| { | |
| "best_metric": 16.52112384371117, | |
| "best_model_checkpoint": "./whisper-medium-korean/checkpoint-9000", | |
| "epoch": 3.461405330564209, | |
| "eval_steps": 500, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.008653513326410523, | |
| "grad_norm": 9.3709077835083, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.8506, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.017307026652821047, | |
| "grad_norm": 5.6410064697265625, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.7759, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02596053997923157, | |
| "grad_norm": 6.114657878875732, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.627, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.034614053305642094, | |
| "grad_norm": 4.652756690979004, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.4309, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04326756663205261, | |
| "grad_norm": 4.665812015533447, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.3935, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05192107995846314, | |
| "grad_norm": 4.651270389556885, | |
| "learning_rate": 3e-06, | |
| "loss": 0.365, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.060574593284873655, | |
| "grad_norm": 4.162643909454346, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.3651, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.06922810661128419, | |
| "grad_norm": 3.771670341491699, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.3666, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0778816199376947, | |
| "grad_norm": 4.835504531860352, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.3505, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.08653513326410522, | |
| "grad_norm": 5.903632164001465, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3394, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09518864659051575, | |
| "grad_norm": 4.5515336990356445, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.3298, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.10384215991692627, | |
| "grad_norm": 4.889475345611572, | |
| "learning_rate": 6e-06, | |
| "loss": 0.3493, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.1124956732433368, | |
| "grad_norm": 4.62334680557251, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.2977, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.12114918656974731, | |
| "grad_norm": 4.044426441192627, | |
| "learning_rate": 7e-06, | |
| "loss": 0.3322, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.12980269989615784, | |
| "grad_norm": 4.073489189147949, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.3359, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.13845621322256838, | |
| "grad_norm": 5.082677841186523, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3426, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1471097265489789, | |
| "grad_norm": 4.216761589050293, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3317, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.1557632398753894, | |
| "grad_norm": 5.513101577758789, | |
| "learning_rate": 9e-06, | |
| "loss": 0.3044, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.16441675320179994, | |
| "grad_norm": 4.340704917907715, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.3284, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.17307026652821045, | |
| "grad_norm": 4.588245391845703, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3114, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.17307026652821045, | |
| "eval_loss": 0.2976242005825043, | |
| "eval_runtime": 6820.6363, | |
| "eval_samples_per_second": 1.506, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 24.376984674858484, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.181723779854621, | |
| "grad_norm": 3.102633237838745, | |
| "learning_rate": 9.973684210526316e-06, | |
| "loss": 0.2907, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.1903772931810315, | |
| "grad_norm": 4.586500644683838, | |
| "learning_rate": 9.947368421052632e-06, | |
| "loss": 0.2977, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.199030806507442, | |
| "grad_norm": 4.89168119430542, | |
| "learning_rate": 9.921052631578947e-06, | |
| "loss": 0.2984, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.20768431983385255, | |
| "grad_norm": 4.386935710906982, | |
| "learning_rate": 9.894736842105264e-06, | |
| "loss": 0.3103, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.21633783316026306, | |
| "grad_norm": 4.106662750244141, | |
| "learning_rate": 9.868421052631579e-06, | |
| "loss": 0.2919, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.2249913464866736, | |
| "grad_norm": 4.3443827629089355, | |
| "learning_rate": 9.842105263157896e-06, | |
| "loss": 0.3029, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2336448598130841, | |
| "grad_norm": 4.201015472412109, | |
| "learning_rate": 9.815789473684212e-06, | |
| "loss": 0.2764, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.24229837313949462, | |
| "grad_norm": 4.213696002960205, | |
| "learning_rate": 9.789473684210527e-06, | |
| "loss": 0.289, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25095188646590516, | |
| "grad_norm": 4.507613182067871, | |
| "learning_rate": 9.763157894736844e-06, | |
| "loss": 0.2937, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.25960539979231567, | |
| "grad_norm": 4.777523994445801, | |
| "learning_rate": 9.736842105263159e-06, | |
| "loss": 0.3014, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.2682589131187262, | |
| "grad_norm": 3.791252613067627, | |
| "learning_rate": 9.710526315789474e-06, | |
| "loss": 0.2873, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.27691242644513675, | |
| "grad_norm": 4.165738105773926, | |
| "learning_rate": 9.68421052631579e-06, | |
| "loss": 0.2959, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.28556593977154726, | |
| "grad_norm": 4.374238967895508, | |
| "learning_rate": 9.657894736842106e-06, | |
| "loss": 0.286, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.2942194530979578, | |
| "grad_norm": 4.140951633453369, | |
| "learning_rate": 9.631578947368422e-06, | |
| "loss": 0.2824, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.3028729664243683, | |
| "grad_norm": 3.896604537963867, | |
| "learning_rate": 9.605263157894737e-06, | |
| "loss": 0.2774, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.3115264797507788, | |
| "grad_norm": 4.016544818878174, | |
| "learning_rate": 9.578947368421054e-06, | |
| "loss": 0.2629, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.32017999307718936, | |
| "grad_norm": 3.743680000305176, | |
| "learning_rate": 9.552631578947369e-06, | |
| "loss": 0.2585, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.3288335064035999, | |
| "grad_norm": 3.727074384689331, | |
| "learning_rate": 9.526315789473684e-06, | |
| "loss": 0.2608, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.3374870197300104, | |
| "grad_norm": 3.8571677207946777, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.273, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.3461405330564209, | |
| "grad_norm": 4.067797660827637, | |
| "learning_rate": 9.473684210526315e-06, | |
| "loss": 0.2765, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3461405330564209, | |
| "eval_loss": 0.26566416025161743, | |
| "eval_runtime": 6826.7941, | |
| "eval_samples_per_second": 1.505, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 21.943600717934558, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3547940463828314, | |
| "grad_norm": 3.9521491527557373, | |
| "learning_rate": 9.447368421052632e-06, | |
| "loss": 0.2696, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.363447559709242, | |
| "grad_norm": 4.91107177734375, | |
| "learning_rate": 9.421052631578949e-06, | |
| "loss": 0.2748, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3721010730356525, | |
| "grad_norm": 3.5369150638580322, | |
| "learning_rate": 9.394736842105264e-06, | |
| "loss": 0.272, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.380754586362063, | |
| "grad_norm": 3.5748019218444824, | |
| "learning_rate": 9.36842105263158e-06, | |
| "loss": 0.2572, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.3894080996884735, | |
| "grad_norm": 4.190589904785156, | |
| "learning_rate": 9.342105263157895e-06, | |
| "loss": 0.2932, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.398061613014884, | |
| "grad_norm": 3.347419023513794, | |
| "learning_rate": 9.315789473684212e-06, | |
| "loss": 0.2656, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.4067151263412946, | |
| "grad_norm": 2.9487226009368896, | |
| "learning_rate": 9.289473684210527e-06, | |
| "loss": 0.2666, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.4153686396677051, | |
| "grad_norm": 3.4610071182250977, | |
| "learning_rate": 9.263157894736842e-06, | |
| "loss": 0.2526, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.4240221529941156, | |
| "grad_norm": 4.092984676361084, | |
| "learning_rate": 9.236842105263159e-06, | |
| "loss": 0.2707, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.4326756663205261, | |
