039, | |
| "step": 240 | |
| }, | |
| { | |
| "loss": 0.2421, | |
| "grad_norm": 7.0008673667907715, | |
| "learning_rate": 9.54417670682731e-06, | |
| "epoch": 0.10179153094462541, | |
| "step": 250 | |
| }, | |
| { | |
| "loss": 0.2073, | |
| "grad_norm": 8.804052352905273, | |
| "learning_rate": 9.52409638554217e-06, | |
| "epoch": 0.10586319218241043, | |
| "step": 260 | |
| }, | |
| { | |
| "loss": 0.2504, | |
| "grad_norm": 6.55370569229126, | |
| "learning_rate": 9.504016064257029e-06, | |
| "epoch": 0.10993485342019543, | |
| "step": 270 | |
| }, | |
| { | |
| "loss": 0.1924, | |
| "grad_norm": 4.867704391479492, | |
| "learning_rate": 9.483935742971888e-06, | |
| "epoch": 0.11400651465798045, | |
| "step": 280 | |
| }, | |
| { | |
| "loss": 0.2319, | |
| "grad_norm": 5.787832260131836, | |
| "learning_rate": 9.463855421686746e-06, | |
| "epoch": 0.11807817589576547, | |
| "step": 290 | |
| }, | |
| { | |
| "loss": 0.235, | |
| "grad_norm": 5.37102746963501, | |
| "learning_rate": 9.443775100401608e-06, | |
| "epoch": 0.12214983713355049, | |
| "step": 300 | |
| }, | |
| { | |
| "loss": 0.2322, | |
| "grad_norm": 8.406855583190918, | |
| "learning_rate": 9.423694779116466e-06, | |
| "epoch": 0.1262214983713355, | |
| "step": 310 | |
| }, | |
| { | |
| "loss": 0.2085, | |
| "grad_norm": 5.433464527130127, | |
| "learning_rate": 9.403614457831326e-06, | |
| "epoch": 0.13029315960912052, | |
| "step": 320 | |
| }, | |
| { | |
| "loss": 0.2052, | |
| "grad_norm": 5.475189685821533, | |
| "learning_rate": 9.383534136546186e-06, | |
| "epoch": 0.13436482084690554, | |
| "step": 330 | |
| }, | |
| { | |
| "loss": 0.1985, | |
| "grad_norm": 9.26491928100586, | |
| "learning_rate": 9.363453815261045e-06, | |
| "epoch": 0.13843648208469056, | |
| "step": 340 | |
| }, | |
| { | |
| "loss": 0.1835, | |
| "grad_norm": 8.15815544128418, | |
| "learning_rate": 9.343373493975905e-06, | |
| "epoch": 0.14250814332247558, | |
| "step": 350 | |
| }, | |
| { | |
| "loss": 0.22, | |
| "grad_norm": 6.896279335021973, | |
| "learning_rate": 9.323293172690763e-06, | |
| "epoch": 0.1465798045602606, | |
| "step": 360 | |
| }, | |
| { | |
| "loss": 0.1886, | |
| "grad_norm": 5.942968368530273, | |
| "learning_rate": 9.303212851405623e-06, | |
| "epoch": 0.15065146579804561, | |
| "step": 370 | |
| }, | |
| { | |
| "loss": 0.1874, | |
| "grad_norm": 7.721154689788818, | |
| "learning_rate": 9.283132530120483e-06, | |
| "epoch": 0.15472312703583063, | |
| "step": 380 | |
| }, | |
| { | |
| "loss": 0.1774, | |
| "grad_norm": 6.1584014892578125, | |
| "learning_rate": 9.263052208835343e-06, | |
| "epoch": 0.15879478827361562, | |
| "step": 390 | |
| }, | |
| { | |
| "loss": 0.1804, | |
| "grad_norm": 6.08401346206665, | |
| "learning_rate": 9.242971887550201e-06, | |
| "epoch": 0.16286644951140064, | |
| "step": 400 | |
| }, | |
| { | |
| "loss": 0.1933, | |
| "grad_norm": 5.218653678894043, | |
| "learning_rate": 9.22289156626506e-06, | |
| "epoch": 0.16693811074918566, | |
| "step": 410 | |
| }, | |
| { | |
| "loss": 0.1946, | |
| "grad_norm": 6.643286228179932, | |
| "learning_rate": 9.20281124497992e-06, | |
| "epoch": 0.17100977198697068, | |
| "step": 420 | |
| }, | |
| { | |
| "loss": 0.1994, | |
| "grad_norm": 7.308907508850098, | |
| "learning_rate": 9.18273092369478e-06, | |
| "epoch": 0.1750814332247557, | |
| "step": 430 | |
| }, | |
| { | |
| "loss": 0.1705, | |
| "grad_norm": 5.21204137802124, | |
| "learning_rate": 9.16265060240964e-06, | |
| "epoch": 0.1791530944625407, | |
| "step": 440 | |
| }, | |
| { | |
| "loss": 0.178, | |
| "grad_norm": 6.208616256713867, | |
| "learning_rate": 9.142570281124498e-06, | |
| "epoch": 0.18322475570032573, | |
| "step": 450 | |
| }, | |
| { | |
| "loss": 0.185, | |
| "grad_norm": 4.3676605224609375, | |
| "learning_rate": 9.122489959839358e-06, | |
| "epoch": 0.18729641693811075, | |
| "step": 460 | |
| }, | |
| { | |
| "loss": 0.1861, | |
| "grad_norm": 6.359353065490723, | |
| "learning_rate": 9.102409638554218e-06, | |
| "epoch": 0.19136807817589577, | |
| "step": 470 | |
| }, | |
| { | |
| "loss": 0.1848, | |
| "grad_norm": 5.313674449920654, | |
| "learning_rate": 9.082329317269078e-06, | |
| "epoch": 0.19543973941368079, | |
| "step": 480 | |
| }, | |
| { | |
| "loss": 0.1654, | |
| "grad_norm": 5.061492443084717, | |
| "learning_rate": 9.062248995983937e-06, | |
| "epoch": 0.1995114006514658, | |
| "step": 490 | |
| }, | |
| { | |
| "loss": 0.1642, | |
| "grad_norm": 8.568002700805664, | |
| "learning_rate": 9.042168674698795e-06, | |
| "epoch": 0.20358306188925082, | |
| "step": 500 | |
| }, | |
| { | |
| "eval_loss": 0.17640618979930878, | |
| "eval_wer": 0.5859276588087098, | |
| "eval_runtime": 1092.231, | |
| "eval_samples_per_second": 1.0, | |
| "eval_steps_per_second": 0.125, | |
| "epoch": 0.20358306188925082, | |
| "step": 500 | |
| }, | |
| { | |
| "loss": 0.1597, | |
| "grad_norm": 5.107157230377197, | |
| "learning_rate": 9.022088353413655e-06, | |
| "epoch": 0.20765472312703584, | |
| "step": 510 | |
| }, | |
| { | |
| "loss": 0.1628, | |
| "grad_norm": 4.68630313873291, | |
| "learning_rate": 9.002008032128515e-06, | |
| "epoch": 0.21172638436482086, | |
| "step": 520 | |
| }, | |
| { | |
| "loss": 0.1721, | |
| "grad_norm": 5.860337734222412, | |
| "learning_rate": 8.981927710843375e-06, | |
| "epoch": 0.21579804560260588, | |
| "step": 530 | |
| }, | |
| { | |
| "loss": 0.1872, | |
| "grad_norm": 5.301535606384277, | |
| "learning_rate": 8.961847389558233e-06, | |
| "epoch": 0.21986970684039087, | |
| "step": 540 | |
| }, | |
| { | |
| "loss": 0.1617, | |
| "grad_norm": 3.8535797595977783, | |
| "learning_rate": 8.941767068273093e-06, | |
| "epoch": 0.22394136807817588, | |
| "step": 550 | |
| }, | |
| { | |
| "loss": 0.1779, | |
| "grad_norm": 3.181271553039551, | |
| "learning_rate": 8.921686746987953e-06, | |
| "epoch": 0.2280130293159609, | |
| "step": 560 | |
| }, | |
| { | |
| "loss": 0.1692, | |
| "grad_norm": 4.963680267333984, | |
| "learning_rate": 8.901606425702812e-06, | |
| "epoch": 0.23208469055374592, | |
| "step": 570 | |
| }, | |
| { | |
| "loss": 0.1778, | |
| "grad_norm": 5.0773773193359375, | |
| "learning_rate": 8.881526104417672e-06, | |
| "epoch": 0.23615635179153094, | |
| "step": 580 | |
| }, | |
| { | |
| "loss": 0.1686, | |
| "grad_norm": 4.0839409828186035, | |
| "learning_rate": 8.86144578313253e-06, | |
| "epoch": 0.24022801302931596, | |
| "step": 590 | |
| }, | |
| { | |
| "loss": 0.163, | |
| "grad_norm": 5.539490222930908, | |
| "learning_rate": 8.84136546184739e-06, | |
| "epoch": 0.24429967426710097, | |
| "step": 600 | |
| }, | |
| { | |
| "loss": 0.1503, | |
| "grad_norm": 7.714535713195801, | |
| "learning_rate": 8.82128514056225e-06, | |
| "epoch": 0.248371335504886, | |
| "step": 610 | |
| }, | |
| { | |
| "loss": 0.1519, | |
| "grad_norm": 5.541062831878662, | |
| "learning_rate": 8.80120481927711e-06, | |
| "epoch": 0.252442996742671, | |
| "step": 620 | |
| }, | |
| { | |
| "loss": 0.1517, | |
| "grad_norm": 4.906521320343018, | |
| "learning_rate": 8.781124497991968e-06, | |
| "epoch": 0.25651465798045603, | |
| "step": 630 | |
| }, | |
| { | |
| "loss": 0.1669, | |
| "grad_norm": 5.647503852844238, | |
| "learning_rate": 8.761044176706828e-06, | |
| "epoch": 0.26058631921824105, | |
| "step": 640 | |
| }, | |
| { | |
| "loss": 0.1974, | |
| "grad_norm": 9.37059211730957, | |
| "learning_rate": 8.740963855421687e-06, | |
| "epoch": 0.26465798045602607, | |
| "step": 650 | |
| }, | |
| { | |
| "loss": 0.1342, | |
| "grad_norm": 4.161442756652832, | |
| "learning_rate": 8.720883534136547e-06, | |
| "epoch": 0.2687296416938111, | |
| "step": 660 | |
| }, | |
| { | |
| "loss": 0.1646, | |
| "grad_norm": 6.539823532104492, | |
| "learning_rate": 8.700803212851407e-06, | |
| "epoch": 0.2728013029315961, | |
| "step": 670 | |
| }, | |
| { | |
| "loss": 0.1736, | |
| "grad_norm": 4.1773881912231445, | |
| "learning_rate": 8.680722891566265e-06, | |
| "epoch": 0.2768729641693811, | |
| "step": 680 | |
| }, | |
| { | |
| "loss": 0.1457, | |
| "grad_norm": 5.162631988525391, | |
| "learning_rate": 8.660642570281125e-06, | |
| "epoch": 0.28094462540716614, | |
| "step": 690 | |
| }, | |
| { | |
| "loss": 0.1421, | |
| "grad_norm": 5.8395490646362305, | |
| "learning_rate": 8.640562248995985e-06, | |
| "epoch": 0.28501628664495116, | |
| "step": 700 | |
| }, | |
| { | |
| "loss": 0.1482, | |
| "grad_norm": 5.02829647064209, | |
| "learning_rate": 8.620481927710845e-06, | |
| "epoch": 0.2890879478827362, | |
| "step": 710 | |
| }, | |
| { | |
| "loss": 0.1729, | |
| "grad_norm": 4.14141321182251, | |
| "learning_rate": 8.600401606425703e-06, | |
| "epoch": 0.2931596091205212, | |
| "step": 720 | |
| }, | |
| { | |
| "loss": 0.1507, | |
| "grad_norm": 4.756168365478516, | |
| "learning_rate": 8.580321285140562e-06, | |
| "epoch": 0.2972312703583062, | |
| "step": 730 | |
| }, | |
| { | |
| "loss": 0.1403, | |
| "grad_norm": 5.376400947570801, | |
| "learning_rate": 8.560240963855422e-06, | |
| "epoch": 0.30130293159609123, | |
| "step": 740 | |
| }, | |
| { | |
| "loss": 0.1492, | |
| "grad_norm": 3.6600284576416016, | |
| "learning_rate": 8.540160642570282e-06, | |
| "epoch": 0.30537459283387625, | |
| "step": 750 | |
| }, | |
| { | |
| "loss": 0.1512, | |
| "grad_norm": 4.753564357757568, | |
| "learning_rate": 8.520080321285142e-06, | |
| "epoch": 0.30944625407166126, | |
| "step": 760 | |
| }, | |
| { | |
| "loss": 0.1595, | |
| "grad_norm": 5.130186557769775, | |
| "learning_rate": 8.5e-06, | |
| "epoch": 0.3135179153094462, | |
| "step": 770 | |
| }, | |
| { | |
| "loss": 0.1524, | |
| "grad_norm": 4.184345245361328, | |
| "learning_rate": 8.47991967871486e-06, | |
| "epoch": 0.31758957654723124, | |
| "step": 780 | |
| }, | |
| { | |
| "loss": 0.1511, | |
| "grad_norm": 5.108601093292236, | |
| "learning_rate": 8.45983935742972e-06, | |
| "epoch": 0.32166123778501626, | |
| "step": 790 | |
| }, | |
| { | |
| "loss": 0.179, | |
| "grad_norm": 3.4104130268096924, | |
| "learning_rate": 8.43975903614458e-06, | |
| "epoch": 0.3257328990228013, | |
| "step": 800 | |
| }, | |
| { | |
| "loss": 0.1678, | |
| "grad_norm": 3.956003189086914, | |
| "learning_rate": 8.419678714859439e-06, | |
| "epoch": 0.3298045602605863, | |
| "step": 810 | |
| }, | |
| { | |
| "loss": 0.1557, | |
| "grad_norm": 4.92228364944458, | |
| "learning_rate": 8.399598393574297e-06, | |
| "epoch": 0.3338762214983713, | |
| "step": 820 | |
| }, | |
| { | |
| "loss": 0.1273, | |
| "grad_norm": 4.740561485290527, | |
| "learning_rate": 8.379518072289157e-06, | |
| "epoch": 0.33794788273615634, | |
| "step": 830 | |
| }, | |
| { | |
| "loss": 0.1308, | |
| "grad_norm": 4.872125148773193, | |
| "learning_rate": 8.359437751004017e-06, | |
| "epoch": 0.34201954397394135, | |
| "step": 840 | |
| }, | |
| { | |
| "loss": 0.1333, | |