| "grad_norm": 4.32016658782959, | |
| "learning_rate": 9.210526315789474e-06, | |
| "loss": 0.2716, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.44132917964693663, | |
| "grad_norm": 4.7245330810546875, | |
| "learning_rate": 9.18421052631579e-06, | |
| "loss": 0.2554, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.4499826929733472, | |
| "grad_norm": 4.063606262207031, | |
| "learning_rate": 9.157894736842105e-06, | |
| "loss": 0.2617, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.4586362062997577, | |
| "grad_norm": 3.9536917209625244, | |
| "learning_rate": 9.131578947368422e-06, | |
| "loss": 0.2471, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.4672897196261682, | |
| "grad_norm": 3.8031530380249023, | |
| "learning_rate": 9.105263157894739e-06, | |
| "loss": 0.2687, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.47594323295257873, | |
| "grad_norm": 3.8580405712127686, | |
| "learning_rate": 9.078947368421054e-06, | |
| "loss": 0.2592, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.48459674627898924, | |
| "grad_norm": 3.3053033351898193, | |
| "learning_rate": 9.05263157894737e-06, | |
| "loss": 0.2412, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.4932502596053998, | |
| "grad_norm": 3.1904115676879883, | |
| "learning_rate": 9.026315789473685e-06, | |
| "loss": 0.2672, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.5019037729318103, | |
| "grad_norm": 2.96549916267395, | |
| "learning_rate": 9e-06, | |
| "loss": 0.2425, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.5105572862582208, | |
| "grad_norm": 3.1574394702911377, | |
| "learning_rate": 8.973684210526317e-06, | |
| "loss": 0.2536, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.5192107995846313, | |
| "grad_norm": 4.734163284301758, | |
| "learning_rate": 8.947368421052632e-06, | |
| "loss": 0.2737, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5192107995846313, | |
| "eval_loss": 0.24431759119033813, | |
| "eval_runtime": 6895.0818, | |
| "eval_samples_per_second": 1.49, | |
| "eval_steps_per_second": 0.093, | |
| "eval_wer": 20.694981361314372, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5278643129110419, | |
| "grad_norm": 4.172092437744141, | |
| "learning_rate": 8.921052631578949e-06, | |
| "loss": 0.2561, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.5365178262374524, | |
| "grad_norm": 4.126692295074463, | |
| "learning_rate": 8.894736842105264e-06, | |
| "loss": 0.2582, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.5451713395638629, | |
| "grad_norm": 3.5045013427734375, | |
| "learning_rate": 8.86842105263158e-06, | |
| "loss": 0.2563, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.5538248528902735, | |
| "grad_norm": 3.760963201522827, | |
| "learning_rate": 8.842105263157895e-06, | |
| "loss": 0.2591, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.562478366216684, | |
| "grad_norm": 3.065091848373413, | |
| "learning_rate": 8.81578947368421e-06, | |
| "loss": 0.2512, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.5711318795430945, | |
| "grad_norm": 3.6731486320495605, | |
| "learning_rate": 8.789473684210527e-06, | |
| "loss": 0.243, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.579785392869505, | |
| "grad_norm": 4.433581352233887, | |
| "learning_rate": 8.763157894736842e-06, | |
| "loss": 0.2265, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.5884389061959155, | |
| "grad_norm": 3.274906635284424, | |
| "learning_rate": 8.736842105263158e-06, | |
| "loss": 0.2495, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5970924195223261, | |
| "grad_norm": 3.508803129196167, | |
| "learning_rate": 8.710526315789475e-06, | |
| "loss": 0.2397, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.6057459328487366, | |
| "grad_norm": 4.499612331390381, | |
| "learning_rate": 8.68421052631579e-06, | |
| "loss": 0.2592, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.6143994461751471, | |
| "grad_norm": 3.7848877906799316, | |
| "learning_rate": 8.657894736842107e-06, | |
| "loss": 0.2438, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.6230529595015576, | |
| "grad_norm": 3.3857030868530273, | |
| "learning_rate": 8.631578947368422e-06, | |
| "loss": 0.2353, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.6317064728279681, | |
| "grad_norm": 4.552176475524902, | |
| "learning_rate": 8.605263157894738e-06, | |
| "loss": 0.2537, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.6403599861543787, | |
| "grad_norm": 3.2606961727142334, | |
| "learning_rate": 8.578947368421053e-06, | |
| "loss": 0.2579, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.6490134994807892, | |
| "grad_norm": 4.423758506774902, | |
| "learning_rate": 8.552631578947368e-06, | |
| "loss": 0.2443, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.6576670128071997, | |
| "grad_norm": 3.7660324573516846, | |
| "learning_rate": 8.526315789473685e-06, | |
| "loss": 0.2353, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.6663205261336103, | |
| "grad_norm": 3.4143283367156982, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.232, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.6749740394600208, | |
| "grad_norm": 3.8295090198516846, | |
| "learning_rate": 8.473684210526317e-06, | |
| "loss": 0.2451, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.6836275527864313, | |
| "grad_norm": 2.5151150226593018, | |
| "learning_rate": 8.447368421052632e-06, | |
| "loss": 0.2443, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.6922810661128418, | |
| "grad_norm": 3.353902816772461, | |
| "learning_rate": 8.421052631578948e-06, | |
| "loss": 0.2421, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6922810661128418, | |
| "eval_loss": 0.2288905382156372, | |
| "eval_runtime": 6825.4022, | |
| "eval_samples_per_second": 1.505, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 19.04683832666022, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.7009345794392523, | |
| "grad_norm": 3.2154219150543213, | |
| "learning_rate": 8.394736842105263e-06, | |
| "loss": 0.2369, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.7095880927656628, | |
| "grad_norm": 3.2622313499450684, | |
| "learning_rate": 8.36842105263158e-06, | |
| "loss": 0.2323, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.7182416060920734, | |
| "grad_norm": 4.255507946014404, | |
| "learning_rate": 8.342105263157897e-06, | |
| "loss": 0.2383, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.726895119418484, | |
| "grad_norm": 3.3413267135620117, | |
| "learning_rate": 8.315789473684212e-06, | |
| "loss": 0.2647, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.7355486327448945, | |
| "grad_norm": 3.5837645530700684, | |
| "learning_rate": 8.289473684210526e-06, | |
| "loss": 0.2209, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.744202146071305, | |
| "grad_norm": 3.2607805728912354, | |
| "learning_rate": 8.263157894736843e-06, | |
| "loss": 0.2338, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.7528556593977155, | |
| "grad_norm": 4.119529724121094, | |
| "learning_rate": 8.236842105263158e-06, | |
| "loss": 0.2279, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.761509172724126, | |
| "grad_norm": 3.5672607421875, | |
| "learning_rate": 8.210526315789475e-06, | |
| "loss": 0.2376, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.7701626860505365, | |
| "grad_norm": 4.001561164855957, | |
| "learning_rate": 8.18421052631579e-06, | |
| "loss": 0.2426, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.778816199376947, | |
| "grad_norm": 3.481905698776245, | |
| "learning_rate": 8.157894736842106e-06, | |
| "loss": 0.203, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.7874697127033575, | |
| "grad_norm": 3.281797170639038, | |
| "learning_rate": 8.131578947368421e-06, | |
| "loss": 0.2119, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.796123226029768, | |
| "grad_norm": 3.9247775077819824, | |
| "learning_rate": 8.105263157894736e-06, | |
| "loss": 0.2161, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.8047767393561787, | |
| "grad_norm": 3.621325731277466, | |
| "learning_rate": 8.078947368421053e-06, | |
| "loss": 0.2402, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.8134302526825892, | |
| "grad_norm": 3.3069469928741455, | |
| "learning_rate": 8.052631578947368e-06, | |
| "loss": 0.2236, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.8220837660089997, | |
| "grad_norm": 2.723513603210449, | |
| "learning_rate": 8.026315789473685e-06, | |
| "loss": 0.2356, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.8307372793354102, | |
| "grad_norm": 3.1568410396575928, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2054, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.8393907926618207, | |
| "grad_norm": 3.152918815612793, | |
| "learning_rate": 7.973684210526316e-06, | |
| "loss": 0.2292, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.8480443059882312, | |
| "grad_norm": 3.4354190826416016, | |
| "learning_rate": 7.947368421052633e-06, | |
| "loss": 0.2228, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.8566978193146417, | |
| "grad_norm": 3.4018874168395996, | |