| "grad_norm": 4.819728851318359, | |
| "learning_rate": 8.339357429718877e-06, | |
| "epoch": 0.34609120521172637, | |
| "step": 850 | |
| }, | |
| { | |
| "loss": 0.1657, | |
| "grad_norm": 4.517435550689697, | |
| "learning_rate": 8.319277108433735e-06, | |
| "epoch": 0.3501628664495114, | |
| "step": 860 | |
| }, | |
| { | |
| "loss": 0.1361, | |
| "grad_norm": 4.909110069274902, | |
| "learning_rate": 8.299196787148595e-06, | |
| "epoch": 0.3542345276872964, | |
| "step": 870 | |
| }, | |
| { | |
| "loss": 0.1408, | |
| "grad_norm": 4.956007480621338, | |
| "learning_rate": 8.279116465863454e-06, | |
| "epoch": 0.3583061889250814, | |
| "step": 880 | |
| }, | |
| { | |
| "loss": 0.1368, | |
| "grad_norm": 4.8904643058776855, | |
| "learning_rate": 8.259036144578314e-06, | |
| "epoch": 0.36237785016286644, | |
| "step": 890 | |
| }, | |
| { | |
| "loss": 0.1341, | |
| "grad_norm": 4.813321113586426, | |
| "learning_rate": 8.238955823293174e-06, | |
| "epoch": 0.36644951140065146, | |
| "step": 900 | |
| }, | |
| { | |
| "loss": 0.131, | |
| "grad_norm": 3.7235021591186523, | |
| "learning_rate": 8.218875502008032e-06, | |
| "epoch": 0.3705211726384365, | |
| "step": 910 | |
| }, | |
| { | |
| "loss": 0.1301, | |
| "grad_norm": 5.953886985778809, | |
| "learning_rate": 8.198795180722894e-06, | |
| "epoch": 0.3745928338762215, | |
| "step": 920 | |
| }, | |
| { | |
| "loss": 0.1282, | |
| "grad_norm": 4.092996120452881, | |
| "learning_rate": 8.178714859437752e-06, | |
| "epoch": 0.3786644951140065, | |
| "step": 930 | |
| }, | |
| { | |
| "loss": 0.1249, | |
| "grad_norm": 4.137129783630371, | |
| "learning_rate": 8.158634538152611e-06, | |
| "epoch": 0.38273615635179153, | |
| "step": 940 | |
| }, | |
| { | |
| "loss": 0.1616, | |
| "grad_norm": 4.147120952606201, | |
| "learning_rate": 8.13855421686747e-06, | |
| "epoch": 0.38680781758957655, | |
| "step": 950 | |
| }, | |
| { | |
| "loss": 0.134, | |
| "grad_norm": 4.717681407928467, | |
| "learning_rate": 8.11847389558233e-06, | |
| "epoch": 0.39087947882736157, | |
| "step": 960 | |
| }, | |
| { | |
| "loss": 0.126, | |
| "grad_norm": 2.7443370819091797, | |
| "learning_rate": 8.09839357429719e-06, | |
| "epoch": 0.3949511400651466, | |
| "step": 970 | |
| }, | |
| { | |
| "loss": 0.1428, | |
| "grad_norm": 3.535062551498413, | |
| "learning_rate": 8.078313253012049e-06, | |
| "epoch": 0.3990228013029316, | |
| "step": 980 | |
| }, | |
| { | |
| "loss": 0.1483, | |
| "grad_norm": 4.808443069458008, | |
| "learning_rate": 8.058232931726909e-06, | |
| "epoch": 0.4030944625407166, | |
| "step": 990 | |
| }, | |
| { | |
| "loss": 0.1501, | |
| "grad_norm": 4.920691013336182, | |
| "learning_rate": 8.038152610441767e-06, | |
| "epoch": 0.40716612377850164, | |
| "step": 1000 | |
| }, | |
| { | |
| "eval_loss": 0.1334841102361679, | |
| "eval_wer": 0.4332673504888729, | |
| "eval_runtime": 1085.2379, | |
| "eval_samples_per_second": 1.006, | |
| "eval_steps_per_second": 0.126, | |
| "epoch": 0.40716612377850164, | |
| "step": 1000 | |
| }, | |
| { | |
| "loss": 0.1452, | |
| "grad_norm": 3.7596540451049805, | |
| "learning_rate": 8.018072289156628e-06, | |
| "epoch": 0.41123778501628666, | |
| "step": 1010 | |
| }, | |
| { | |
| "loss": 0.1477, | |
| "grad_norm": 6.351055145263672, | |
| "learning_rate": 7.997991967871486e-06, | |
| "epoch": 0.4153094462540717, | |
| "step": 1020 | |
| }, | |
| { | |
| "loss": 0.1241, | |
| "grad_norm": 4.097475528717041, | |
| "learning_rate": 7.977911646586346e-06, | |
| "epoch": 0.4193811074918567, | |
| "step": 1030 | |
| }, | |
| { | |
| "loss": 0.1458, | |
| "grad_norm": 5.231217861175537, | |
| "learning_rate": 7.957831325301204e-06, | |
| "epoch": 0.4234527687296417, | |
| "step": 1040 | |
| }, | |
| { | |
| "loss": 0.1613, | |
| "grad_norm": 4.824036598205566, | |
| "learning_rate": 7.937751004016064e-06, | |
| "epoch": 0.42752442996742673, | |
| "step": 1050 | |
| }, | |
| { | |
| "loss": 0.1434, | |
| "grad_norm": 10.601489067077637, | |
| "learning_rate": 7.917670682730924e-06, | |
| "epoch": 0.43159609120521175, | |
| "step": 1060 | |
| }, | |
| { | |
| "loss": 0.1402, | |
| "grad_norm": 3.7711360454559326, | |
| "learning_rate": 7.897590361445784e-06, | |
| "epoch": 0.43566775244299677, | |
| "step": 1070 | |
| }, | |
| { | |
| "loss": 0.1208, | |
| "grad_norm": 3.999556064605713, | |
| "learning_rate": 7.877510040160644e-06, | |
| "epoch": 0.43973941368078173, | |
| "step": 1080 | |
| }, | |
| { | |
| "loss": 0.1345, | |
| "grad_norm": 3.9999871253967285, | |
| "learning_rate": 7.857429718875502e-06, | |
| "epoch": 0.44381107491856675, | |
| "step": 1090 | |
| }, | |
| { | |
| "loss": 0.1447, | |
| "grad_norm": 4.634665012359619, | |
| "learning_rate": 7.837349397590363e-06, | |
| "epoch": 0.44788273615635177, | |
| "step": 1100 | |
| }, | |
| { | |
| "loss": 0.1547, | |
| "grad_norm": 5.034597396850586, | |
| "learning_rate": 7.817269076305221e-06, | |
| "epoch": 0.4519543973941368, | |
| "step": 1110 | |
| }, | |
| { | |
| "loss": 0.1507, | |
| "grad_norm": 5.1984734535217285, | |
| "learning_rate": 7.797188755020081e-06, | |
| "epoch": 0.4560260586319218, | |
| "step": 1120 | |
| }, | |
| { | |
| "loss": 0.1409, | |
| "grad_norm": 3.9791903495788574, | |
| "learning_rate": 7.777108433734941e-06, | |
| "epoch": 0.4600977198697068, | |
| "step": 1130 | |
| }, | |
| { | |
| "loss": 0.1444, | |
| "grad_norm": 5.183425426483154, | |
| "learning_rate": 7.7570281124498e-06, | |
| "epoch": 0.46416938110749184, | |
| "step": 1140 | |
| }, | |
| { | |
| "loss": 0.1295, | |
| "grad_norm": 4.427835464477539, | |
| "learning_rate": 7.736947791164659e-06, | |
| "epoch": 0.46824104234527686, | |
| "step": 1150 | |
| }, | |
| { | |
| "loss": 0.119, | |
| "grad_norm": 3.312504529953003, | |
| "learning_rate": 7.716867469879519e-06, | |
| "epoch": 0.4723127035830619, | |
| "step": 1160 | |
| }, | |
| { | |
| "loss": 0.1211, | |
| "grad_norm": 3.6821799278259277, | |
| "learning_rate": 7.696787148594378e-06, | |
| "epoch": 0.4763843648208469, | |
| "step": 1170 | |
| }, | |
| { | |
| "loss": 0.137, | |
| "grad_norm": 4.483107566833496, | |
| "learning_rate": 7.676706827309236e-06, | |
| "epoch": 0.4804560260586319, | |
| "step": 1180 | |
| }, | |
| { | |
| "loss": 0.1166, | |
| "grad_norm": 2.978266954421997, | |
| "learning_rate": 7.656626506024098e-06, | |
| "epoch": 0.48452768729641693, | |
| "step": 1190 | |
| }, | |
| { | |
| "loss": 0.1222, | |
| "grad_norm": 5.3204779624938965, | |
| "learning_rate": 7.636546184738956e-06, | |
| "epoch": 0.48859934853420195, | |
| "step": 1200 | |
| }, | |
| { | |
| "loss": 0.1081, | |
| "grad_norm": 4.243503570556641, | |
| "learning_rate": 7.616465863453816e-06, | |
| "epoch": 0.49267100977198697, | |
| "step": 1210 | |
| }, | |
| { | |
| "loss": 0.1294, | |
| "grad_norm": 5.24080228805542, | |
| "learning_rate": 7.596385542168675e-06, | |
| "epoch": 0.496742671009772, | |
| "step": 1220 | |
| }, | |
| { | |
| "loss": 0.1177, | |
| "grad_norm": 4.412652969360352, | |
| "learning_rate": 7.5763052208835355e-06, | |
| "epoch": 0.500814332247557, | |
| "step": 1230 | |
| }, | |
| { | |
| "loss": 0.1119, | |
| "grad_norm": 4.703248500823975, | |
| "learning_rate": 7.5562248995983944e-06, | |
| "epoch": 0.504885993485342, | |
| "step": 1240 | |
| }, | |
| { | |
| "loss": 0.132, | |
| "grad_norm": 2.9482319355010986, | |
| "learning_rate": 7.536144578313253e-06, | |
| "epoch": 0.508957654723127, | |
| "step": 1250 | |
| }, | |
| { | |
| "loss": 0.1221, | |
| "grad_norm": 5.138255596160889, | |
| "learning_rate": 7.516064257028113e-06, | |
| "epoch": 0.5130293159609121, | |
| "step": 1260 | |
| }, | |
| { | |
| "loss": 0.1238, | |
| "grad_norm": 2.7300281524658203, | |
| "learning_rate": 7.495983935742972e-06, | |
| "epoch": 0.5171009771986971, | |
| "step": 1270 | |
| }, | |
| { | |
| "loss": 0.1166, | |
| "grad_norm": 3.802260398864746, | |
| "learning_rate": 7.475903614457832e-06, | |
| "epoch": 0.5211726384364821, | |
| "step": 1280 | |
| }, | |
| { | |
| "loss": 0.1266, | |
| "grad_norm": 4.9052324295043945, | |
| "learning_rate": 7.455823293172692e-06, | |
| "epoch": 0.5252442996742671, | |
| "step": 1290 | |
| }, | |
| { | |
| "loss": 0.1314, | |
| "grad_norm": 2.9627578258514404, | |
| "learning_rate": 7.435742971887551e-06, | |
| "epoch": 0.5293159609120521, | |
| "step": 1300 | |
| }, | |
| { | |
| "loss": 0.119, | |
| "grad_norm": 4.511162281036377, | |
| "learning_rate": 7.41566265060241e-06, | |
| "epoch": 0.5333876221498371, | |
| "step": 1310 | |
| }, | |
| { | |
| "loss": 0.1173, | |
| "grad_norm": 4.310062408447266, | |
| "learning_rate": 7.39558232931727e-06, | |
| "epoch": 0.5374592833876222, | |
| "step": 1320 | |
| }, | |
| { | |
| "loss": 0.1153, | |
| "grad_norm": 3.812718391418457, | |
| "learning_rate": 7.375502008032129e-06, | |
| "epoch": 0.5415309446254072, | |
| "step": 1330 | |
| }, | |
| { | |
| "loss": 0.119, | |
| "grad_norm": 4.304417610168457, | |
| "learning_rate": 7.355421686746988e-06, | |
| "epoch": 0.5456026058631922, | |
| "step": 1340 | |
| }, | |
| { | |
| "loss": 0.119, | |
| "grad_norm": 4.510639667510986, | |
| "learning_rate": 7.335341365461848e-06, | |
| "epoch": 0.5496742671009772, | |
| "step": 1350 | |
| }, | |
| { | |
| "loss": 0.1125, | |
| "grad_norm": 4.046252727508545, | |
| "learning_rate": 7.315261044176707e-06, | |
| "epoch": 0.5537459283387622, | |
| "step": 1360 | |
| }, | |
| { | |
| "loss": 0.1132, | |
| "grad_norm": 4.546014308929443, | |
| "learning_rate": 7.295180722891568e-06, | |
| "epoch": 0.5578175895765473, | |
| "step": 1370 | |
| }, | |
| { | |
| "loss": 0.1159, | |
| "grad_norm": 2.908959150314331, | |
| "learning_rate": 7.2751004016064266e-06, | |
| "epoch": 0.5618892508143323, | |
| "step": 1380 | |
| }, | |
| { | |
| "loss": 0.1078, | |
| "grad_norm": 3.725773811340332, | |
| "learning_rate": 7.2550200803212855e-06, | |
| "epoch": 0.5659609120521173, | |
| "step": 1390 | |
| }, | |
| { | |
| "loss": 0.1197, | |
| "grad_norm": 3.7473294734954834, | |
| "learning_rate": 7.2349397590361445e-06, | |
| "epoch": 0.5700325732899023, | |
| "step": 1400 | |
| }, | |
| { | |
| "loss": 0.1177, | |
| "grad_norm": 3.5435144901275635, | |
| "learning_rate": 7.214859437751005e-06, | |
| "epoch": 0.5741042345276873, | |
| "step": 1410 | |
| }, | |
| { | |
| "loss": 0.1158, | |
| "grad_norm": 4.562975883483887, | |
| "learning_rate": 7.194779116465864e-06, | |
| "epoch": 0.5781758957654723, | |
| "step": 1420 | |
| }, | |
| { | |
| "loss": 0.123, | |
| "grad_norm": 4.734598636627197, | |
| "learning_rate": 7.174698795180723e-06, | |
| "epoch": 0.5822475570032574, | |
| "step": 1430 | |
| }, | |
| { | |
| "loss": 0.1122, | |
| "grad_norm": 2.8323323726654053, | |
| "learning_rate": 7.154618473895583e-06, | |
| "epoch": 0.5863192182410424, | |
| "step": 1440 | |
| }, | |
| { | |
| "loss": 0.1205, | |
| "grad_norm": 3.141563653945923, | |