| "learning_rate": 7.921052631578948e-06, | |
| "loss": 0.2346, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.8653513326410522, | |
| "grad_norm": 2.7067339420318604, | |
| "learning_rate": 7.894736842105265e-06, | |
| "loss": 0.2237, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.8653513326410522, | |
| "eval_loss": 0.22117741405963898, | |
| "eval_runtime": 6835.5047, | |
| "eval_samples_per_second": 1.503, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 18.436766533204473, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.8740048459674628, | |
| "grad_norm": 3.9850449562072754, | |
| "learning_rate": 7.86842105263158e-06, | |
| "loss": 0.2334, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.8826583592938733, | |
| "grad_norm": 3.366445302963257, | |
| "learning_rate": 7.842105263157895e-06, | |
| "loss": 0.2507, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.8913118726202839, | |
| "grad_norm": 2.6732399463653564, | |
| "learning_rate": 7.815789473684211e-06, | |
| "loss": 0.23, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.8999653859466944, | |
| "grad_norm": 3.804313898086548, | |
| "learning_rate": 7.789473684210526e-06, | |
| "loss": 0.2527, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.9086188992731049, | |
| "grad_norm": 4.052364826202393, | |
| "learning_rate": 7.763157894736843e-06, | |
| "loss": 0.2459, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.9172724125995154, | |
| "grad_norm": 3.691915988922119, | |
| "learning_rate": 7.736842105263158e-06, | |
| "loss": 0.2333, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.9259259259259259, | |
| "grad_norm": 3.742799758911133, | |
| "learning_rate": 7.710526315789474e-06, | |
| "loss": 0.2318, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.9345794392523364, | |
| "grad_norm": 3.024702310562134, | |
| "learning_rate": 7.68421052631579e-06, | |
| "loss": 0.2229, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.943232952578747, | |
| "grad_norm": 2.7311902046203613, | |
| "learning_rate": 7.657894736842106e-06, | |
| "loss": 0.201, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.9518864659051575, | |
| "grad_norm": 4.097088813781738, | |
| "learning_rate": 7.631578947368423e-06, | |
| "loss": 0.2375, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.960539979231568, | |
| "grad_norm": 3.8325319290161133, | |
| "learning_rate": 7.605263157894738e-06, | |
| "loss": 0.256, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.9691934925579785, | |
| "grad_norm": 3.960142135620117, | |
| "learning_rate": 7.578947368421054e-06, | |
| "loss": 0.2536, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.9778470058843891, | |
| "grad_norm": 2.902967929840088, | |
| "learning_rate": 7.552631578947369e-06, | |
| "loss": 0.2343, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.9865005192107996, | |
| "grad_norm": 3.7276077270507812, | |
| "learning_rate": 7.526315789473685e-06, | |
| "loss": 0.2103, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.9951540325372101, | |
| "grad_norm": 3.44114089012146, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.2212, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.0038075458636206, | |
| "grad_norm": 2.8075897693634033, | |
| "learning_rate": 7.473684210526316e-06, | |
| "loss": 0.1869, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.0124610591900312, | |
| "grad_norm": 2.7247374057769775, | |
| "learning_rate": 7.447368421052632e-06, | |
| "loss": 0.1468, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.0211145725164417, | |
| "grad_norm": 2.5912511348724365, | |
| "learning_rate": 7.421052631578948e-06, | |
| "loss": 0.1507, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.0297680858428522, | |
| "grad_norm": 3.2269246578216553, | |
| "learning_rate": 7.3947368421052635e-06, | |
| "loss": 0.1598, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.0384215991692627, | |
| "grad_norm": 2.3670835494995117, | |
| "learning_rate": 7.368421052631579e-06, | |
| "loss": 0.1301, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.0384215991692627, | |
| "eval_loss": 0.21585418283939362, | |
| "eval_runtime": 6825.6837, | |
| "eval_samples_per_second": 1.505, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 17.737815822173133, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.0470751124956732, | |
| "grad_norm": 3.005739450454712, | |
| "learning_rate": 7.342105263157895e-06, | |
| "loss": 0.1571, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.0557286258220837, | |
| "grad_norm": 3.31783127784729, | |
| "learning_rate": 7.315789473684212e-06, | |
| "loss": 0.1368, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.0643821391484942, | |
| "grad_norm": 2.8186709880828857, | |
| "learning_rate": 7.289473684210528e-06, | |
| "loss": 0.1494, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.0730356524749047, | |
| "grad_norm": 3.0247764587402344, | |
| "learning_rate": 7.263157894736843e-06, | |
| "loss": 0.1382, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.0816891658013152, | |
| "grad_norm": 3.01029372215271, | |
| "learning_rate": 7.236842105263158e-06, | |
| "loss": 0.1518, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.0903426791277258, | |
| "grad_norm": 3.1356372833251953, | |
| "learning_rate": 7.210526315789474e-06, | |
| "loss": 0.1557, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.0989961924541363, | |
| "grad_norm": 3.427821636199951, | |
| "learning_rate": 7.18421052631579e-06, | |
| "loss": 0.1672, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.107649705780547, | |
| "grad_norm": 2.804636240005493, | |
| "learning_rate": 7.157894736842106e-06, | |
| "loss": 0.1489, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.1163032191069575, | |
| "grad_norm": 3.353768825531006, | |
| "learning_rate": 7.131578947368422e-06, | |
| "loss": 0.1446, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.124956732433368, | |
| "grad_norm": 2.2061870098114014, | |
| "learning_rate": 7.1052631578947375e-06, | |
| "loss": 0.1392, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.1336102457597785, | |
| "grad_norm": 3.4037885665893555, | |
| "learning_rate": 7.078947368421053e-06, | |
| "loss": 0.1361, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.142263759086189, | |
| "grad_norm": 3.9204328060150146, | |
| "learning_rate": 7.052631578947369e-06, | |
| "loss": 0.154, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.1509172724125996, | |
| "grad_norm": 2.112093925476074, | |
| "learning_rate": 7.026315789473684e-06, | |
| "loss": 0.1636, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.15957078573901, | |
| "grad_norm": 2.0618863105773926, | |
| "learning_rate": 7e-06, | |
| "loss": 0.1382, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.1682242990654206, | |
| "grad_norm": 3.1674695014953613, | |
| "learning_rate": 6.973684210526316e-06, | |
| "loss": 0.1437, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.176877812391831, | |
| "grad_norm": 3.2779016494750977, | |
| "learning_rate": 6.947368421052632e-06, | |
| "loss": 0.1421, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.1855313257182416, | |
| "grad_norm": 2.3959224224090576, | |
| "learning_rate": 6.921052631578948e-06, | |
| "loss": 0.1366, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.1941848390446521, | |
| "grad_norm": 3.081360340118408, | |
| "learning_rate": 6.894736842105264e-06, | |
| "loss": 0.1425, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.2028383523710626, | |
| "grad_norm": 2.6062395572662354, | |
| "learning_rate": 6.86842105263158e-06, | |
| "loss": 0.1413, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.2114918656974731, | |
| "grad_norm": 3.2448296546936035, | |
| "learning_rate": 6.842105263157896e-06, | |
| "loss": 0.1391, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.2114918656974731, | |
| "eval_loss": 0.21471112966537476, | |
| "eval_runtime": 6831.4547, | |
| "eval_samples_per_second": 1.504, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 18.2641861107276, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.2201453790238836, | |
| "grad_norm": 2.590219259262085, | |
| "learning_rate": 6.8157894736842115e-06, | |
| "loss": 0.1417, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.2287988923502942, | |
| "grad_norm": 1.8229866027832031, | |
| "learning_rate": 6.789473684210527e-06, | |
| "loss": 0.1407, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.2374524056767047, | |
| "grad_norm": 2.4515280723571777, | |
| "learning_rate": 6.763157894736842e-06, | |
| "loss": 0.1587, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.2461059190031152, | |
| "grad_norm": 2.929910182952881, | |
| "learning_rate": 6.736842105263158e-06, | |
| "loss": 0.1626, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.254759432329526, | |
| "grad_norm": 2.7540080547332764, | |
| "learning_rate": 6.710526315789474e-06, | |
| "loss": 0.1714, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.2634129456559364, | |
| "grad_norm": 3.262491226196289, | |
| "learning_rate": 6.68421052631579e-06, | |
| "loss": 0.1474, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.272066458982347, | |
| "grad_norm": 2.461406946182251, | |