| "learning_rate": 7.134538152610442e-06, | |
| "epoch": 0.5903908794788274, | |
| "step": 1450 | |
| }, | |
| { | |
| "loss": 0.1252, | |
| "grad_norm": 4.478124618530273, | |
| "learning_rate": 7.114457831325302e-06, | |
| "epoch": 0.5944625407166124, | |
| "step": 1460 | |
| }, | |
| { | |
| "loss": 0.1319, | |
| "grad_norm": 4.2382307052612305, | |
| "learning_rate": 7.094377510040161e-06, | |
| "epoch": 0.5985342019543974, | |
| "step": 1470 | |
| }, | |
| { | |
| "loss": 0.1001, | |
| "grad_norm": 3.496803045272827, | |
| "learning_rate": 7.07429718875502e-06, | |
| "epoch": 0.6026058631921825, | |
| "step": 1480 | |
| }, | |
| { | |
| "loss": 0.1257, | |
| "grad_norm": 4.4111104011535645, | |
| "learning_rate": 7.054216867469879e-06, | |
| "epoch": 0.6066775244299675, | |
| "step": 1490 | |
| }, | |
| { | |
| "loss": 0.1201, | |
| "grad_norm": 3.838998556137085, | |
| "learning_rate": 7.03413654618474e-06, | |
| "epoch": 0.6107491856677525, | |
| "step": 1500 | |
| }, | |
| { | |
| "eval_loss": 0.11762961745262146, | |
| "eval_wer": 0.586767440465479, | |
| "eval_runtime": 1062.8817, | |
| "eval_samples_per_second": 1.027, | |
| "eval_steps_per_second": 0.129, | |
| "epoch": 0.6107491856677525, | |
| "step": 1500 | |
| }, | |
| { | |
| "loss": 0.1086, | |
| "grad_norm": 3.556838274002075, | |
| "learning_rate": 7.014056224899599e-06, | |
| "epoch": 0.6148208469055375, | |
| "step": 1510 | |
| }, | |
| { | |
| "loss": 0.107, | |
| "grad_norm": 5.520222187042236, | |
| "learning_rate": 6.993975903614459e-06, | |
| "epoch": 0.6188925081433225, | |
| "step": 1520 | |
| }, | |
| { | |
| "loss": 0.1101, | |
| "grad_norm": 4.722973823547363, | |
| "learning_rate": 6.973895582329318e-06, | |
| "epoch": 0.6229641693811075, | |
| "step": 1530 | |
| }, | |
| { | |
| "loss": 0.1028, | |
| "grad_norm": 3.2113094329833984, | |
| "learning_rate": 6.9538152610441774e-06, | |
| "epoch": 0.6270358306188925, | |
| "step": 1540 | |
| }, | |
| { | |
| "loss": 0.1281, | |
| "grad_norm": 4.3559651374816895, | |
| "learning_rate": 6.933734939759037e-06, | |
| "epoch": 0.6311074918566775, | |
| "step": 1550 | |
| }, | |
| { | |
| "loss": 0.1379, | |
| "grad_norm": 4.843337059020996, | |
| "learning_rate": 6.913654618473896e-06, | |
| "epoch": 0.6351791530944625, | |
| "step": 1560 | |
| }, | |
| { | |
| "loss": 0.1233, | |
| "grad_norm": 4.652430057525635, | |
| "learning_rate": 6.893574297188755e-06, | |
| "epoch": 0.6392508143322475, | |
| "step": 1570 | |
| }, | |
| { | |
| "loss": 0.1161, | |
| "grad_norm": 2.430050849914551, | |
| "learning_rate": 6.873493975903615e-06, | |
| "epoch": 0.6433224755700325, | |
| "step": 1580 | |
| }, | |
| { | |
| "loss": 0.1028, | |
| "grad_norm": 2.606083631515503, | |
| "learning_rate": 6.853413654618475e-06, | |
| "epoch": 0.6473941368078175, | |
| "step": 1590 | |
| }, | |
| { | |
| "loss": 0.1059, | |
| "grad_norm": 4.149251461029053, | |
| "learning_rate": 6.833333333333334e-06, | |
| "epoch": 0.6514657980456026, | |
| "step": 1600 | |
| }, | |
| { | |
| "loss": 0.1217, | |
| "grad_norm": 5.203061580657959, | |
| "learning_rate": 6.8132530120481935e-06, | |
| "epoch": 0.6555374592833876, | |
| "step": 1610 | |
| }, | |
| { | |
| "loss": 0.1216, | |
| "grad_norm": 3.6398723125457764, | |
| "learning_rate": 6.7931726907630524e-06, | |
| "epoch": 0.6596091205211726, | |
| "step": 1620 | |
| }, | |
| { | |
| "loss": 0.1153, | |
| "grad_norm": 2.7058064937591553, | |
| "learning_rate": 6.773092369477913e-06, | |
| "epoch": 0.6636807817589576, | |
| "step": 1630 | |
| }, | |
| { | |
| "loss": 0.1274, | |
| "grad_norm": 4.477231979370117, | |
| "learning_rate": 6.753012048192772e-06, | |
| "epoch": 0.6677524429967426, | |
| "step": 1640 | |
| }, | |
| { | |
| "loss": 0.0873, | |
| "grad_norm": 4.188817977905273, | |
| "learning_rate": 6.732931726907631e-06, | |
| "epoch": 0.6718241042345277, | |
| "step": 1650 | |
| }, | |
| { | |
| "loss": 0.111, | |
| "grad_norm": 4.296275615692139, | |
| "learning_rate": 6.71285140562249e-06, | |
| "epoch": 0.6758957654723127, | |
| "step": 1660 | |
| }, | |
| { | |
| "loss": 0.1154, | |
| "grad_norm": 3.5604984760284424, | |
| "learning_rate": 6.69277108433735e-06, | |
| "epoch": 0.6799674267100977, | |
| "step": 1670 | |
| }, | |
| { | |
| "loss": 0.1308, | |
| "grad_norm": 5.463046550750732, | |
| "learning_rate": 6.6726907630522096e-06, | |
| "epoch": 0.6840390879478827, | |
| "step": 1680 | |
| }, | |
| { | |
| "loss": 0.1079, | |
| "grad_norm": 3.9705495834350586, | |
| "learning_rate": 6.652610441767069e-06, | |
| "epoch": 0.6881107491856677, | |
| "step": 1690 | |
| }, | |
| { | |
| "loss": 0.1159, | |
| "grad_norm": 2.7398242950439453, | |
| "learning_rate": 6.632530120481928e-06, | |
| "epoch": 0.6921824104234527, | |
| "step": 1700 | |
| }, | |
| { | |
| "loss": 0.1227, | |
| "grad_norm": 2.828465700149536, | |
| "learning_rate": 6.612449799196787e-06, | |
| "epoch": 0.6962540716612378, | |
| "step": 1710 | |
| }, | |
| { | |
| "loss": 0.0939, | |
| "grad_norm": 3.963291883468628, | |
| "learning_rate": 6.592369477911648e-06, | |
| "epoch": 0.7003257328990228, | |
| "step": 1720 | |
| }, | |
| { | |
| "loss": 0.1035, | |
| "grad_norm": 2.91257643699646, | |
| "learning_rate": 6.572289156626507e-06, | |
| "epoch": 0.7043973941368078, | |
| "step": 1730 | |
| }, | |
| { | |
| "loss": 0.1236, | |
| "grad_norm": 5.31588077545166, | |
| "learning_rate": 6.552208835341366e-06, | |
| "epoch": 0.7084690553745928, | |
| "step": 1740 | |
| }, | |
| { | |
| "loss": 0.1144, | |
| "grad_norm": 3.4339985847473145, | |
| "learning_rate": 6.532128514056225e-06, | |
| "epoch": 0.7125407166123778, | |
| "step": 1750 | |
| }, | |
| { | |
| "loss": 0.1034, | |
| "grad_norm": 4.348824501037598, | |
| "learning_rate": 6.5120481927710846e-06, | |
| "epoch": 0.7166123778501629, | |
| "step": 1760 | |
| }, | |
| { | |
| "loss": 0.0964, | |
| "grad_norm": 5.164637565612793, | |
| "learning_rate": 6.491967871485944e-06, | |
| "epoch": 0.7206840390879479, | |
| "step": 1770 | |
| }, | |
| { | |
| "loss": 0.1253, | |
| "grad_norm": 5.915904998779297, | |
| "learning_rate": 6.471887550200804e-06, | |
| "epoch": 0.7247557003257329, | |
| "step": 1780 | |
| }, | |
| { | |
| "loss": 0.1166, | |
| "grad_norm": 3.2082207202911377, | |
| "learning_rate": 6.451807228915663e-06, | |
| "epoch": 0.7288273615635179, | |
| "step": 1790 | |
| }, | |
| { | |
| "loss": 0.102, | |
| "grad_norm": 3.8156161308288574, | |
| "learning_rate": 6.431726907630522e-06, | |
| "epoch": 0.7328990228013029, | |
| "step": 1800 | |
| }, | |
| { | |
| "loss": 0.1049, | |
| "grad_norm": 3.3919761180877686, | |
| "learning_rate": 6.411646586345383e-06, | |
| "epoch": 0.7369706840390879, | |
| "step": 1810 | |
| }, | |
| { | |
| "loss": 0.1091, | |
| "grad_norm": 3.612900733947754, | |
| "learning_rate": 6.391566265060242e-06, | |
| "epoch": 0.741042345276873, | |
| "step": 1820 | |
| }, | |
| { | |
| "loss": 0.1013, | |
| "grad_norm": 2.8713858127593994, | |
| "learning_rate": 6.371485943775101e-06, | |
| "epoch": 0.745114006514658, | |
| "step": 1830 | |
| }, | |
| { | |
| "loss": 0.0987, | |
| "grad_norm": 3.605447292327881, | |
| "learning_rate": 6.3514056224899604e-06, | |
| "epoch": 0.749185667752443, | |
| "step": 1840 | |
| }, | |
| { | |
| "loss": 0.1187, | |
| "grad_norm": 4.194695949554443, | |
| "learning_rate": 6.331325301204819e-06, | |
| "epoch": 0.753257328990228, | |
| "step": 1850 | |
| }, | |
| { | |
| "loss": 0.1208, | |
| "grad_norm": 4.349564552307129, | |
| "learning_rate": 6.311244979919679e-06, | |
| "epoch": 0.757328990228013, | |
| "step": 1860 | |
| }, | |
| { | |
| "loss": 0.1074, | |
| "grad_norm": 4.614382743835449, | |
| "learning_rate": 6.291164658634539e-06, | |
| "epoch": 0.761400651465798, | |
| "step": 1870 | |
| }, | |
| { | |
| "loss": 0.0874, | |
| "grad_norm": 2.872790575027466, | |
| "learning_rate": 6.271084337349398e-06, | |
| "epoch": 0.7654723127035831, | |
| "step": 1880 | |
| }, | |
| { | |
| "loss": 0.1025, | |
| "grad_norm": 3.846503734588623, | |
| "learning_rate": 6.251004016064257e-06, | |
| "epoch": 0.7695439739413681, | |
| "step": 1890 | |
| }, | |
| { | |
| "loss": 0.1106, | |
| "grad_norm": 5.955644607543945, | |
| "learning_rate": 6.2309236947791175e-06, | |
| "epoch": 0.7736156351791531, | |
| "step": 1900 | |
| }, | |
| { | |
| "loss": 0.1037, | |
| "grad_norm": 4.284672260284424, | |
| "learning_rate": 6.2108433734939765e-06, | |
| "epoch": 0.7776872964169381, | |
| "step": 1910 | |
| }, | |
| { | |
| "loss": 0.0981, | |
| "grad_norm": 3.2877254486083984, | |
| "learning_rate": 6.1907630522088354e-06, | |
| "epoch": 0.7817589576547231, | |
| "step": 1920 | |
| }, | |
| { | |
| "loss": 0.1143, | |
| "grad_norm": 4.010598182678223, | |
| "learning_rate": 6.170682730923695e-06, | |
| "epoch": 0.7858306188925082, | |
| "step": 1930 | |
| }, | |
| { | |
| "loss": 0.1191, | |
| "grad_norm": 3.8233859539031982, | |
| "learning_rate": 6.150602409638555e-06, | |
| "epoch": 0.7899022801302932, | |
| "step": 1940 | |
| }, | |
| { | |
| "loss": 0.1117, | |
| "grad_norm": 4.0363054275512695, | |
| "learning_rate": 6.130522088353415e-06, | |
| "epoch": 0.7939739413680782, | |
| "step": 1950 | |
| }, | |
| { | |
| "loss": 0.1119, | |
| "grad_norm": 2.8114421367645264, | |
| "learning_rate": 6.110441767068274e-06, | |
| "epoch": 0.7980456026058632, | |
| "step": 1960 | |
| }, | |
| { | |
| "loss": 0.0929, | |
| "grad_norm": 4.236625671386719, | |
| "learning_rate": 6.090361445783133e-06, | |
| "epoch": 0.8021172638436482, | |
| "step": 1970 | |
| }, | |
| { | |
| "loss": 0.0979, | |
| "grad_norm": 3.6054937839508057, | |
| "learning_rate": 6.070281124497992e-06, | |
| "epoch": 0.8061889250814332, | |
| "step": 1980 | |
| }, | |
| { | |
| "loss": 0.0991, | |
| "grad_norm": 3.8201963901519775, | |
| "learning_rate": 6.050200803212852e-06, | |
| "epoch": 0.8102605863192183, | |
| "step": 1990 | |
| }, | |
| { | |
| "loss": 0.0958, | |
| "grad_norm": 3.7878975868225098, | |
| "learning_rate": 6.030120481927711e-06, | |
| "epoch": 0.8143322475570033, | |
| "step": 2000 | |
| }, | |
| { | |
| "eval_loss": 0.10570257157087326, | |
| "eval_wer": 0.47717593425709315, | |
| "eval_runtime": 1094.9352, | |
| "eval_samples_per_second": 0.997, | |
| "eval_steps_per_second": 0.125, | |
| "epoch": 0.8143322475570033, | |
| "step": 2000 | |
| }, | |
| { | |
| "loss": 0.1005, | |
| "grad_norm": 4.2161993980407715, | |
| "learning_rate": 6.010040160642571e-06, | |
| "epoch": 0.8184039087947883, | |
| "step": 2010 | |
| }, | |
| { | |
| "loss": 0.1103, | |
| "grad_norm": 3.233527660369873, | |
| "learning_rate": 5.98995983935743e-06, | |
| "epoch": 0.8224755700325733, | |
| "step": 2020 | |
| }, | |
| { | |
| "loss": 0.112, | |
| "grad_norm": 5.469659805297852, | |
| "learning_rate": 5.96987951807229e-06, | |
| "epoch": 0.8265472312703583, | |
| "step": 2030 | |
| }, | |
| { | |
| "loss": 0.1129, | |
| "grad_norm": 3.275724172592163, | |
| "learning_rate": 5.94979919678715e-06, | |
| "epoch": 0.8306188925081434, | |
| "step": 2040 | |
| }, | |
| { | |
| "loss": 0.1062, | |
| "grad_norm": 2.848862886428833, | |
| "learning_rate": 5.929718875502009e-06, | |
| "epoch": 0.8346905537459284, | |
| "step": 2050 | |
| }, | |
| { | |
| "loss": 0.0921, | |
| "grad_norm": 2.5302412509918213, | |
| "learning_rate": 5.9096385542168676e-06, | |
| "epoch": 0.8387622149837134, | |
| "step": 2060 | |
| }, | |
| { | |
| "loss": 0.1211, | |
| "grad_norm": 3.372823476791382, | |
| "learning_rate": 5.8895582329317265e-06, | |
| "epoch": 0.8428338762214984, | |
| "step": 2070 | |
| }, | |
| { | |
| "loss": 0.0983, | |
| "grad_norm": 2.7902920246124268, | |
| "learning_rate": 5.869477911646587e-06, | |
| "epoch": 0.8469055374592834, | |
| "step": 2080 | |
| }, | |
| { | |
| "loss": 0.122, | |
| "grad_norm": 4.941180229187012, | |
| "learning_rate": 5.849397590361446e-06, | |
| "epoch": 0.8509771986970684, | |
| "step": 2090 | |
| }, | |
| { | |
| "loss": 0.1019, | |
| "grad_norm": 5.100167274475098, | |
| "learning_rate": 5.829317269076306e-06, | |
| "epoch": 0.8550488599348535, | |
| "step": 2100 | |
| }, | |
| { | |
| "loss": 0.111, | |
| "grad_norm": 3.684011936187744, | |
| "learning_rate": 5.809236947791165e-06, | |
| "epoch": 0.8591205211726385, | |
| "step": 2110 | |
| }, | |
| { | |
| "loss": 0.0952, | |
| "grad_norm": 3.7681939601898193, | |
| "learning_rate": 5.789156626506025e-06, | |
| "epoch": 0.8631921824104235, | |
| "step": 2120 | |
| }, | |
| { | |
| "loss": 0.1263, | |
| "grad_norm": 3.5337135791778564, | |
| "learning_rate": 5.7690763052208845e-06, | |
| "epoch": 0.8672638436482085, | |
| "step": 2130 | |
| }, | |
| { | |
| "loss": 0.098, | |
| "grad_norm": 3.6547176837921143, | |
| "learning_rate": 5.748995983935743e-06, | |
| "epoch": 0.8713355048859935, | |
| "step": 2140 | |
| }, | |
| { | |
| "loss": 0.0995, | |
| "grad_norm": 3.4289116859436035, | |
| "learning_rate": 5.728915662650602e-06, | |
| "epoch": 0.8754071661237784, | |
| "step": 2150 | |
| }, | |
| { | |
| "loss": 0.1106, | |
| "grad_norm": 2.800053596496582, | |
| "learning_rate": 5.708835341365462e-06, | |
| "epoch": 0.8794788273615635, | |
| "step": 2160 | |
| }, | |
| { | |
| "loss": 0.1104, | |
| "grad_norm": 3.710017204284668, | |
| "learning_rate": 5.688755020080322e-06, | |
| "epoch": 0.8835504885993485, | |
| "step": 2170 | |
| }, | |
| { | |
| "loss": 0.1123, | |
| "grad_norm": 3.7521555423736572, | |
| "learning_rate": 5.668674698795181e-06, | |
| "epoch": 0.8876221498371335, | |
| "step": 2180 | |
| }, | |
| { | |
| "loss": 0.106, | |
| "grad_norm": 2.9758145809173584, | |
| "learning_rate": 5.648594377510041e-06, | |
| "epoch": 0.8916938110749185, | |
| "step": 2190 | |
| }, | |
| { | |
| "loss": 0.1039, | |
| "grad_norm": 4.036179542541504, | |
| "learning_rate": 5.6285140562249e-06, | |
| "epoch": 0.8957654723127035, | |
| "step": 2200 | |
| }, | |
| { | |
| "loss": 0.1057, | |
| "grad_norm": 6.8129777908325195, | |
| "learning_rate": 5.60843373493976e-06, | |
| "epoch": 0.8998371335504886, | |
| "step": 2210 | |
| }, | |
| { | |
| "loss": 0.1143, | |
| "grad_norm": 3.602363109588623, | |
| "learning_rate": 5.588353413654619e-06, | |
| "epoch": 0.9039087947882736, | |
| "step": 2220 | |
| }, | |
| { | |
| "loss": 0.0989, | |
| "grad_norm": 3.6404459476470947, | |
| "learning_rate": 5.568273092369478e-06, | |
| "epoch": 0.9079804560260586, | |
| "step": 2230 | |
| }, | |
| { | |
| "loss": 0.1005, | |
| "grad_norm": 2.946787118911743, | |
| "learning_rate": 5.548192771084337e-06, | |
| "epoch": 0.9120521172638436, | |
| "step": 2240 | |
| }, | |
| { | |
| "loss": 0.109, | |
| "grad_norm": 5.762280464172363, | |
| "learning_rate": 5.528112449799197e-06, | |
| "epoch": 0.9161237785016286, | |
| "step": 2250 | |
| }, | |
| { | |
| "loss": 0.0979, | |
| "grad_norm": 2.7821807861328125, | |
| "learning_rate": 5.508032128514057e-06, | |
| "epoch": 0.9201954397394136, | |
| "step": 2260 | |
| }, | |
| { | |
| "loss": 0.1247, | |
| "grad_norm": 3.976362943649292, | |
| "learning_rate": 5.487951807228917e-06, | |
| "epoch": 0.9242671009771987, | |
| "step": 2270 | |
| }, | |
| { | |
| "loss": 0.1033, | |
| "grad_norm": 3.5761988162994385, | |
| "learning_rate": 5.4678714859437755e-06, | |
| "epoch": 0.9283387622149837, | |
| "step": 2280 | |
| }, | |
| { | |
| "loss": 0.0961, | |
| "grad_norm": 3.7126190662384033, | |
| "learning_rate": 5.4477911646586345e-06, | |
| "epoch": 0.9324104234527687, | |
| "step": 2290 | |
| }, | |
| { | |
| "loss": 0.0914, | |
| "grad_norm": 3.133615016937256, | |
| "learning_rate": 5.427710843373495e-06, | |
| "epoch": 0.9364820846905537, | |
| "step": 2300 | |
| }, | |
| { | |
| "loss": 0.1067, | |
| "grad_norm": 2.7286133766174316, | |
| "learning_rate": 5.407630522088354e-06, | |
| "epoch": 0.9405537459283387, | |
| "step": 2310 | |
| }, | |
| { | |
| "loss": 0.1118, | |
| "grad_norm": 2.833508253097534, | |
| "learning_rate": 5.387550200803213e-06, | |
| "epoch": 0.9446254071661238, | |
| "step": 2320 | |
| }, | |
| { | |
| "loss": 0.0912, | |
| "grad_norm": 2.9205315113067627, | |
| "learning_rate": 5.367469879518072e-06, | |
| "epoch": 0.9486970684039088, | |
| "step": 2330 | |
| }, | |
| { | |
| "loss": 0.098, | |
| "grad_norm": 4.562292575836182, | |
| "learning_rate": 5.347389558232933e-06, | |
| "epoch": 0.9527687296416938, | |
| "step": 2340 | |
| }, | |
| { | |
| "loss": 0.1012, | |
| "grad_norm": 5.536325931549072, | |
| "learning_rate": 5.327309236947792e-06, | |
| "epoch": 0.9568403908794788, | |
| "step": 2350 | |
| }, | |
| { | |
| "loss": 0.0906, | |
| "grad_norm": 4.97511100769043, | |
| "learning_rate": 5.307228915662651e-06, | |
| "epoch": 0.9609120521172638, | |
| "step": 2360 | |
| }, | |
| { | |
| "loss": 0.1094, | |
| "grad_norm": 3.348374843597412, | |
| "learning_rate": 5.28714859437751e-06, | |
| "epoch": 0.9649837133550488, | |
| "step": 2370 | |
| }, | |
| { | |
| "loss": 0.0958, | |
| "grad_norm": 3.38018536567688, | |
| "learning_rate": 5.267068273092369e-06, | |
| "epoch": 0.9690553745928339, | |
| "step": 2380 | |
| }, | |
| { | |
| "loss": 0.107, | |
| "grad_norm": 3.7269461154937744, | |
| "learning_rate": 5.24698795180723e-06, | |
| "epoch": 0.9731270358306189, | |
| "step": 2390 | |
| }, | |
| { | |
| "loss": 0.1018, | |
| "grad_norm": 3.6639297008514404, | |
| "learning_rate": 5.226907630522089e-06, | |
| "epoch": 0.9771986970684039, | |
| "step": 2400 | |
| }, | |
| { | |
| "loss": 0.1114, | |
| "grad_norm": 3.246769666671753, | |
| "learning_rate": 5.206827309236948e-06, | |
| "epoch": 0.9812703583061889, | |
| "step": 2410 | |
| }, | |
| { | |
| "loss": 0.0945, | |
| "grad_norm": 2.663567066192627, | |
| "learning_rate": 5.186746987951808e-06, | |
| "epoch": 0.9853420195439739, | |
| "step": 2420 | |
| }, | |
| { | |
| "loss": 0.1057, | |
| "grad_norm": 2.9218430519104004, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "epoch": 0.989413680781759, | |
| "step": 2430 | |
| }, | |
| { | |
| "loss": 0.0881, | |
| "grad_norm": 3.1941118240356445, | |
| "learning_rate": 5.146586345381526e-06, | |
| "epoch": 0.993485342019544, | |
| "step": 2440 | |
| }, | |
| { | |
| "loss": 0.1024, | |
| "grad_norm": 3.0851552486419678, | |
| "learning_rate": 5.126506024096386e-06, | |
| "epoch": 0.997557003257329, | |
| "step": 2450 | |
| }, | |
| { | |
| "loss": 0.0963, | |
| "grad_norm": 2.6105995178222656, | |
| "learning_rate": 5.106425702811245e-06, | |
| "epoch": 1.001628664495114, | |
| "step": 2460 | |
| }, | |
| { | |
| "loss": 0.0751, | |
| "grad_norm": 2.5004005432128906, | |
| "learning_rate": 5.086345381526104e-06, | |
| "epoch": 1.005700325732899, | |
| "step": 2470 | |
| }, | |
| { | |
| "loss": 0.0731, | |
| "grad_norm": 2.020967483520508, | |
| "learning_rate": 5.066265060240965e-06, | |
| "epoch": 1.009771986970684, | |
| "step": 2480 | |
| }, | |
| { | |
| "loss": 0.0576, | |
| "grad_norm": 2.1105527877807617, | |
| "learning_rate": 5.046184738955824e-06, | |
| "epoch": 1.013843648208469, | |
| "step": 2490 | |
| }, | |
| { | |
| "loss": 0.0679, | |
| "grad_norm": 3.5036275386810303, | |
| "learning_rate": 5.026104417670683e-06, | |
| "epoch": 1.017915309446254, | |
| "step": 2500 | |
| }, | |
| { | |
| "eval_loss": 0.09774903953075409, | |
| "eval_wer": 0.37232319596904806, | |
| "eval_runtime": 1091.5675, | |
| "eval_samples_per_second": 1.0, | |
| "eval_steps_per_second": 0.126, | |
| "epoch": 1.017915309446254, | |
| "step": 2500 | |
| }, | |
| { | |
| "loss": 0.0644, | |
| "grad_norm": 3.5613856315612793, | |
| "learning_rate": 5.0060240963855425e-06, | |
| "epoch": 1.021986970684039, | |
| "step": 2510 | |
| }, | |
| { | |
| "loss": 0.0631, | |
| "grad_norm": 3.0361297130584717, | |
| "learning_rate": 4.985943775100402e-06, | |
| "epoch": 1.0260586319218241, | |
| "step": 2520 | |
| }, | |
| { | |
| "loss": 0.0793, | |
| "grad_norm": 1.6461753845214844, | |
| "learning_rate": 4.965863453815262e-06, | |
| "epoch": 1.030130293159609, | |
| "step": 2530 | |
| }, | |
| { | |
| "loss": 0.0597, | |
| "grad_norm": 3.5274806022644043, | |
| "learning_rate": 4.945783132530121e-06, | |
| "epoch": 1.0342019543973942, | |
| "step": 2540 | |
| }, | |
| { | |
| "loss": 0.079, | |
| "grad_norm": 3.901984691619873, | |
| "learning_rate": 4.92570281124498e-06, | |
| "epoch": 1.038273615635179, | |
| "step": 2550 | |
| }, | |
| { | |
| "loss": 0.0724, | |
| "grad_norm": 2.7136402130126953, | |
| "learning_rate": 4.90562248995984e-06, | |
| "epoch": 1.0423452768729642, | |
| "step": 2560 | |
| }, | |
| { | |
| "loss": 0.08, | |
| "grad_norm": 3.487898826599121, | |
| "learning_rate": 4.885542168674699e-06, | |
| "epoch": 1.046416938110749, | |
| "step": 2570 | |
| }, | |
| { | |
| "loss": 0.0631, | |
| "grad_norm": 3.0557966232299805, | |
| "learning_rate": 4.8654618473895585e-06, | |
| "epoch": 1.0504885993485342, | |
| "step": 2580 | |
| }, | |
| { | |
| "loss": 0.0869, | |
| "grad_norm": 5.151049613952637, | |
| "learning_rate": 4.845381526104418e-06, | |
| "epoch": 1.0545602605863191, | |
| "step": 2590 | |
| }, | |
| { | |
| "loss": 0.0715, | |
| "grad_norm": 1.9614373445510864, | |
| "learning_rate": 4.825301204819277e-06, | |
| "epoch": 1.0586319218241043, | |
| "step": 2600 | |
| }, | |
| { | |
| "loss": 0.0629, | |
| "grad_norm": 1.9923406839370728, | |
| "learning_rate": 4.805220883534137e-06, | |
| "epoch": 1.0627035830618892, | |
| "step": 2610 | |
| }, | |
| { | |
| "loss": 0.0734, | |
| "grad_norm": 3.557880401611328, | |
| "learning_rate": 4.785140562248997e-06, | |
| "epoch": 1.0667752442996743, | |
| "step": 2620 | |
| }, | |
| { | |
| "loss": 0.0582, | |
| "grad_norm": 2.4320619106292725, | |
| "learning_rate": 4.765060240963856e-06, | |
| "epoch": 1.0708469055374592, | |
| "step": 2630 | |
| }, | |
| { | |