| "learning_rate": 6.6578947368421055e-06, | |
| "loss": 0.1451, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.2807199723087574, | |
| "grad_norm": 2.6654164791107178, | |
| "learning_rate": 6.631578947368421e-06, | |
| "loss": 0.1598, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.289373485635168, | |
| "grad_norm": 2.850276231765747, | |
| "learning_rate": 6.605263157894738e-06, | |
| "loss": 0.1452, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.2980269989615785, | |
| "grad_norm": 3.434420585632324, | |
| "learning_rate": 6.578947368421054e-06, | |
| "loss": 0.149, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.306680512287989, | |
| "grad_norm": 2.520908832550049, | |
| "learning_rate": 6.55263157894737e-06, | |
| "loss": 0.1444, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.3153340256143995, | |
| "grad_norm": 1.7677472829818726, | |
| "learning_rate": 6.526315789473685e-06, | |
| "loss": 0.1589, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.32398753894081, | |
| "grad_norm": 2.4634532928466797, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.1491, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.3326410522672205, | |
| "grad_norm": 2.644317626953125, | |
| "learning_rate": 6.473684210526316e-06, | |
| "loss": 0.1546, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.341294565593631, | |
| "grad_norm": 2.459674596786499, | |
| "learning_rate": 6.447368421052632e-06, | |
| "loss": 0.1411, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.3499480789200415, | |
| "grad_norm": 2.2588868141174316, | |
| "learning_rate": 6.421052631578948e-06, | |
| "loss": 0.1335, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.358601592246452, | |
| "grad_norm": 3.8076562881469727, | |
| "learning_rate": 6.394736842105264e-06, | |
| "loss": 0.1456, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.3672551055728626, | |
| "grad_norm": 2.1609466075897217, | |
| "learning_rate": 6.3684210526315795e-06, | |
| "loss": 0.1519, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.375908618899273, | |
| "grad_norm": 2.537875175476074, | |
| "learning_rate": 6.342105263157895e-06, | |
| "loss": 0.1435, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.3845621322256836, | |
| "grad_norm": 2.0628812313079834, | |
| "learning_rate": 6.31578947368421e-06, | |
| "loss": 0.1406, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.3845621322256836, | |
| "eval_loss": 0.2134682536125183, | |
| "eval_runtime": 6847.272, | |
| "eval_samples_per_second": 1.5, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 18.063992820654427, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.393215645552094, | |
| "grad_norm": 2.8480277061462402, | |
| "learning_rate": 6.289473684210526e-06, | |
| "loss": 0.1569, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.4018691588785046, | |
| "grad_norm": 2.712129831314087, | |
| "learning_rate": 6.263157894736842e-06, | |
| "loss": 0.1417, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.4105226722049151, | |
| "grad_norm": 2.2308459281921387, | |
| "learning_rate": 6.236842105263159e-06, | |
| "loss": 0.1348, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.4191761855313256, | |
| "grad_norm": 2.9454047679901123, | |
| "learning_rate": 6.2105263157894745e-06, | |
| "loss": 0.1396, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.4278296988577361, | |
| "grad_norm": 3.238121509552002, | |
| "learning_rate": 6.18421052631579e-06, | |
| "loss": 0.1491, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.4364832121841467, | |
| "grad_norm": 2.46944260597229, | |
| "learning_rate": 6.157894736842106e-06, | |
| "loss": 0.1362, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.4451367255105572, | |
| "grad_norm": 2.5275824069976807, | |
| "learning_rate": 6.131578947368422e-06, | |
| "loss": 0.131, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.4537902388369677, | |
| "grad_norm": 3.005089521408081, | |
| "learning_rate": 6.105263157894738e-06, | |
| "loss": 0.1588, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.4624437521633784, | |
| "grad_norm": 3.0360686779022217, | |
| "learning_rate": 6.0789473684210535e-06, | |
| "loss": 0.1565, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.471097265489789, | |
| "grad_norm": 2.272226333618164, | |
| "learning_rate": 6.0526315789473685e-06, | |
| "loss": 0.1478, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.4797507788161994, | |
| "grad_norm": 2.23938250541687, | |
| "learning_rate": 6.026315789473684e-06, | |
| "loss": 0.1523, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.48840429214261, | |
| "grad_norm": 3.1530332565307617, | |
| "learning_rate": 6e-06, | |
| "loss": 0.1462, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.4970578054690205, | |
| "grad_norm": 2.3748743534088135, | |
| "learning_rate": 5.973684210526316e-06, | |
| "loss": 0.1549, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.505711318795431, | |
| "grad_norm": 2.5215299129486084, | |
| "learning_rate": 5.947368421052632e-06, | |
| "loss": 0.1418, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.5143648321218415, | |
| "grad_norm": 1.9181177616119385, | |
| "learning_rate": 5.921052631578948e-06, | |
| "loss": 0.1452, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.523018345448252, | |
| "grad_norm": 3.432988166809082, | |
| "learning_rate": 5.8947368421052634e-06, | |
| "loss": 0.1692, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.5316718587746625, | |
| "grad_norm": 2.3905420303344727, | |
| "learning_rate": 5.86842105263158e-06, | |
| "loss": 0.1428, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.540325372101073, | |
| "grad_norm": 2.4743220806121826, | |
| "learning_rate": 5.842105263157896e-06, | |
| "loss": 0.1487, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.5489788854274835, | |
| "grad_norm": 3.31463623046875, | |
| "learning_rate": 5.815789473684212e-06, | |
| "loss": 0.154, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.557632398753894, | |
| "grad_norm": 2.186415672302246, | |
| "learning_rate": 5.789473684210527e-06, | |
| "loss": 0.1421, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.557632398753894, | |
| "eval_loss": 0.20704275369644165, | |
| "eval_runtime": 6831.8854, | |
| "eval_samples_per_second": 1.504, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 17.19073588292144, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.5662859120803048, | |
| "grad_norm": 1.9734889268875122, | |
| "learning_rate": 5.7631578947368425e-06, | |
| "loss": 0.1348, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.5749394254067153, | |
| "grad_norm": 3.3035688400268555, | |
| "learning_rate": 5.736842105263158e-06, | |
| "loss": 0.1294, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.5835929387331258, | |
| "grad_norm": 2.1477127075195312, | |
| "learning_rate": 5.710526315789474e-06, | |
| "loss": 0.1463, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.5922464520595363, | |
| "grad_norm": 2.911785364151001, | |
| "learning_rate": 5.68421052631579e-06, | |
| "loss": 0.1425, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.6008999653859468, | |
| "grad_norm": 2.9243428707122803, | |
| "learning_rate": 5.657894736842106e-06, | |
| "loss": 0.1437, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.6095534787123573, | |
| "grad_norm": 1.9402472972869873, | |
| "learning_rate": 5.631578947368422e-06, | |
| "loss": 0.1448, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.6182069920387678, | |
| "grad_norm": 2.558166980743408, | |
| "learning_rate": 5.605263157894737e-06, | |
| "loss": 0.162, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.6268605053651783, | |
| "grad_norm": 2.497208833694458, | |
| "learning_rate": 5.578947368421052e-06, | |
| "loss": 0.1534, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.6355140186915889, | |
| "grad_norm": 3.3218460083007812, | |
| "learning_rate": 5.552631578947368e-06, | |
| "loss": 0.1491, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.6441675320179994, | |
| "grad_norm": 2.5457661151885986, | |
| "learning_rate": 5.526315789473685e-06, | |
| "loss": 0.1491, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.6528210453444099, | |
| "grad_norm": 2.2707815170288086, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.1363, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.6614745586708204, | |
| "grad_norm": 2.816852569580078, | |
| "learning_rate": 5.4736842105263165e-06, | |
| "loss": 0.1404, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.670128071997231, | |
| "grad_norm": 2.974275827407837, | |
| "learning_rate": 5.447368421052632e-06, | |
| "loss": 0.1372, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.6787815853236414, | |
| "grad_norm": 2.9549601078033447, | |
| "learning_rate": 5.421052631578948e-06, | |
| "loss": 0.1342, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.687435098650052, | |
| "grad_norm": 3.5474634170532227, | |
| "learning_rate": 5.394736842105264e-06, | |
| "loss": 0.1481, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.6960886119764624, | |
| "grad_norm": 3.140803813934326, | |
| "learning_rate": 5.36842105263158e-06, | |