| "loss": 0.0708, | |
| "grad_norm": 2.9941861629486084, | |
| "learning_rate": 4.744979919678716e-06, | |
| "epoch": 1.0749185667752443, | |
| "step": 2640 | |
| }, | |
| { | |
| "loss": 0.0649, | |
| "grad_norm": 5.872101306915283, | |
| "learning_rate": 4.724899598393575e-06, | |
| "epoch": 1.0789902280130292, | |
| "step": 2650 | |
| }, | |
| { | |
| "loss": 0.0558, | |
| "grad_norm": 1.6835378408432007, | |
| "learning_rate": 4.7048192771084335e-06, | |
| "epoch": 1.0830618892508144, | |
| "step": 2660 | |
| }, | |
| { | |
| "loss": 0.0646, | |
| "grad_norm": 1.938002586364746, | |
| "learning_rate": 4.684738955823293e-06, | |
| "epoch": 1.0871335504885993, | |
| "step": 2670 | |
| }, | |
| { | |
| "loss": 0.0603, | |
| "grad_norm": 2.7401139736175537, | |
| "learning_rate": 4.664658634538153e-06, | |
| "epoch": 1.0912052117263844, | |
| "step": 2680 | |
| }, | |
| { | |
| "loss": 0.0874, | |
| "grad_norm": 5.646732330322266, | |
| "learning_rate": 4.644578313253013e-06, | |
| "epoch": 1.0952768729641693, | |
| "step": 2690 | |
| }, | |
| { | |
| "loss": 0.0679, | |
| "grad_norm": 3.446324110031128, | |
| "learning_rate": 4.624497991967872e-06, | |
| "epoch": 1.0993485342019544, | |
| "step": 2700 | |
| }, | |
| { | |
| "loss": 0.0724, | |
| "grad_norm": 3.958233594894409, | |
| "learning_rate": 4.604417670682732e-06, | |
| "epoch": 1.1034201954397393, | |
| "step": 2710 | |
| }, | |
| { | |
| "loss": 0.0712, | |
| "grad_norm": 2.991860866546631, | |
| "learning_rate": 4.584337349397591e-06, | |
| "epoch": 1.1074918566775245, | |
| "step": 2720 | |
| }, | |
| { | |
| "loss": 0.0743, | |
| "grad_norm": 3.0944552421569824, | |
| "learning_rate": 4.5642570281124504e-06, | |
| "epoch": 1.1115635179153094, | |
| "step": 2730 | |
| }, | |
| { | |
| "loss": 0.0689, | |
| "grad_norm": 3.3628759384155273, | |
| "learning_rate": 4.544176706827309e-06, | |
| "epoch": 1.1156351791530945, | |
| "step": 2740 | |
| }, | |
| { | |
| "loss": 0.0656, | |
| "grad_norm": 2.7449183464050293, | |
| "learning_rate": 4.524096385542169e-06, | |
| "epoch": 1.1197068403908794, | |
| "step": 2750 | |
| }, | |
| { | |
| "loss": 0.06, | |
| "grad_norm": 2.636051893234253, | |
| "learning_rate": 4.504016064257028e-06, | |
| "epoch": 1.1237785016286646, | |
| "step": 2760 | |
| }, | |
| { | |
| "loss": 0.0951, | |
| "grad_norm": 3.0840206146240234, | |
| "learning_rate": 4.483935742971888e-06, | |
| "epoch": 1.1278501628664495, | |
| "step": 2770 | |
| }, | |
| { | |
| "loss": 0.0577, | |
| "grad_norm": 2.543952465057373, | |
| "learning_rate": 4.463855421686748e-06, | |
| "epoch": 1.1319218241042346, | |
| "step": 2780 | |
| }, | |
| { | |
| "loss": 0.081, | |
| "grad_norm": 3.951514482498169, | |
| "learning_rate": 4.443775100401607e-06, | |
| "epoch": 1.1359934853420195, | |
| "step": 2790 | |
| }, | |
| { | |
| "loss": 0.0704, | |
| "grad_norm": 5.900193691253662, | |
| "learning_rate": 4.4236947791164665e-06, | |
| "epoch": 1.1400651465798046, | |
| "step": 2800 | |
| }, | |
| { | |
| "loss": 0.0583, | |
| "grad_norm": 2.5971202850341797, | |
| "learning_rate": 4.4036144578313255e-06, | |
| "epoch": 1.1441368078175895, | |
| "step": 2810 | |
| }, | |
| { | |
| "loss": 0.0693, | |
| "grad_norm": 3.4016895294189453, | |
| "learning_rate": 4.383534136546185e-06, | |
| "epoch": 1.1482084690553747, | |
| "step": 2820 | |
| }, | |
| { | |
| "loss": 0.0689, | |
| "grad_norm": 2.5758798122406006, | |
| "learning_rate": 4.363453815261044e-06, | |
| "epoch": 1.1522801302931596, | |
| "step": 2830 | |
| }, | |
| { | |
| "loss": 0.0651, | |
| "grad_norm": 3.1408371925354004, | |
| "learning_rate": 4.343373493975904e-06, | |
| "epoch": 1.1563517915309447, | |
| "step": 2840 | |
| }, | |
| { | |
| "loss": 0.0523, | |
| "grad_norm": 2.7517447471618652, | |
| "learning_rate": 4.323293172690764e-06, | |
| "epoch": 1.1604234527687296, | |
| "step": 2850 | |
| }, | |
| { | |
| "loss": 0.0694, | |
| "grad_norm": 2.3461155891418457, | |
| "learning_rate": 4.303212851405623e-06, | |
| "epoch": 1.1644951140065147, | |
| "step": 2860 | |
| }, | |
| { | |
| "loss": 0.0729, | |
| "grad_norm": 1.9746013879776, | |
| "learning_rate": 4.2831325301204826e-06, | |
| "epoch": 1.1685667752442996, | |
| "step": 2870 | |
| }, | |
| { | |
| "loss": 0.0691, | |
| "grad_norm": 3.0223231315612793, | |
| "learning_rate": 4.2630522088353415e-06, | |
| "epoch": 1.1726384364820848, | |
| "step": 2880 | |
| }, | |
| { | |
| "loss": 0.0797, | |
| "grad_norm": 3.556380271911621, | |
| "learning_rate": 4.242971887550201e-06, | |
| "epoch": 1.1767100977198697, | |
| "step": 2890 | |
| }, | |
| { | |
| "loss": 0.0655, | |
| "grad_norm": 3.123739719390869, | |
| "learning_rate": 4.22289156626506e-06, | |
| "epoch": 1.1807817589576548, | |
| "step": 2900 | |
| }, | |
| { | |
| "loss": 0.0627, | |
| "grad_norm": 3.2453410625457764, | |
| "learning_rate": 4.20281124497992e-06, | |
| "epoch": 1.1848534201954397, | |
| "step": 2910 | |
| }, | |
| { | |
| "loss": 0.0694, | |
| "grad_norm": 3.2520205974578857, | |
| "learning_rate": 4.182730923694779e-06, | |
| "epoch": 1.1889250814332248, | |
| "step": 2920 | |
| }, | |
| { | |
| "loss": 0.0636, | |
| "grad_norm": 2.269685745239258, | |
| "learning_rate": 4.162650602409639e-06, | |
| "epoch": 1.1929967426710097, | |
| "step": 2930 | |
| }, | |
| { | |
| "loss": 0.0743, | |
| "grad_norm": 3.3884530067443848, | |
| "learning_rate": 4.142570281124499e-06, | |
| "epoch": 1.1970684039087949, | |
| "step": 2940 | |
| }, | |
| { | |
| "loss": 0.0643, | |
| "grad_norm": 2.7258195877075195, | |
| "learning_rate": 4.122489959839358e-06, | |
| "epoch": 1.2011400651465798, | |
| "step": 2950 | |
| }, | |
| { | |
| "loss": 0.08, | |
| "grad_norm": 4.698688507080078, | |
| "learning_rate": 4.102409638554217e-06, | |
| "epoch": 1.205211726384365, | |
| "step": 2960 | |
| }, | |
| { | |
| "loss": 0.0656, | |
| "grad_norm": 3.4465513229370117, | |
| "learning_rate": 4.082329317269076e-06, | |
| "epoch": 1.2092833876221498, | |
| "step": 2970 | |
| }, | |
| { | |
| "loss": 0.0711, | |
| "grad_norm": 3.1135551929473877, | |
| "learning_rate": 4.062248995983936e-06, | |
| "epoch": 1.213355048859935, | |
| "step": 2980 | |
| }, | |
| { | |
| "loss": 0.0848, | |
| "grad_norm": 2.4550602436065674, | |
| "learning_rate": 4.042168674698795e-06, | |
| "epoch": 1.2174267100977199, | |
| "step": 2990 | |
| }, | |
| { | |
| "loss": 0.067, | |
| "grad_norm": 2.5689499378204346, | |
| "learning_rate": 4.022088353413655e-06, | |
| "epoch": 1.221498371335505, | |
| "step": 3000 | |
| }, | |
| { | |
| "eval_loss": 0.09280993044376373, | |
| "eval_wer": 0.46721852318397217, | |
| "eval_runtime": 1096.286, | |
| "eval_samples_per_second": 0.996, | |
| "eval_steps_per_second": 0.125, | |
| "epoch": 1.221498371335505, | |
| "step": 3000 | |
| }, | |
| { | |
| "loss": 0.0726, | |
| "grad_norm": 2.809565544128418, | |
| "learning_rate": 4.002008032128515e-06, | |
| "epoch": 1.22557003257329, | |
| "step": 3010 | |
| }, | |
| { | |
| "loss": 0.0748, | |
| "grad_norm": 5.701145648956299, | |
| "learning_rate": 3.981927710843374e-06, | |
| "epoch": 1.2296416938110748, | |
| "step": 3020 | |
| }, | |
| { | |
| "loss": 0.0646, | |
| "grad_norm": 3.123772144317627, | |
| "learning_rate": 3.9618473895582334e-06, | |
| "epoch": 1.23371335504886, | |
| "step": 3030 | |
| }, | |
| { | |
| "loss": 0.0622, | |
| "grad_norm": 4.213207721710205, | |
| "learning_rate": 3.941767068273093e-06, | |
| "epoch": 1.237785016286645, | |
| "step": 3040 | |
| }, | |
| { | |
| "loss": 0.0608, | |
| "grad_norm": 2.7219290733337402, | |
| "learning_rate": 3.921686746987952e-06, | |
| "epoch": 1.24185667752443, | |
| "step": 3050 | |
| }, | |
| { | |
| "loss": 0.0654, | |
| "grad_norm": 2.914076566696167, | |
| "learning_rate": 3.901606425702811e-06, | |
| "epoch": 1.2459283387622149, | |
| "step": 3060 | |
| }, | |
| { | |
| "loss": 0.0565, | |
| "grad_norm": 2.7983219623565674, | |
| "learning_rate": 3.881526104417671e-06, | |
| "epoch": 1.25, | |
| "step": 3070 | |
| }, | |
| { | |
| "loss": 0.0591, | |
| "grad_norm": 2.7069931030273438, | |
| "learning_rate": 3.86144578313253e-06, | |
| "epoch": 1.2540716612377851, | |
| "step": 3080 | |
| }, | |
| { | |
| "loss": 0.0627, | |
| "grad_norm": 3.389723777770996, | |
| "learning_rate": 3.84136546184739e-06, | |
| "epoch": 1.25814332247557, | |
| "step": 3090 | |
| }, | |
| { | |
| "loss": 0.0705, | |
| "grad_norm": 3.272932529449463, | |
| "learning_rate": 3.8212851405622495e-06, | |
| "epoch": 1.262214983713355, | |
| "step": 3100 | |
| }, | |
| { | |
| "loss": 0.0675, | |
| "grad_norm": 1.9180351495742798, | |
| "learning_rate": 3.801204819277109e-06, | |
| "epoch": 1.26628664495114, | |
| "step": 3110 | |
| }, | |
| { | |
| "loss": 0.0555, | |
| "grad_norm": 2.3821358680725098, | |
| "learning_rate": 3.7811244979919683e-06, | |
| "epoch": 1.2703583061889252, | |
| "step": 3120 | |
| }, | |
| { | |
| "loss": 0.0539, | |
| "grad_norm": 3.1405835151672363, | |
| "learning_rate": 3.761044176706828e-06, | |
| "epoch": 1.27442996742671, | |
| "step": 3130 | |
| }, | |
| { | |
| "loss": 0.0623, | |
| "grad_norm": 2.8730640411376953, | |
| "learning_rate": 3.740963855421687e-06, | |
| "epoch": 1.278501628664495, | |
| "step": 3140 | |
| }, | |
| { | |
| "loss": 0.0681, | |
| "grad_norm": 3.1091349124908447, | |
| "learning_rate": 3.7208835341365464e-06, | |
| "epoch": 1.2825732899022801, | |
| "step": 3150 | |
| }, | |
| { | |
| "loss": 0.0714, | |
| "grad_norm": 2.3350601196289062, | |
| "learning_rate": 3.700803212851406e-06, | |
| "epoch": 1.2866449511400653, | |
| "step": 3160 | |
| }, | |
| { | |
| "loss": 0.0581, | |
| "grad_norm": 2.4262986183166504, | |
| "learning_rate": 3.680722891566265e-06, | |
| "epoch": 1.2907166123778502, | |
| "step": 3170 | |
| }, | |
| { | |
| "loss": 0.0785, | |
| "grad_norm": 2.667527198791504, | |
| "learning_rate": 3.660642570281125e-06, | |
| "epoch": 1.294788273615635, | |
| "step": 3180 | |
| }, | |
| { | |
| "loss": 0.0555, | |
| "grad_norm": 3.6449811458587646, | |
| "learning_rate": 3.640562248995984e-06, | |
| "epoch": 1.2988599348534202, | |
| "step": 3190 | |
| }, | |
| { | |
| "loss": 0.0527, | |
| "grad_norm": 2.295485258102417, | |
| "learning_rate": 3.6204819277108437e-06, | |
| "epoch": 1.3029315960912053, | |
| "step": 3200 | |
| }, | |
| { | |
| "loss": 0.0601, | |
| "grad_norm": 2.301023244857788, | |
| "learning_rate": 3.600401606425703e-06, | |
| "epoch": 1.3070032573289903, | |
| "step": 3210 | |
| }, | |
| { | |
| "loss": 0.0723, | |
| "grad_norm": 2.470319986343384, | |
| "learning_rate": 3.580321285140563e-06, | |
| "epoch": 1.3110749185667752, | |
| "step": 3220 | |
| }, | |
| { | |
| "loss": 0.0742, | |
| "grad_norm": 4.1519083976745605, | |
| "learning_rate": 3.560240963855422e-06, | |
| "epoch": 1.3151465798045603, | |
| "step": 3230 | |
| }, | |
| { | |
| "loss": 0.0569, | |