| "loss": 0.1406, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.704742125302873, | |
| "grad_norm": 3.267249584197998, | |
| "learning_rate": 5.342105263157895e-06, | |
| "loss": 0.1503, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.7133956386292835, | |
| "grad_norm": 3.2628836631774902, | |
| "learning_rate": 5.315789473684211e-06, | |
| "loss": 0.154, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.722049151955694, | |
| "grad_norm": 3.291492462158203, | |
| "learning_rate": 5.289473684210526e-06, | |
| "loss": 0.1526, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.7307026652821045, | |
| "grad_norm": 2.379425525665283, | |
| "learning_rate": 5.263157894736842e-06, | |
| "loss": 0.1427, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.7307026652821045, | |
| "eval_loss": 0.20547065138816833, | |
| "eval_runtime": 6830.2168, | |
| "eval_samples_per_second": 1.504, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.991405494960652, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.739356178608515, | |
| "grad_norm": 2.3537607192993164, | |
| "learning_rate": 5.236842105263158e-06, | |
| "loss": 0.1389, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.7480096919349255, | |
| "grad_norm": 2.943369150161743, | |
| "learning_rate": 5.210526315789474e-06, | |
| "loss": 0.1521, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.756663205261336, | |
| "grad_norm": 3.2738773822784424, | |
| "learning_rate": 5.18421052631579e-06, | |
| "loss": 0.1601, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.7653167185877465, | |
| "grad_norm": 2.1923158168792725, | |
| "learning_rate": 5.157894736842106e-06, | |
| "loss": 0.1631, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.773970231914157, | |
| "grad_norm": 3.9332115650177, | |
| "learning_rate": 5.131578947368422e-06, | |
| "loss": 0.1455, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.7826237452405675, | |
| "grad_norm": 2.184039354324341, | |
| "learning_rate": 5.105263157894738e-06, | |
| "loss": 0.1565, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.791277258566978, | |
| "grad_norm": 3.008984327316284, | |
| "learning_rate": 5.078947368421053e-06, | |
| "loss": 0.1395, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.7999307718933886, | |
| "grad_norm": 2.965576648712158, | |
| "learning_rate": 5.052631578947369e-06, | |
| "loss": 0.1398, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.808584285219799, | |
| "grad_norm": 2.224731922149658, | |
| "learning_rate": 5.026315789473685e-06, | |
| "loss": 0.1327, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.8172377985462098, | |
| "grad_norm": 2.2881205081939697, | |
| "learning_rate": 5e-06, | |
| "loss": 0.1515, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.8258913118726203, | |
| "grad_norm": 2.452091932296753, | |
| "learning_rate": 4.973684210526316e-06, | |
| "loss": 0.1433, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.8345448251990308, | |
| "grad_norm": 2.7906880378723145, | |
| "learning_rate": 4.947368421052632e-06, | |
| "loss": 0.1459, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.8431983385254413, | |
| "grad_norm": 2.501685619354248, | |
| "learning_rate": 4.921052631578948e-06, | |
| "loss": 0.1471, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.8518518518518519, | |
| "grad_norm": 2.82496976852417, | |
| "learning_rate": 4.894736842105264e-06, | |
| "loss": 0.1339, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.8605053651782624, | |
| "grad_norm": 2.912050485610962, | |
| "learning_rate": 4.8684210526315795e-06, | |
| "loss": 0.1359, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.8691588785046729, | |
| "grad_norm": 2.363729953765869, | |
| "learning_rate": 4.842105263157895e-06, | |
| "loss": 0.1577, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.8778123918310834, | |
| "grad_norm": 2.8046460151672363, | |
| "learning_rate": 4.815789473684211e-06, | |
| "loss": 0.1519, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.886465905157494, | |
| "grad_norm": 3.004221200942993, | |
| "learning_rate": 4.789473684210527e-06, | |
| "loss": 0.1486, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.8951194184839044, | |
| "grad_norm": 2.8844528198242188, | |
| "learning_rate": 4.763157894736842e-06, | |
| "loss": 0.1309, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.9037729318103151, | |
| "grad_norm": 2.8573217391967773, | |
| "learning_rate": 4.736842105263158e-06, | |
| "loss": 0.1429, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.9037729318103151, | |
| "eval_loss": 0.20240795612335205, | |
| "eval_runtime": 6794.1597, | |
| "eval_samples_per_second": 1.512, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.77567996686456, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.9124264451367257, | |
| "grad_norm": 2.1657752990722656, | |
| "learning_rate": 4.710526315789474e-06, | |
| "loss": 0.1384, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.9210799584631362, | |
| "grad_norm": 2.8199362754821777, | |
| "learning_rate": 4.68421052631579e-06, | |
| "loss": 0.1383, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.9297334717895467, | |
| "grad_norm": 2.596308469772339, | |
| "learning_rate": 4.657894736842106e-06, | |
| "loss": 0.1368, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.9383869851159572, | |
| "grad_norm": 2.499068260192871, | |
| "learning_rate": 4.631578947368421e-06, | |
| "loss": 0.1425, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.9470404984423677, | |
| "grad_norm": 2.7802367210388184, | |
| "learning_rate": 4.605263157894737e-06, | |
| "loss": 0.1381, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.9556940117687782, | |
| "grad_norm": 2.1181230545043945, | |
| "learning_rate": 4.578947368421053e-06, | |
| "loss": 0.1427, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.9643475250951887, | |
| "grad_norm": 2.876237630844116, | |
| "learning_rate": 4.552631578947369e-06, | |
| "loss": 0.1325, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.9730010384215992, | |
| "grad_norm": 2.6661200523376465, | |
| "learning_rate": 4.526315789473685e-06, | |
| "loss": 0.1401, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.9816545517480098, | |
| "grad_norm": 2.1553337574005127, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.1479, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.9903080650744203, | |
| "grad_norm": 1.7381737232208252, | |
| "learning_rate": 4.473684210526316e-06, | |
| "loss": 0.1386, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.9989615784008308, | |
| "grad_norm": 2.856213331222534, | |
| "learning_rate": 4.447368421052632e-06, | |
| "loss": 0.147, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 2.0076150917272413, | |
| "grad_norm": 2.158287286758423, | |
| "learning_rate": 4.4210526315789476e-06, | |
| "loss": 0.0941, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.016268605053652, | |
| "grad_norm": 1.5348066091537476, | |
| "learning_rate": 4.394736842105263e-06, | |
| "loss": 0.0839, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 2.0249221183800623, | |
| "grad_norm": 2.0257511138916016, | |
| "learning_rate": 4.368421052631579e-06, | |
| "loss": 0.0836, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.033575631706473, | |
| "grad_norm": 2.420926809310913, | |
| "learning_rate": 4.342105263157895e-06, | |
| "loss": 0.0912, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.0422291450328833, | |
| "grad_norm": 1.870279312133789, | |
| "learning_rate": 4.315789473684211e-06, | |
| "loss": 0.0785, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.050882658359294, | |
| "grad_norm": 2.135033130645752, | |
| "learning_rate": 4.289473684210527e-06, | |
| "loss": 0.0808, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 2.0595361716857044, | |
| "grad_norm": 1.9781912565231323, | |
| "learning_rate": 4.2631578947368425e-06, | |
| "loss": 0.0937, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.068189685012115, | |
| "grad_norm": 2.5464608669281006, | |
| "learning_rate": 4.236842105263158e-06, | |
| "loss": 0.0824, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 2.0768431983385254, | |
| "grad_norm": 2.4031896591186523, | |
| "learning_rate": 4.210526315789474e-06, | |
| "loss": 0.0893, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.0768431983385254, | |
| "eval_loss": 0.20871570706367493, | |
| "eval_runtime": 6782.3344, | |
| "eval_samples_per_second": 1.515, | |
| "eval_steps_per_second": 0.095, | |
| "eval_wer": 16.56685765566754, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.085496711664936, | |
| "grad_norm": 1.7652385234832764, | |
| "learning_rate": 4.18421052631579e-06, | |
| "loss": 0.085, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 2.0941502249913464, | |
| "grad_norm": 2.4497318267822266, | |
| "learning_rate": 4.157894736842106e-06, | |
| "loss": 0.0825, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.102803738317757, | |
| "grad_norm": 1.7798666954040527, | |
| "learning_rate": 4.1315789473684216e-06, | |
| "loss": 0.077, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 2.1114572516441674, | |
| "grad_norm": 1.8536590337753296, | |