| "grad_norm": 2.4179742336273193, | |
| "learning_rate": 3.5401606425702816e-06, | |
| "epoch": 1.3192182410423452, | |
| "step": 3240 | |
| }, | |
| { | |
| "loss": 0.0798, | |
| "grad_norm": 3.407655954360962, | |
| "learning_rate": 3.520080321285141e-06, | |
| "epoch": 1.3232899022801303, | |
| "step": 3250 | |
| }, | |
| { | |
| "loss": 0.0548, | |
| "grad_norm": 3.6222007274627686, | |
| "learning_rate": 3.5e-06, | |
| "epoch": 1.3273615635179152, | |
| "step": 3260 | |
| }, | |
| { | |
| "loss": 0.0705, | |
| "grad_norm": 2.585489511489868, | |
| "learning_rate": 3.4799196787148597e-06, | |
| "epoch": 1.3314332247557004, | |
| "step": 3270 | |
| }, | |
| { | |
| "loss": 0.0618, | |
| "grad_norm": 3.0058841705322266, | |
| "learning_rate": 3.459839357429719e-06, | |
| "epoch": 1.3355048859934853, | |
| "step": 3280 | |
| }, | |
| { | |
| "loss": 0.0669, | |
| "grad_norm": 2.6239447593688965, | |
| "learning_rate": 3.439759036144579e-06, | |
| "epoch": 1.3395765472312704, | |
| "step": 3290 | |
| }, | |
| { | |
| "loss": 0.0634, | |
| "grad_norm": 2.1419737339019775, | |
| "learning_rate": 3.419678714859438e-06, | |
| "epoch": 1.3436482084690553, | |
| "step": 3300 | |
| }, | |
| { | |
| "loss": 0.0629, | |
| "grad_norm": 3.5120849609375, | |
| "learning_rate": 3.3995983935742977e-06, | |
| "epoch": 1.3477198697068404, | |
| "step": 3310 | |
| }, | |
| { | |
| "loss": 0.0661, | |
| "grad_norm": 1.8860255479812622, | |
| "learning_rate": 3.379518072289157e-06, | |
| "epoch": 1.3517915309446253, | |
| "step": 3320 | |
| }, | |
| { | |
| "loss": 0.0685, | |
| "grad_norm": 3.449812650680542, | |
| "learning_rate": 3.3594377510040164e-06, | |
| "epoch": 1.3558631921824105, | |
| "step": 3330 | |
| }, | |
| { | |
| "loss": 0.055, | |
| "grad_norm": 3.1233980655670166, | |
| "learning_rate": 3.339357429718876e-06, | |
| "epoch": 1.3599348534201954, | |
| "step": 3340 | |
| }, | |
| { | |
| "loss": 0.0708, | |
| "grad_norm": 2.702009677886963, | |
| "learning_rate": 3.3192771084337348e-06, | |
| "epoch": 1.3640065146579805, | |
| "step": 3350 | |
| }, | |
| { | |
| "loss": 0.061, | |
| "grad_norm": 3.161761522293091, | |
| "learning_rate": 3.2991967871485946e-06, | |
| "epoch": 1.3680781758957654, | |
| "step": 3360 | |
| }, | |
| { | |
| "loss": 0.0667, | |
| "grad_norm": 2.9508256912231445, | |
| "learning_rate": 3.279116465863454e-06, | |
| "epoch": 1.3721498371335505, | |
| "step": 3370 | |
| }, | |
| { | |
| "loss": 0.076, | |
| "grad_norm": 2.473778247833252, | |
| "learning_rate": 3.2590361445783137e-06, | |
| "epoch": 1.3762214983713354, | |
| "step": 3380 | |
| }, | |
| { | |
| "loss": 0.0489, | |
| "grad_norm": 2.781268835067749, | |
| "learning_rate": 3.2389558232931727e-06, | |
| "epoch": 1.3802931596091206, | |
| "step": 3390 | |
| }, | |
| { | |
| "loss": 0.0655, | |
| "grad_norm": 2.8619117736816406, | |
| "learning_rate": 3.2188755020080325e-06, | |
| "epoch": 1.3843648208469055, | |
| "step": 3400 | |
| }, | |
| { | |
| "loss": 0.076, | |
| "grad_norm": 1.4964746236801147, | |
| "learning_rate": 3.198795180722892e-06, | |
| "epoch": 1.3884364820846906, | |
| "step": 3410 | |
| }, | |
| { | |
| "loss": 0.0637, | |
| "grad_norm": 2.054722785949707, | |
| "learning_rate": 3.1787148594377517e-06, | |
| "epoch": 1.3925081433224755, | |
| "step": 3420 | |
| }, | |
| { | |
| "loss": 0.0776, | |
| "grad_norm": 3.710057258605957, | |
| "learning_rate": 3.1586345381526106e-06, | |
| "epoch": 1.3965798045602607, | |
| "step": 3430 | |
| }, | |
| { | |
| "loss": 0.0629, | |
| "grad_norm": 3.1431212425231934, | |
| "learning_rate": 3.13855421686747e-06, | |
| "epoch": 1.4006514657980456, | |
| "step": 3440 | |
| }, | |
| { | |
| "loss": 0.0604, | |
| "grad_norm": 3.1654181480407715, | |
| "learning_rate": 3.11847389558233e-06, | |
| "epoch": 1.4047231270358307, | |
| "step": 3450 | |
| }, | |
| { | |
| "loss": 0.0726, | |
| "grad_norm": 3.2606966495513916, | |
| "learning_rate": 3.0983935742971887e-06, | |
| "epoch": 1.4087947882736156, | |
| "step": 3460 | |
| }, | |
| { | |
| "loss": 0.0668, | |
| "grad_norm": 6.117660045623779, | |
| "learning_rate": 3.0783132530120485e-06, | |
| "epoch": 1.4128664495114007, | |
| "step": 3470 | |
| }, | |
| { | |
| "loss": 0.0807, | |
| "grad_norm": 3.8929762840270996, | |
| "learning_rate": 3.058232931726908e-06, | |
| "epoch": 1.4169381107491856, | |
| "step": 3480 | |
| }, | |
| { | |
| "loss": 0.0664, | |
| "grad_norm": 3.6813697814941406, | |
| "learning_rate": 3.0381526104417673e-06, | |
| "epoch": 1.4210097719869708, | |
| "step": 3490 | |
| }, | |
| { | |
| "loss": 0.0603, | |
| "grad_norm": 2.45691180229187, | |
| "learning_rate": 3.0180722891566267e-06, | |
| "epoch": 1.4250814332247557, | |
| "step": 3500 | |
| }, | |
| { | |
| "eval_loss": 0.08894983679056168, | |
| "eval_wer": 0.6577289904624798, | |
| "eval_runtime": 1113.3082, | |
| "eval_samples_per_second": 0.981, | |
| "eval_steps_per_second": 0.123, | |
| "epoch": 1.4250814332247557, | |
| "step": 3500 | |
| }, | |
| { | |
| "loss": 0.0572, | |
| "grad_norm": 2.4227569103240967, | |
| "learning_rate": 2.9979919678714865e-06, | |
| "epoch": 1.4291530944625408, | |
| "step": 3510 | |
| }, | |
| { | |
| "loss": 0.0784, | |
| "grad_norm": 2.883768320083618, | |
| "learning_rate": 2.9779116465863454e-06, | |
| "epoch": 1.4332247557003257, | |
| "step": 3520 | |
| }, | |
| { | |
| "loss": 0.0677, | |
| "grad_norm": 3.615967273712158, | |
| "learning_rate": 2.9578313253012052e-06, | |
| "epoch": 1.4372964169381108, | |
| "step": 3530 | |
| }, | |
| { | |
| "loss": 0.0658, | |
| "grad_norm": 2.8562071323394775, | |
| "learning_rate": 2.9377510040160646e-06, | |
| "epoch": 1.4413680781758957, | |
| "step": 3540 | |
| }, | |
| { | |
| "loss": 0.0631, | |
| "grad_norm": 2.794185161590576, | |
| "learning_rate": 2.9176706827309236e-06, | |
| "epoch": 1.4454397394136809, | |
| "step": 3550 | |
| }, | |
| { | |
| "loss": 0.0631, | |
| "grad_norm": 5.115529537200928, | |
| "learning_rate": 2.8975903614457834e-06, | |
| "epoch": 1.4495114006514658, | |
| "step": 3560 | |
| }, | |
| { | |
| "loss": 0.0666, | |
| "grad_norm": 3.6309163570404053, | |
| "learning_rate": 2.8775100401606427e-06, | |
| "epoch": 1.4535830618892507, | |
| "step": 3570 | |
| }, | |
| { | |
| "loss": 0.0468, | |
| "grad_norm": 3.5887033939361572, | |
| "learning_rate": 2.8574297188755025e-06, | |
| "epoch": 1.4576547231270358, | |
| "step": 3580 | |
| }, | |
| { | |
| "loss": 0.0574, | |
| "grad_norm": 4.536567211151123, | |
| "learning_rate": 2.8373493975903615e-06, | |
| "epoch": 1.461726384364821, | |
| "step": 3590 | |
| }, | |
| { | |
| "loss": 0.0475, | |
| "grad_norm": 2.421926498413086, | |
| "learning_rate": 2.8172690763052213e-06, | |
| "epoch": 1.4657980456026058, | |
| "step": 3600 | |
| }, | |
| { | |
| "loss": 0.0679, | |
| "grad_norm": 1.6686556339263916, | |
| "learning_rate": 2.7971887550200807e-06, | |
| "epoch": 1.4698697068403908, | |
| "step": 3610 | |
| }, | |
| { | |
| "loss": 0.0678, | |
| "grad_norm": 2.364711046218872, | |
| "learning_rate": 2.77710843373494e-06, | |
| "epoch": 1.4739413680781759, | |
| "step": 3620 | |
| }, | |
| { | |
| "loss": 0.0623, | |
| "grad_norm": 2.1344664096832275, | |
| "learning_rate": 2.7570281124497994e-06, | |
| "epoch": 1.478013029315961, | |
| "step": 3630 | |
| }, | |
| { | |
| "loss": 0.077, | |
| "grad_norm": 2.8033318519592285, | |
| "learning_rate": 2.7369477911646584e-06, | |
| "epoch": 1.482084690553746, | |
| "step": 3640 | |
| }, | |
| { | |
| "loss": 0.053, | |
| "grad_norm": 3.7287914752960205, | |
| "learning_rate": 2.716867469879518e-06, | |
| "epoch": 1.4861563517915308, | |
| "step": 3650 | |
| }, | |
| { | |
| "loss": 0.0616, | |
| "grad_norm": 1.9503307342529297, | |
| "learning_rate": 2.6967871485943776e-06, | |
| "epoch": 1.490228013029316, | |
| "step": 3660 | |
| }, | |
| { | |
| "loss": 0.0632, | |
| "grad_norm": 2.4476473331451416, | |
| "learning_rate": 2.6767068273092374e-06, | |
| "epoch": 1.494299674267101, | |
| "step": 3670 | |
| }, | |
| { | |
| "loss": 0.0561, | |
| "grad_norm": 2.103212833404541, | |
| "learning_rate": 2.6566265060240963e-06, | |
| "epoch": 1.498371335504886, | |
| "step": 3680 | |
| }, | |
| { | |
| "loss": 0.0615, | |
| "grad_norm": 3.7654311656951904, | |
| "learning_rate": 2.636546184738956e-06, | |
| "epoch": 1.502442996742671, | |
| "step": 3690 | |
| }, | |
| { | |
| "loss": 0.0498, | |
| "grad_norm": 2.4322330951690674, | |
| "learning_rate": 2.6164658634538155e-06, | |
| "epoch": 1.506514657980456, | |
| "step": 3700 | |
| }, | |
| { | |
| "loss": 0.0502, | |
| "grad_norm": 5.046417236328125, | |
| "learning_rate": 2.5963855421686753e-06, | |
| "epoch": 1.5105863192182412, | |
| "step": 3710 | |
| }, | |
| { | |
| "loss": 0.0568, | |
| "grad_norm": 2.5868771076202393, | |
| "learning_rate": 2.5763052208835342e-06, | |
| "epoch": 1.514657980456026, | |
| "step": 3720 | |
| }, | |
| { | |
| "loss": 0.0696, | |
| "grad_norm": 4.398423671722412, | |
| "learning_rate": 2.556224899598394e-06, | |
| "epoch": 1.518729641693811, | |
| "step": 3730 | |
| }, | |
| { | |
| "loss": 0.062, | |
| "grad_norm": 3.1776480674743652, | |
| "learning_rate": 2.5361445783132534e-06, | |
| "epoch": 1.522801302931596, | |
| "step": 3740 | |
| }, | |
| { | |
| "loss": 0.0617, | |
| "grad_norm": 3.0078420639038086, | |
| "learning_rate": 2.5160642570281124e-06, | |
| "epoch": 1.5268729641693812, | |
| "step": 3750 | |
| }, | |
| { | |
| "loss": 0.0641, | |
| "grad_norm": 4.2703118324279785, | |
| "learning_rate": 2.495983935742972e-06, | |
| "epoch": 1.5309446254071661, | |
| "step": 3760 | |
| }, | |
| { | |
| "loss": 0.0633, | |
| "grad_norm": 2.4665615558624268, | |
| "learning_rate": 2.4759036144578315e-06, | |
| "epoch": 1.535016286644951, | |
| "step": 3770 | |
| }, | |
| { | |
| "loss": 0.0483, | |
| "grad_norm": 2.7302215099334717, | |
| "learning_rate": 2.455823293172691e-06, | |
| "epoch": 1.5390879478827362, | |
| "step": 3780 | |
| }, | |
| { | |
| "loss": 0.0529, | |
| "grad_norm": 2.41924786567688, | |
| "learning_rate": 2.4357429718875503e-06, | |
| "epoch": 1.5431596091205213, | |
| "step": 3790 | |
| }, | |
| { | |
| "loss": 0.0765, | |
| "grad_norm": 3.919454574584961, | |
| "learning_rate": 2.4156626506024097e-06, | |
| "epoch": 1.5472312703583062, | |
| "step": 3800 | |
| }, | |
| { | |
| "loss": 0.0618, | |
| "grad_norm": 3.1884820461273193, | |
| "learning_rate": 2.395582329317269e-06, | |
| "epoch": 1.5513029315960911, | |
| "step": 3810 | |
| }, | |
| { | |
| "loss": 0.0692, | |
| "grad_norm": 3.261373996734619, | |
| "learning_rate": 2.375502008032129e-06, | |
| "epoch": 1.5553745928338762, | |
| "step": 3820 | |
| }, | |
| { | |
| "loss": 0.0644, | |
| "grad_norm": 3.2010719776153564, | |
| "learning_rate": 2.3554216867469882e-06, | |
| "epoch": 1.5594462540716614, | |
| "step": 3830 | |
| }, | |
| { | |
| "loss": 0.0546, | |