| "learning_rate": 4.105263157894737e-06, | |
| "loss": 0.0799, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.120110764970578, | |
| "grad_norm": 1.8329837322235107, | |
| "learning_rate": 4.078947368421053e-06, | |
| "loss": 0.0819, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 2.1287642782969884, | |
| "grad_norm": 1.9826552867889404, | |
| "learning_rate": 4.052631578947368e-06, | |
| "loss": 0.0953, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.137417791623399, | |
| "grad_norm": 1.9027239084243774, | |
| "learning_rate": 4.026315789473684e-06, | |
| "loss": 0.0889, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 2.1460713049498095, | |
| "grad_norm": 1.8616188764572144, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0869, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.15472481827622, | |
| "grad_norm": 1.987467885017395, | |
| "learning_rate": 3.9736842105263165e-06, | |
| "loss": 0.0854, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 2.1633783316026305, | |
| "grad_norm": 1.5226447582244873, | |
| "learning_rate": 3.947368421052632e-06, | |
| "loss": 0.0747, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.172031844929041, | |
| "grad_norm": 2.361544370651245, | |
| "learning_rate": 3.921052631578947e-06, | |
| "loss": 0.0749, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 2.1806853582554515, | |
| "grad_norm": 2.086289882659912, | |
| "learning_rate": 3.894736842105263e-06, | |
| "loss": 0.0824, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.189338871581862, | |
| "grad_norm": 1.9018774032592773, | |
| "learning_rate": 3.868421052631579e-06, | |
| "loss": 0.076, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 2.1979923849082725, | |
| "grad_norm": 2.450321912765503, | |
| "learning_rate": 3.842105263157895e-06, | |
| "loss": 0.0861, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.2066458982346835, | |
| "grad_norm": 2.0160300731658936, | |
| "learning_rate": 3.815789473684211e-06, | |
| "loss": 0.0832, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 2.215299411561094, | |
| "grad_norm": 2.0652830600738525, | |
| "learning_rate": 3.789473684210527e-06, | |
| "loss": 0.0844, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.2239529248875045, | |
| "grad_norm": 1.6188254356384277, | |
| "learning_rate": 3.7631578947368426e-06, | |
| "loss": 0.081, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 2.232606438213915, | |
| "grad_norm": 2.319303035736084, | |
| "learning_rate": 3.736842105263158e-06, | |
| "loss": 0.0894, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.2412599515403255, | |
| "grad_norm": 2.1996448040008545, | |
| "learning_rate": 3.710526315789474e-06, | |
| "loss": 0.0843, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 2.249913464866736, | |
| "grad_norm": 2.6930758953094482, | |
| "learning_rate": 3.6842105263157896e-06, | |
| "loss": 0.0821, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.249913464866736, | |
| "eval_loss": 0.21049684286117554, | |
| "eval_runtime": 6797.122, | |
| "eval_samples_per_second": 1.511, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.584978600027615, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.2585669781931466, | |
| "grad_norm": 2.480389356613159, | |
| "learning_rate": 3.657894736842106e-06, | |
| "loss": 0.0847, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 2.267220491519557, | |
| "grad_norm": 2.3303518295288086, | |
| "learning_rate": 3.6315789473684217e-06, | |
| "loss": 0.0878, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.2758740048459676, | |
| "grad_norm": 2.0243282318115234, | |
| "learning_rate": 3.605263157894737e-06, | |
| "loss": 0.0864, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 2.284527518172378, | |
| "grad_norm": 2.6420583724975586, | |
| "learning_rate": 3.578947368421053e-06, | |
| "loss": 0.0884, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.2931810314987886, | |
| "grad_norm": 2.4340221881866455, | |
| "learning_rate": 3.5526315789473687e-06, | |
| "loss": 0.0894, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 2.301834544825199, | |
| "grad_norm": 2.174543857574463, | |
| "learning_rate": 3.5263157894736846e-06, | |
| "loss": 0.0814, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.3104880581516096, | |
| "grad_norm": 2.3189802169799805, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.0857, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 2.31914157147802, | |
| "grad_norm": 2.114579439163208, | |
| "learning_rate": 3.473684210526316e-06, | |
| "loss": 0.0907, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.3277950848044306, | |
| "grad_norm": 1.8805601596832275, | |
| "learning_rate": 3.447368421052632e-06, | |
| "loss": 0.0869, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 2.336448598130841, | |
| "grad_norm": 1.9643834829330444, | |
| "learning_rate": 3.421052631578948e-06, | |
| "loss": 0.0878, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.3451021114572517, | |
| "grad_norm": 1.8426399230957031, | |
| "learning_rate": 3.3947368421052636e-06, | |
| "loss": 0.0815, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 2.353755624783662, | |
| "grad_norm": 2.246006965637207, | |
| "learning_rate": 3.368421052631579e-06, | |
| "loss": 0.0839, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.3624091381100727, | |
| "grad_norm": 2.0131287574768066, | |
| "learning_rate": 3.342105263157895e-06, | |
| "loss": 0.0773, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 2.371062651436483, | |
| "grad_norm": 1.8699673414230347, | |
| "learning_rate": 3.3157894736842107e-06, | |
| "loss": 0.0834, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.3797161647628937, | |
| "grad_norm": 1.833447813987732, | |
| "learning_rate": 3.289473684210527e-06, | |
| "loss": 0.0817, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 2.3883696780893042, | |
| "grad_norm": 1.5582021474838257, | |
| "learning_rate": 3.2631578947368423e-06, | |
| "loss": 0.0746, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.3970231914157147, | |
| "grad_norm": 1.5744956731796265, | |
| "learning_rate": 3.236842105263158e-06, | |
| "loss": 0.0773, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 2.4056767047421252, | |
| "grad_norm": 2.09190034866333, | |
| "learning_rate": 3.210526315789474e-06, | |
| "loss": 0.0756, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.4143302180685358, | |
| "grad_norm": 2.2656619548797607, | |
| "learning_rate": 3.1842105263157898e-06, | |
| "loss": 0.0814, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 2.4229837313949463, | |
| "grad_norm": 1.941187858581543, | |
| "learning_rate": 3.157894736842105e-06, | |
| "loss": 0.0784, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.4229837313949463, | |
| "eval_loss": 0.21001853048801422, | |
| "eval_runtime": 6836.3983, | |
| "eval_samples_per_second": 1.503, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 17.051808642827556, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.431637244721357, | |
| "grad_norm": 1.3389655351638794, | |
| "learning_rate": 3.131578947368421e-06, | |
| "loss": 0.0818, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 2.4402907580477673, | |
| "grad_norm": 1.9891241788864136, | |
| "learning_rate": 3.1052631578947372e-06, | |
| "loss": 0.0858, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.448944271374178, | |
| "grad_norm": 1.729161024093628, | |
| "learning_rate": 3.078947368421053e-06, | |
| "loss": 0.0792, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 2.4575977847005883, | |
| "grad_norm": 2.3100719451904297, | |
| "learning_rate": 3.052631578947369e-06, | |
| "loss": 0.0773, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.466251298026999, | |
| "grad_norm": 1.6781518459320068, | |
| "learning_rate": 3.0263157894736843e-06, | |
| "loss": 0.0862, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 2.4749048113534093, | |
| "grad_norm": 2.458012342453003, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0901, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.48355832467982, | |
| "grad_norm": 2.350067377090454, | |
| "learning_rate": 2.973684210526316e-06, | |
| "loss": 0.0859, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 2.4922118380062304, | |
| "grad_norm": 2.4465065002441406, | |
| "learning_rate": 2.9473684210526317e-06, | |
| "loss": 0.0793, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.5008653513326413, | |
| "grad_norm": 2.160317897796631, | |
| "learning_rate": 2.921052631578948e-06, | |
| "loss": 0.0847, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 2.509518864659052, | |
| "grad_norm": 1.6664947271347046, | |
| "learning_rate": 2.8947368421052634e-06, | |
| "loss": 0.0941, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.5181723779854623, | |
| "grad_norm": 2.054807424545288, | |
| "learning_rate": 2.868421052631579e-06, | |
| "loss": 0.0811, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 2.526825891311873, | |
| "grad_norm": 2.8626656532287598, | |
| "learning_rate": 2.842105263157895e-06, | |
| "loss": 0.0893, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.5354794046382834, | |
| "grad_norm": 2.172909736633301, | |
| "learning_rate": 2.815789473684211e-06, | |
| "loss": 0.0828, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 2.544132917964694, | |
| "grad_norm": 2.266849994659424, | |
| "learning_rate": 2.789473684210526e-06, | |
| "loss": 0.0763, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 2.5527864312911044, | |
| "grad_norm": 1.8911911249160767, | |
| "learning_rate": 2.7631578947368424e-06, | |
| "loss": 0.0839, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 2.561439944617515, | |
| "grad_norm": 2.2773964405059814, | |
| "learning_rate": 2.7368421052631583e-06, | |
| "loss": 0.0773, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.5700934579439254, | |
| "grad_norm": 1.95570707321167, | |
| "learning_rate": 2.710526315789474e-06, | |
| "loss": 0.0835, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 2.578746971270336, | |
| "grad_norm": 1.5412108898162842, | |
| "learning_rate": 2.68421052631579e-06, | |
| "loss": 0.0847, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 2.5874004845967464, | |
| "grad_norm": 2.286245107650757, | |
| "learning_rate": 2.6578947368421053e-06, | |
| "loss": 0.0893, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 2.596053997923157, | |
| "grad_norm": 2.308685779571533, | |
| "learning_rate": 2.631578947368421e-06, | |
| "loss": 0.0785, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.596053997923157, | |
| "eval_loss": 0.2088063508272171, | |
| "eval_runtime": 6812.1745, | |
| "eval_samples_per_second": 1.508, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.74634129504349, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.6047075112495675, | |
| "grad_norm": 2.2219200134277344, | |
| "learning_rate": 2.605263157894737e-06, | |
| "loss": 0.0774, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 2.613361024575978, | |
| "grad_norm": 3.492546558380127, | |
| "learning_rate": 2.578947368421053e-06, | |
| "loss": 0.0852, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 2.6220145379023885, | |
| "grad_norm": 1.7338074445724487, | |
| "learning_rate": 2.552631578947369e-06, | |
| "loss": 0.0825, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 2.630668051228799, | |
| "grad_norm": 2.582502603530884, | |
| "learning_rate": 2.5263157894736844e-06, | |
| "loss": 0.0758, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.6393215645552095, | |
| "grad_norm": 3.0539708137512207, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0943, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 2.64797507788162, | |
| "grad_norm": 2.1495137214660645, | |
| "learning_rate": 2.473684210526316e-06, | |
| "loss": 0.0817, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 2.6566285912080305, | |
| "grad_norm": 1.9127858877182007, | |
| "learning_rate": 2.447368421052632e-06, | |
| "loss": 0.0867, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 2.665282104534441, | |
| "grad_norm": 1.831830382347107, | |
| "learning_rate": 2.4210526315789477e-06, | |
| "loss": 0.0768, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.6739356178608515, | |
| "grad_norm": 1.5215665102005005, | |
| "learning_rate": 2.3947368421052635e-06, | |
| "loss": 0.0862, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 2.682589131187262, | |
| "grad_norm": 2.2080304622650146, | |
| "learning_rate": 2.368421052631579e-06, | |
| "loss": 0.0876, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 2.6912426445136726, | |
| "grad_norm": 3.0281341075897217, | |
| "learning_rate": 2.342105263157895e-06, | |
| "loss": 0.0815, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 2.699896157840083, | |
| "grad_norm": 2.798521041870117, | |
| "learning_rate": 2.3157894736842105e-06, | |
| "loss": 0.0965, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.7085496711664936, | |
| "grad_norm": 2.5977230072021484, | |
| "learning_rate": 2.2894736842105263e-06, | |
| "loss": 0.0878, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 2.717203184492904, | |
| "grad_norm": 2.576024055480957, | |
| "learning_rate": 2.2631578947368426e-06, | |
| "loss": 0.0848, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 2.7258566978193146, | |
| "grad_norm": 2.787841796875, | |
| "learning_rate": 2.236842105263158e-06, | |
| "loss": 0.0867, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 2.734510211145725, | |
| "grad_norm": 3.0872480869293213, | |
| "learning_rate": 2.2105263157894738e-06, | |
| "loss": 0.0931, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.7431637244721356, | |
| "grad_norm": 1.9903494119644165, | |
| "learning_rate": 2.1842105263157896e-06, | |
| "loss": 0.0824, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 2.751817237798546, | |
| "grad_norm": 1.7111889123916626, | |
| "learning_rate": 2.1578947368421054e-06, | |
| "loss": 0.0828, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 2.7604707511249567, | |
| "grad_norm": 2.4591681957244873, | |
| "learning_rate": 2.1315789473684212e-06, | |
| "loss": 0.0847, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 2.769124264451367, | |
| "grad_norm": 1.9559268951416016, | |
| "learning_rate": 2.105263157894737e-06, | |
| "loss": 0.072, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.769124264451367, | |
| "eval_loss": 0.20778915286064148, | |
| "eval_runtime": 6824.3281, | |
| "eval_samples_per_second": 1.505, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.5470109070827, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 2.040600299835205, | |
| "learning_rate": 2.078947368421053e-06, | |
| "loss": 0.0878, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 2.786431291104188, | |
| "grad_norm": 2.3346126079559326, | |
| "learning_rate": 2.0526315789473687e-06, | |
| "loss": 0.0811, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 2.7950848044305987, | |
| "grad_norm": 2.295974016189575, | |
| "learning_rate": 2.026315789473684e-06, | |
| "loss": 0.0793, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 2.803738317757009, | |
| "grad_norm": 1.8700435161590576, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.0791, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.8123918310834197, | |
| "grad_norm": 1.5720834732055664, | |
| "learning_rate": 1.973684210526316e-06, | |
| "loss": 0.0768, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 2.8210453444098302, | |
| "grad_norm": 2.1670758724212646, | |
| "learning_rate": 1.9473684210526315e-06, | |
| "loss": 0.075, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 2.8296988577362407, | |
| "grad_norm": 2.925609588623047, | |
| "learning_rate": 1.9210526315789474e-06, | |
| "loss": 0.0752, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 2.8383523710626513, | |
| "grad_norm": 2.2166748046875, | |
| "learning_rate": 1.8947368421052634e-06, | |
| "loss": 0.0924, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.8470058843890618, | |
| "grad_norm": 2.2777693271636963, | |
| "learning_rate": 1.868421052631579e-06, | |
| "loss": 0.0868, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 2.8556593977154723, | |
| "grad_norm": 2.00960111618042, | |
| "learning_rate": 1.8421052631578948e-06, | |
| "loss": 0.0696, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 2.864312911041883, | |
| "grad_norm": 1.7867509126663208, | |
| "learning_rate": 1.8157894736842109e-06, | |
| "loss": 0.0756, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 2.8729664243682933, | |
| "grad_norm": 1.8508473634719849, | |
| "learning_rate": 1.7894736842105265e-06, | |
| "loss": 0.0887, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.881619937694704, | |
| "grad_norm": 2.443462371826172, | |
| "learning_rate": 1.7631578947368423e-06, | |
| "loss": 0.0819, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 2.8902734510211143, | |
| "grad_norm": 1.782518982887268, | |
| "learning_rate": 1.736842105263158e-06, | |
| "loss": 0.077, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 2.898926964347525, | |
| "grad_norm": 2.163900852203369, | |
| "learning_rate": 1.710526315789474e-06, | |
| "loss": 0.0806, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 2.9075804776739353, | |
| "grad_norm": 2.4447150230407715, | |
| "learning_rate": 1.6842105263157895e-06, | |
| "loss": 0.0807, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.9162339910003463, | |
| "grad_norm": 2.5183095932006836, | |
| "learning_rate": 1.6578947368421053e-06, | |
| "loss": 0.0843, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 2.924887504326757, | |
| "grad_norm": 1.8049046993255615, | |
| "learning_rate": 1.6315789473684212e-06, | |
| "loss": 0.0841, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 2.9335410176531673, | |
| "grad_norm": 2.3873138427734375, | |
| "learning_rate": 1.605263157894737e-06, | |
| "loss": 0.0892, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 2.942194530979578, | |
| "grad_norm": 3.381866693496704, | |
| "learning_rate": 1.5789473684210526e-06, | |
| "loss": 0.086, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.942194530979578, | |
| "eval_loss": 0.20743827521800995, | |
| "eval_runtime": 6784.8177, | |
| "eval_samples_per_second": 1.514, | |
| "eval_steps_per_second": 0.095, | |
| "eval_wer": 16.289003175479774, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.9508480443059883, | |