| "grad_norm": 2.0534307956695557, | |
| "learning_rate": 2.3353413654618476e-06, | |
| "epoch": 1.5635179153094463, | |
| "step": 3840 | |
| }, | |
| { | |
| "loss": 0.0525, | |
| "grad_norm": 2.69657301902771, | |
| "learning_rate": 2.315261044176707e-06, | |
| "epoch": 1.5675895765472312, | |
| "step": 3850 | |
| }, | |
| { | |
| "loss": 0.0648, | |
| "grad_norm": 2.6026852130889893, | |
| "learning_rate": 2.2951807228915664e-06, | |
| "epoch": 1.5716612377850163, | |
| "step": 3860 | |
| }, | |
| { | |
| "loss": 0.0623, | |
| "grad_norm": 2.7913529872894287, | |
| "learning_rate": 2.275100401606426e-06, | |
| "epoch": 1.5757328990228014, | |
| "step": 3870 | |
| }, | |
| { | |
| "loss": 0.0754, | |
| "grad_norm": 2.727436065673828, | |
| "learning_rate": 2.2550200803212855e-06, | |
| "epoch": 1.5798045602605864, | |
| "step": 3880 | |
| }, | |
| { | |
| "loss": 0.063, | |
| "grad_norm": 1.8480360507965088, | |
| "learning_rate": 2.2349397590361445e-06, | |
| "epoch": 1.5838762214983713, | |
| "step": 3890 | |
| }, | |
| { | |
| "loss": 0.0522, | |
| "grad_norm": 2.362602710723877, | |
| "learning_rate": 2.2148594377510043e-06, | |
| "epoch": 1.5879478827361564, | |
| "step": 3900 | |
| }, | |
| { | |
| "loss": 0.0672, | |
| "grad_norm": 3.3140482902526855, | |
| "learning_rate": 2.1947791164658637e-06, | |
| "epoch": 1.5920195439739415, | |
| "step": 3910 | |
| }, | |
| { | |
| "loss": 0.0548, | |
| "grad_norm": 3.847317695617676, | |
| "learning_rate": 2.174698795180723e-06, | |
| "epoch": 1.5960912052117264, | |
| "step": 3920 | |
| }, | |
| { | |
| "loss": 0.0605, | |
| "grad_norm": 4.0564069747924805, | |
| "learning_rate": 2.1546184738955824e-06, | |
| "epoch": 1.6001628664495113, | |
| "step": 3930 | |
| }, | |
| { | |
| "loss": 0.0651, | |
| "grad_norm": 3.2962307929992676, | |
| "learning_rate": 2.134538152610442e-06, | |
| "epoch": 1.6042345276872965, | |
| "step": 3940 | |
| }, | |
| { | |
| "loss": 0.0605, | |
| "grad_norm": 3.0828750133514404, | |
| "learning_rate": 2.1144578313253016e-06, | |
| "epoch": 1.6083061889250816, | |
| "step": 3950 | |
| }, | |
| { | |
| "loss": 0.0639, | |
| "grad_norm": 2.274686574935913, | |
| "learning_rate": 2.094377510040161e-06, | |
| "epoch": 1.6123778501628665, | |
| "step": 3960 | |
| }, | |
| { | |
| "loss": 0.0629, | |
| "grad_norm": 2.720350980758667, | |
| "learning_rate": 2.0742971887550203e-06, | |
| "epoch": 1.6164495114006514, | |
| "step": 3970 | |
| }, | |
| { | |
| "loss": 0.0467, | |
| "grad_norm": 3.0196821689605713, | |
| "learning_rate": 2.0542168674698797e-06, | |
| "epoch": 1.6205211726384365, | |
| "step": 3980 | |
| }, | |
| { | |
| "loss": 0.0552, | |
| "grad_norm": 5.02653169631958, | |
| "learning_rate": 2.034136546184739e-06, | |
| "epoch": 1.6245928338762217, | |
| "step": 3990 | |
| }, | |
| { | |
| "loss": 0.0621, | |
| "grad_norm": 3.200761556625366, | |
| "learning_rate": 2.0140562248995985e-06, | |
| "epoch": 1.6286644951140063, | |
| "step": 4000 | |
| }, | |
| { | |
| "eval_loss": 0.08594676852226257, | |
| "eval_wer": 0.6990582448563374, | |
| "eval_runtime": 1102.0109, | |
| "eval_samples_per_second": 0.991, | |
| "eval_steps_per_second": 0.124, | |
| "epoch": 1.6286644951140063, | |
| "step": 4000 | |
| }, | |
| { | |
| "loss": 0.064, | |
| "grad_norm": 1.623856782913208, | |
| "learning_rate": 1.993975903614458e-06, | |
| "epoch": 1.6327361563517915, | |
| "step": 4010 | |
| }, | |
| { | |
| "loss": 0.0523, | |
| "grad_norm": 2.1615750789642334, | |
| "learning_rate": 1.9738955823293172e-06, | |
| "epoch": 1.6368078175895766, | |
| "step": 4020 | |
| }, | |
| { | |
| "loss": 0.0721, | |
| "grad_norm": 2.6709165573120117, | |
| "learning_rate": 1.953815261044177e-06, | |
| "epoch": 1.6408794788273615, | |
| "step": 4030 | |
| }, | |
| { | |
| "loss": 0.0645, | |
| "grad_norm": 4.282010078430176, | |
| "learning_rate": 1.9337349397590364e-06, | |
| "epoch": 1.6449511400651464, | |
| "step": 4040 | |
| }, | |
| { | |
| "loss": 0.0685, | |
| "grad_norm": 2.3375027179718018, | |
| "learning_rate": 1.9136546184738958e-06, | |
| "epoch": 1.6490228013029316, | |
| "step": 4050 | |
| }, | |
| { | |
| "loss": 0.0542, | |
| "grad_norm": 3.5294740200042725, | |
| "learning_rate": 1.8935742971887552e-06, | |
| "epoch": 1.6530944625407167, | |
| "step": 4060 | |
| }, | |
| { | |
| "loss": 0.0565, | |
| "grad_norm": 2.618698835372925, | |
| "learning_rate": 1.8734939759036147e-06, | |
| "epoch": 1.6571661237785016, | |
| "step": 4070 | |
| }, | |
| { | |
| "loss": 0.066, | |
| "grad_norm": 3.0897600650787354, | |
| "learning_rate": 1.8534136546184741e-06, | |
| "epoch": 1.6612377850162865, | |
| "step": 4080 | |
| }, | |
| { | |
| "loss": 0.0658, | |
| "grad_norm": 2.3968629837036133, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "epoch": 1.6653094462540716, | |
| "step": 4090 | |
| }, | |
| { | |
| "loss": 0.0643, | |
| "grad_norm": 2.8940470218658447, | |
| "learning_rate": 1.8132530120481929e-06, | |
| "epoch": 1.6693811074918568, | |
| "step": 4100 | |
| }, | |
| { | |
| "loss": 0.0583, | |
| "grad_norm": 2.8800227642059326, | |
| "learning_rate": 1.7931726907630523e-06, | |
| "epoch": 1.6734527687296417, | |
| "step": 4110 | |
| }, | |
| { | |
| "loss": 0.0646, | |
| "grad_norm": 3.087979555130005, | |
| "learning_rate": 1.7730923694779118e-06, | |
| "epoch": 1.6775244299674266, | |
| "step": 4120 | |
| }, | |
| { | |
| "loss": 0.0536, | |
| "grad_norm": 2.7075448036193848, | |
| "learning_rate": 1.7530120481927712e-06, | |
| "epoch": 1.6815960912052117, | |
| "step": 4130 | |
| }, | |
| { | |
| "loss": 0.052, | |
| "grad_norm": 1.812359094619751, | |
| "learning_rate": 1.7329317269076306e-06, | |
| "epoch": 1.6856677524429968, | |
| "step": 4140 | |
| }, | |
| { | |
| "loss": 0.0487, | |
| "grad_norm": 3.518482208251953, | |
| "learning_rate": 1.7128514056224902e-06, | |
| "epoch": 1.6897394136807817, | |
| "step": 4150 | |
| }, | |
| { | |
| "loss": 0.0649, | |
| "grad_norm": 1.6757196187973022, | |
| "learning_rate": 1.6927710843373496e-06, | |
| "epoch": 1.6938110749185666, | |
| "step": 4160 | |
| }, | |
| { | |
| "loss": 0.0725, | |
| "grad_norm": 3.218384265899658, | |
| "learning_rate": 1.6726907630522091e-06, | |
| "epoch": 1.6978827361563518, | |
| "step": 4170 | |
| }, | |
| { | |
| "loss": 0.06, | |
| "grad_norm": 3.9945616722106934, | |
| "learning_rate": 1.6526104417670685e-06, | |
| "epoch": 1.701954397394137, | |
| "step": 4180 | |
| }, | |
| { | |
| "loss": 0.0648, | |
| "grad_norm": 1.5746277570724487, | |
| "learning_rate": 1.6325301204819277e-06, | |
| "epoch": 1.7060260586319218, | |
| "step": 4190 | |
| }, | |
| { | |
| "loss": 0.0609, | |
| "grad_norm": 2.5348174571990967, | |
| "learning_rate": 1.6124497991967873e-06, | |
| "epoch": 1.7100977198697067, | |
| "step": 4200 | |
| }, | |
| { | |
| "loss": 0.0604, | |
| "grad_norm": 2.6076154708862305, | |
| "learning_rate": 1.5923694779116467e-06, | |
| "epoch": 1.7141693811074918, | |
| "step": 4210 | |
| }, | |
| { | |
| "loss": 0.0523, | |
| "grad_norm": 2.4243481159210205, | |
| "learning_rate": 1.572289156626506e-06, | |
| "epoch": 1.718241042345277, | |
| "step": 4220 | |
| }, | |
| { | |
| "loss": 0.0679, | |
| "grad_norm": 3.78879451751709, | |
| "learning_rate": 1.5522088353413656e-06, | |
| "epoch": 1.7223127035830619, | |
| "step": 4230 | |
| }, | |
| { | |
| "loss": 0.0773, | |
| "grad_norm": 3.3975095748901367, | |
| "learning_rate": 1.532128514056225e-06, | |
| "epoch": 1.7263843648208468, | |
| "step": 4240 | |
| }, | |
| { | |
| "loss": 0.0592, | |
| "grad_norm": 2.3238937854766846, | |
| "learning_rate": 1.5120481927710846e-06, | |
| "epoch": 1.730456026058632, | |
| "step": 4250 | |
| }, | |
| { | |
| "loss": 0.0564, | |
| "grad_norm": 2.3824143409729004, | |
| "learning_rate": 1.491967871485944e-06, | |
| "epoch": 1.734527687296417, | |
| "step": 4260 | |
| }, | |
| { | |
| "loss": 0.0582, | |
| "grad_norm": 2.944546937942505, | |
| "learning_rate": 1.4718875502008033e-06, | |
| "epoch": 1.738599348534202, | |
| "step": 4270 | |
| }, | |
| { | |
| "loss": 0.0701, | |
| "grad_norm": 2.0511245727539062, | |
| "learning_rate": 1.451807228915663e-06, | |
| "epoch": 1.7426710097719869, | |
| "step": 4280 | |
| }, | |
| { | |
| "loss": 0.0478, | |
| "grad_norm": 3.890240430831909, | |
| "learning_rate": 1.431726907630522e-06, | |
| "epoch": 1.746742671009772, | |
| "step": 4290 | |
| }, | |
| { | |
| "loss": 0.0622, | |
| "grad_norm": 2.7167794704437256, | |
| "learning_rate": 1.4116465863453815e-06, | |
| "epoch": 1.7508143322475571, | |
| "step": 4300 | |
| }, | |
| { | |
| "loss": 0.0729, | |
| "grad_norm": 2.829355001449585, | |
| "learning_rate": 1.391566265060241e-06, | |
| "epoch": 1.754885993485342, | |
| "step": 4310 | |
| }, | |
| { | |
| "loss": 0.0564, | |
| "grad_norm": 2.425443410873413, | |
| "learning_rate": 1.3714859437751004e-06, | |
| "epoch": 1.758957654723127, | |
| "step": 4320 | |
| }, | |
| { | |
| "loss": 0.0609, | |
| "grad_norm": 2.4310548305511475, | |
| "learning_rate": 1.35140562248996e-06, | |
| "epoch": 1.763029315960912, | |
| "step": 4330 | |
| }, | |
| { | |
| "loss": 0.0738, | |
| "grad_norm": 6.1249003410339355, | |
| "learning_rate": 1.3313253012048194e-06, | |
| "epoch": 1.7671009771986972, | |
| "step": 4340 | |
| }, | |
| { | |
| "loss": 0.0587, | |
| "grad_norm": 2.752681016921997, | |
| "learning_rate": 1.3132530120481928e-06, | |
| "epoch": 1.771172638436482, | |
| "step": 4350 | |
| }, | |
| { | |
| "loss": 0.0742, | |
| "grad_norm": 3.6890759468078613, | |
| "learning_rate": 1.2931726907630524e-06, | |
| "epoch": 1.775244299674267, | |
| "step": 4360 | |
| }, | |
| { | |
| "loss": 0.0658, | |
| "grad_norm": 2.928663730621338, | |
| "learning_rate": 1.2730923694779118e-06, | |
| "epoch": 1.7793159609120521, | |
| "step": 4370 | |
| }, | |
| { | |
| "loss": 0.0743, | |
| "grad_norm": 2.4388198852539062, | |
| "learning_rate": 1.2530120481927713e-06, | |
| "epoch": 1.7833876221498373, | |
| "step": 4380 | |
| }, | |
| { | |
| "loss": 0.0543, | |
| "grad_norm": 2.4720866680145264, | |
| "learning_rate": 1.2329317269076305e-06, | |
| "epoch": 1.7874592833876222, | |
| "step": 4390 | |
| }, | |
| { | |
| "loss": 0.0765, | |
| "grad_norm": 4.353959083557129, | |
| "learning_rate": 1.21285140562249e-06, | |
| "epoch": 1.791530944625407, | |
| "step": 4400 | |
| }, | |
| { | |
| "loss": 0.0582, | |
| "grad_norm": 2.662203311920166, | |
| "learning_rate": 1.1927710843373495e-06, | |
| "epoch": 1.7956026058631922, | |
| "step": 4410 | |
| }, | |
| { | |
| "loss": 0.0602, | |
| "grad_norm": 2.8086483478546143, | |
| "learning_rate": 1.172690763052209e-06, | |
| "epoch": 1.7996742671009773, | |
| "step": 4420 | |
| }, | |
| { | |
| "loss": 0.0591, | |
| "grad_norm": 3.4378201961517334, | |
| "learning_rate": 1.1526104417670682e-06, | |
| "epoch": 1.8037459283387622, | |
| "step": 4430 | |
| }, | |
| { | |
| "loss": 0.0562, | |
| "grad_norm": 3.4084293842315674, | |
| "learning_rate": 1.1325301204819278e-06, | |
| "epoch": 1.8078175895765471, | |
| "step": 4440 | |
| }, | |
| { | |
| "loss": 0.054, | |
| "grad_norm": 2.4970297813415527, | |
| "learning_rate": 1.1124497991967872e-06, | |
| "epoch": 1.8118892508143323, | |
| "step": 4450 | |
| }, | |
| { | |
| "loss": 0.0596, | |
| "grad_norm": 3.488628387451172, | |
| "learning_rate": 1.0923694779116468e-06, | |
| "epoch": 1.8159609120521174, | |
| "step": 4460 | |
| }, | |
| { | |
| "loss": 0.0602, | |
| "grad_norm": 4.658164024353027, | |
| "learning_rate": 1.0722891566265062e-06, | |
| "epoch": 1.8200325732899023, | |
| "step": 4470 | |
| }, | |
| { | |
| "loss": 0.0471, | |
| "grad_norm": 2.424395799636841, | |
| "learning_rate": 1.0522088353413655e-06, | |
| "epoch": 1.8241042345276872, | |
| "step": 4480 | |
| }, | |
| { | |
| "loss": 0.0566, | |
| "grad_norm": 2.604961633682251, | |
| "learning_rate": 1.032128514056225e-06, | |
| "epoch": 1.8281758957654723, | |
| "step": 4490 | |
| }, | |
| { | |
| "loss": 0.0612, | |
| "grad_norm": 2.1949777603149414, | |
| "learning_rate": 1.0120481927710845e-06, | |
| "epoch": 1.8322475570032575, | |
| "step": 4500 | |
| }, | |
| { | |
| "eval_loss": 0.08283434808254242, | |
| "eval_wer": 0.6747045768100294, | |
| "eval_runtime": 1066.2713, | |
| "eval_samples_per_second": 1.024, | |
| "eval_steps_per_second": 0.128, | |
| "epoch": 1.8322475570032575, | |
| "step": 4500 | |
| }, | |
| { | |
| "loss": 0.0706, | |
| "grad_norm": 3.848797559738159, | |
| "learning_rate": 9.919678714859439e-07, | |
| "epoch": 1.8363192182410424, | |
| "step": 4510 | |
| }, | |
| { | |
| "loss": 0.0757, | |
| "grad_norm": 3.5777056217193604, | |
| "learning_rate": 9.718875502008035e-07, | |
| "epoch": 1.8403908794788273, | |
| "step": 4520 | |
| }, | |
| { | |
| "loss": 0.0677, | |
| "grad_norm": 3.7705461978912354, | |
| "learning_rate": 9.518072289156627e-07, | |
| "epoch": 1.8444625407166124, | |
| "step": 4530 | |
| }, | |
| { | |
| "loss": 0.0599, | |
| "grad_norm": 2.314631462097168, | |
| "learning_rate": 9.317269076305221e-07, | |
| "epoch": 1.8485342019543975, | |
| "step": 4540 | |
| }, | |
| { | |
| "loss": 0.0576, | |
| "grad_norm": 1.3859368562698364, | |
| "learning_rate": 9.116465863453816e-07, | |
| "epoch": 1.8526058631921825, | |
| "step": 4550 | |
| }, | |
| { | |
| "loss": 0.0649, | |
| "grad_norm": 1.7714697122573853, | |
| "learning_rate": 8.915662650602411e-07, | |
| "epoch": 1.8566775244299674, | |
| "step": 4560 | |
| }, | |
| { | |
| "loss": 0.0614, | |
| "grad_norm": 3.5768227577209473, | |
| "learning_rate": 8.714859437751006e-07, | |
| "epoch": 1.8607491856677525, | |
| "step": 4570 | |
| }, | |
| { | |
| "loss": 0.0554, | |
| "grad_norm": 2.688397169113159, | |
| "learning_rate": 8.514056224899598e-07, | |
| "epoch": 1.8648208469055376, | |
| "step": 4580 | |
| }, | |
| { | |
| "loss": 0.0535, | |
| "grad_norm": 2.1303982734680176, | |
| "learning_rate": 8.313253012048193e-07, | |
| "epoch": 1.8688925081433225, | |
| "step": 4590 | |
| }, | |
| { | |
| "loss": 0.0636, | |
| "grad_norm": 2.2192540168762207, | |
| "learning_rate": 8.112449799196788e-07, | |
| "epoch": 1.8729641693811074, | |
| "step": 4600 | |
| }, | |
| { | |
| "loss": 0.0617, | |
| "grad_norm": 2.636942148208618, | |
| "learning_rate": 7.911646586345383e-07, | |
| "epoch": 1.8770358306188926, | |
| "step": 4610 | |
| }, | |
| { | |
| "loss": 0.0462, | |
| "grad_norm": 2.2228713035583496, | |
| "learning_rate": 7.710843373493978e-07, | |
| "epoch": 1.8811074918566775, | |
| "step": 4620 | |
| }, | |
| { | |
| "loss": 0.0557, | |
| "grad_norm": 2.411165475845337, | |
| "learning_rate": 7.51004016064257e-07, | |
| "epoch": 1.8851791530944624, | |
| "step": 4630 | |
| }, | |
| { | |
| "loss": 0.0554, | |
| "grad_norm": 2.742436408996582, | |
| "learning_rate": 7.309236947791165e-07, | |
| "epoch": 1.8892508143322475, | |
| "step": 4640 | |
| }, | |
| { | |
| "loss": 0.0651, | |
| "grad_norm": 2.5632567405700684, | |
| "learning_rate": 7.10843373493976e-07, | |
| "epoch": 1.8933224755700326, | |
| "step": 4650 | |
| }, | |
| { | |
| "loss": 0.0568, | |
| "grad_norm": 2.021392345428467, | |
| "learning_rate": 6.907630522088355e-07, | |
| "epoch": 1.8973941368078175, | |
| "step": 4660 | |
| }, | |
| { | |
| "loss": 0.0573, | |
| "grad_norm": 1.7550697326660156, | |
| "learning_rate": 6.706827309236949e-07, | |
| "epoch": 1.9014657980456025, | |
| "step": 4670 | |
| }, | |
| { | |
| "loss": 0.0512, | |
| "grad_norm": 3.079806327819824, | |
| "learning_rate": 6.506024096385542e-07, | |
| "epoch": 1.9055374592833876, | |
| "step": 4680 | |
| }, | |
| { | |
| "loss": 0.057, | |
| "grad_norm": 2.4125704765319824, | |
| "learning_rate": 6.305220883534137e-07, | |
| "epoch": 1.9096091205211727, | |
| "step": 4690 | |
| }, | |
| { | |
| "loss": 0.0536, | |
| "grad_norm": 2.7389943599700928, | |
| "learning_rate": 6.104417670682732e-07, | |
| "epoch": 1.9136807817589576, | |
| "step": 4700 | |
| }, | |
| { | |
| "loss": 0.0529, | |
| "grad_norm": 3.2968907356262207, | |
| "learning_rate": 5.903614457831326e-07, | |
| "epoch": 1.9177524429967425, | |
| "step": 4710 | |
| }, | |
| { | |
| "loss": 0.0524, | |
| "grad_norm": 2.5253422260284424, | |
| "learning_rate": 5.702811244979921e-07, | |
| "epoch": 1.9218241042345277, | |
| "step": 4720 | |
| }, | |
| { | |
| "loss": 0.0519, | |
| "grad_norm": 2.081864356994629, | |
| "learning_rate": 5.502008032128514e-07, | |
| "epoch": 1.9258957654723128, | |
| "step": 4730 | |
| }, | |
| { | |
| "loss": 0.0688, | |
| "grad_norm": 4.3627495765686035, | |
| "learning_rate": 5.301204819277109e-07, | |
| "epoch": 1.9299674267100977, | |
| "step": 4740 | |
| }, | |
| { | |
| "loss": 0.0568, | |
| "grad_norm": 2.8700270652770996, | |
| "learning_rate": 5.100401606425703e-07, | |
| "epoch": 1.9340390879478826, | |
| "step": 4750 | |
| }, | |
| { | |
| "loss": 0.0597, | |
| "grad_norm": 2.5363810062408447, | |
| "learning_rate": 4.899598393574298e-07, | |
| "epoch": 1.9381107491856677, | |
| "step": 4760 | |
| }, | |
| { | |
| "loss": 0.0562, | |
| "grad_norm": 2.8106465339660645, | |
| "learning_rate": 4.698795180722892e-07, | |
| "epoch": 1.9421824104234529, | |
| "step": 4770 | |
| }, | |
| { | |
| "loss": 0.0668, | |
| "grad_norm": 3.303130626678467, | |
| "learning_rate": 4.497991967871487e-07, | |
| "epoch": 1.9462540716612378, | |
| "step": 4780 | |
| }, | |
| { | |
| "loss": 0.0534, | |
| "grad_norm": 2.3976430892944336, | |
| "learning_rate": 4.2971887550200806e-07, | |
| "epoch": 1.9503257328990227, | |
| "step": 4790 | |
| }, | |
| { | |
| "loss": 0.0599, | |
| "grad_norm": 5.05959415435791, | |
| "learning_rate": 4.0963855421686754e-07, | |
| "epoch": 1.9543973941368078, | |
| "step": 4800 | |
| }, | |
| { | |
| "loss": 0.0579, | |
| "grad_norm": 2.829632520675659, | |
| "learning_rate": 3.895582329317269e-07, | |
| "epoch": 1.958469055374593, | |
| "step": 4810 | |
| }, | |
| { | |
| "loss": 0.0605, | |
| "grad_norm": 2.824951410293579, | |
| "learning_rate": 3.694779116465864e-07, | |
| "epoch": 1.9625407166123778, | |
| "step": 4820 | |
| }, | |
| { | |
| "loss": 0.0714, | |
| "grad_norm": 2.989415168762207, | |
| "learning_rate": 3.4939759036144583e-07, | |
| "epoch": 1.9666123778501627, | |
| "step": 4830 | |
| }, | |
| { | |
| "loss": 0.0476, | |
| "grad_norm": 2.8402302265167236, | |
| "learning_rate": 3.2931726907630526e-07, | |
| "epoch": 1.9706840390879479, | |
| "step": 4840 | |
| }, | |
| { | |
| "loss": 0.0598, | |
| "grad_norm": 2.0398058891296387, | |
| "learning_rate": 3.092369477911647e-07, | |
| "epoch": 1.974755700325733, | |
| "step": 4850 | |
| }, | |
| { | |
| "loss": 0.0526, | |
| "grad_norm": 3.0668020248413086, | |
| "learning_rate": 2.891566265060241e-07, | |
| "epoch": 1.978827361563518, | |
| "step": 4860 | |
| }, | |
| { | |
| "loss": 0.0623, | |
| "grad_norm": 3.341176748275757, | |
| "learning_rate": 2.6907630522088355e-07, | |
| "epoch": 1.9828990228013028, | |
| "step": 4870 | |
| }, | |
| { | |
| "loss": 0.0791, | |
| "grad_norm": 5.048247814178467, | |
| "learning_rate": 2.48995983935743e-07, | |
| "epoch": 1.986970684039088, | |
| "step": 4880 | |
| }, | |
| { | |
| "loss": 0.0549, | |
| "grad_norm": 2.9377377033233643, | |
| "learning_rate": 2.289156626506024e-07, | |
| "epoch": 1.991042345276873, | |
| "step": 4890 | |
| }, | |
| { | |
| "loss": 0.0507, | |
| "grad_norm": 3.243870496749878, | |
| "learning_rate": 2.088353413654619e-07, | |
| "epoch": 1.995114006514658, | |
| "step": 4900 | |
| }, | |
| { | |
| "loss": 0.0698, | |
| "grad_norm": 3.158334493637085, | |
| "learning_rate": 1.8875502008032132e-07, | |
| "epoch": 1.9991856677524429, | |
| "step": 4910 | |
| }, | |
| { | |
| "loss": 0.0344, | |
| "grad_norm": 2.143577814102173, | |
| "learning_rate": 1.6867469879518075e-07, | |
| "epoch": 2.003257328990228, | |
| "step": 4920 | |
| }, | |
| { | |
| "loss": 0.0377, | |
| "grad_norm": 2.3065574169158936, | |
| "learning_rate": 1.4859437751004018e-07, | |
| "epoch": 2.007328990228013, | |
| "step": 4930 | |
| }, | |
| { | |
| "loss": 0.0465, | |
| "grad_norm": 2.2823996543884277, | |
| "learning_rate": 1.285140562248996e-07, | |
| "epoch": 2.011400651465798, | |
| "step": 4940 | |
| }, | |
| { | |
| "loss": 0.0394, | |
| "grad_norm": 1.8728781938552856, | |
| "learning_rate": 1.0843373493975904e-07, | |
| "epoch": 2.015472312703583, | |
| "step": 4950 | |
| }, | |
| { | |
| "loss": 0.045, | |
| "grad_norm": 2.2312207221984863, | |
| "learning_rate": 8.835341365461847e-08, | |
| "epoch": 2.019543973941368, | |
| "step": 4960 | |
| }, | |
| { | |
| "loss": 0.0312, | |
| "grad_norm": 1.7082674503326416, | |
| "learning_rate": 6.827309236947792e-08, | |
| "epoch": 2.023615635179153, | |
| "step": 4970 | |
| }, | |
| { | |
| "loss": 0.0363, | |
| "grad_norm": 3.192293167114258, | |
| "learning_rate": 4.819277108433736e-08, | |
| "epoch": 2.027687296416938, | |
| "step": 4980 | |
| }, | |
| { | |
| "loss": 0.0441, | |
| "grad_norm": 2.6271777153015137, | |
| "learning_rate": 2.8112449799196793e-08, | |
| "epoch": 2.031758957654723, | |
| "step": 4990 | |
| }, | |
| { | |
| "loss": 0.0477, | |
| "grad_norm": 3.464754581451416, | |
| "learning_rate": 8.032128514056226e-09, | |
| "epoch": 2.035830618892508, | |
| "step": 5000 | |
| }, | |
| { | |
| "eval_loss": 0.0814063549041748, | |
| "eval_wer": 0.6870613640453482, | |
| "eval_runtime": 1070.1708, | |
| "eval_samples_per_second": 1.02, | |
| "eval_steps_per_second": 0.128, | |
| "epoch": 2.035830618892508, | |
| "step": 5000 | |
| }, | |
| { | |
| "train_runtime": 14776.9064, | |
| "train_samples_per_second": 1.353, | |
| "train_steps_per_second": 0.338, | |
| "total_flos": 5.7717080064e+18, | |
| "train_loss": 0.11340971575975418, | |
| "epoch": 2.035830618892508, | |
| "step": 5000 | |
| } | |
| ] |