| "grad_norm": 1.9069478511810303, | |
| "learning_rate": 1.5526315789473686e-06, | |
| "loss": 0.0715, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 2.959501557632399, | |
| "grad_norm": 2.0899953842163086, | |
| "learning_rate": 1.5263157894736844e-06, | |
| "loss": 0.0782, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 2.9681550709588094, | |
| "grad_norm": 2.9195377826690674, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0875, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 2.97680858428522, | |
| "grad_norm": 2.9911153316497803, | |
| "learning_rate": 1.4736842105263159e-06, | |
| "loss": 0.082, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.9854620976116304, | |
| "grad_norm": 2.4869954586029053, | |
| "learning_rate": 1.4473684210526317e-06, | |
| "loss": 0.0798, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 2.994115610938041, | |
| "grad_norm": 2.3882336616516113, | |
| "learning_rate": 1.4210526315789475e-06, | |
| "loss": 0.0711, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 3.0027691242644514, | |
| "grad_norm": 1.6410326957702637, | |
| "learning_rate": 1.394736842105263e-06, | |
| "loss": 0.0807, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 3.011422637590862, | |
| "grad_norm": 0.9989892244338989, | |
| "learning_rate": 1.3684210526315791e-06, | |
| "loss": 0.0455, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 3.0200761509172724, | |
| "grad_norm": 1.602900743484497, | |
| "learning_rate": 1.342105263157895e-06, | |
| "loss": 0.0462, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 3.028729664243683, | |
| "grad_norm": 1.4966108798980713, | |
| "learning_rate": 1.3157894736842106e-06, | |
| "loss": 0.0464, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 3.0373831775700935, | |
| "grad_norm": 1.798976182937622, | |
| "learning_rate": 1.2894736842105266e-06, | |
| "loss": 0.0452, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 3.046036690896504, | |
| "grad_norm": 1.5227916240692139, | |
| "learning_rate": 1.2631578947368422e-06, | |
| "loss": 0.0515, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 3.0546902042229145, | |
| "grad_norm": 1.3021562099456787, | |
| "learning_rate": 1.236842105263158e-06, | |
| "loss": 0.0486, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 3.063343717549325, | |
| "grad_norm": 1.3772845268249512, | |
| "learning_rate": 1.2105263157894738e-06, | |
| "loss": 0.0451, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 3.0719972308757355, | |
| "grad_norm": 2.208716630935669, | |
| "learning_rate": 1.1842105263157894e-06, | |
| "loss": 0.0498, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 3.080650744202146, | |
| "grad_norm": 1.7014740705490112, | |
| "learning_rate": 1.1578947368421053e-06, | |
| "loss": 0.0446, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 3.0893042575285565, | |
| "grad_norm": 1.7459232807159424, | |
| "learning_rate": 1.1315789473684213e-06, | |
| "loss": 0.0442, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 3.097957770854967, | |
| "grad_norm": 1.3665467500686646, | |
| "learning_rate": 1.1052631578947369e-06, | |
| "loss": 0.051, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 3.1066112841813776, | |
| "grad_norm": 1.5824533700942993, | |
| "learning_rate": 1.0789473684210527e-06, | |
| "loss": 0.052, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 3.115264797507788, | |
| "grad_norm": 1.0504260063171387, | |
| "learning_rate": 1.0526315789473685e-06, | |
| "loss": 0.0463, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.115264797507788, | |
| "eval_loss": 0.2164984941482544, | |
| "eval_runtime": 6827.9492, | |
| "eval_samples_per_second": 1.504, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.52112384371117, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.1239183108341986, | |
| "grad_norm": 2.2462034225463867, | |
| "learning_rate": 1.0263157894736843e-06, | |
| "loss": 0.0492, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 3.132571824160609, | |
| "grad_norm": 1.408866286277771, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0483, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 3.1412253374870196, | |
| "grad_norm": 1.656809687614441, | |
| "learning_rate": 9.736842105263158e-07, | |
| "loss": 0.0516, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 3.14987885081343, | |
| "grad_norm": 1.4440449476242065, | |
| "learning_rate": 9.473684210526317e-07, | |
| "loss": 0.0497, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 3.1585323641398406, | |
| "grad_norm": 1.7022453546524048, | |
| "learning_rate": 9.210526315789474e-07, | |
| "loss": 0.0467, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 3.167185877466251, | |
| "grad_norm": 1.5018113851547241, | |
| "learning_rate": 8.947368421052632e-07, | |
| "loss": 0.0476, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 3.1758393907926616, | |
| "grad_norm": 1.9194047451019287, | |
| "learning_rate": 8.68421052631579e-07, | |
| "loss": 0.0487, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 3.184492904119072, | |
| "grad_norm": 1.4882956743240356, | |
| "learning_rate": 8.421052631578948e-07, | |
| "loss": 0.0531, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 3.1931464174454827, | |
| "grad_norm": 1.4099938869476318, | |
| "learning_rate": 8.157894736842106e-07, | |
| "loss": 0.048, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 3.2017999307718936, | |
| "grad_norm": 1.5829529762268066, | |
| "learning_rate": 7.894736842105263e-07, | |
| "loss": 0.0503, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 3.210453444098304, | |
| "grad_norm": 1.280104160308838, | |
| "learning_rate": 7.631578947368422e-07, | |
| "loss": 0.0439, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 3.2191069574247146, | |
| "grad_norm": 1.5087403059005737, | |
| "learning_rate": 7.368421052631579e-07, | |
| "loss": 0.0486, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 3.227760470751125, | |
| "grad_norm": 1.830002784729004, | |
| "learning_rate": 7.105263157894737e-07, | |
| "loss": 0.0497, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 3.2364139840775357, | |
| "grad_norm": 0.9852591156959534, | |
| "learning_rate": 6.842105263157896e-07, | |
| "loss": 0.0526, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 3.245067497403946, | |
| "grad_norm": 1.8316926956176758, | |
| "learning_rate": 6.578947368421053e-07, | |
| "loss": 0.0517, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 3.2537210107303567, | |
| "grad_norm": 1.5552564859390259, | |
| "learning_rate": 6.315789473684211e-07, | |
| "loss": 0.0506, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 3.262374524056767, | |
| "grad_norm": 1.0861077308654785, | |
| "learning_rate": 6.052631578947369e-07, | |
| "loss": 0.0442, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 3.2710280373831777, | |
| "grad_norm": 1.3339163064956665, | |
| "learning_rate": 5.789473684210526e-07, | |
| "loss": 0.0491, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 3.2796815507095882, | |
| "grad_norm": 1.8896433115005493, | |
| "learning_rate": 5.526315789473684e-07, | |
| "loss": 0.0428, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 3.2883350640359987, | |
| "grad_norm": 2.0069425106048584, | |
| "learning_rate": 5.263157894736843e-07, | |
| "loss": 0.047, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.2883350640359987, | |
| "eval_loss": 0.2176118642091751, | |
| "eval_runtime": 6817.142, | |
| "eval_samples_per_second": 1.507, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.54269639652078, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.2969885773624092, | |
| "grad_norm": 1.5549033880233765, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0437, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 3.3056420906888198, | |
| "grad_norm": 1.2936606407165527, | |
| "learning_rate": 4.7368421052631585e-07, | |
| "loss": 0.048, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 3.3142956040152303, | |
| "grad_norm": 1.3265544176101685, | |
| "learning_rate": 4.473684210526316e-07, | |
| "loss": 0.0416, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 3.322949117341641, | |
| "grad_norm": 2.265376567840576, | |
| "learning_rate": 4.210526315789474e-07, | |
| "loss": 0.0462, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 3.3316026306680513, | |
| "grad_norm": 1.4310107231140137, | |
| "learning_rate": 3.9473684210526315e-07, | |
| "loss": 0.0447, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 3.340256143994462, | |
| "grad_norm": 2.204995632171631, | |
| "learning_rate": 3.6842105263157896e-07, | |
| "loss": 0.0492, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 3.3489096573208723, | |
| "grad_norm": 1.6492983102798462, | |
| "learning_rate": 3.421052631578948e-07, | |
| "loss": 0.0477, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 3.357563170647283, | |
| "grad_norm": 1.491167426109314, | |
| "learning_rate": 3.1578947368421055e-07, | |
| "loss": 0.0472, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 3.3662166839736933, | |
| "grad_norm": 1.4204721450805664, | |
| "learning_rate": 2.894736842105263e-07, | |
| "loss": 0.0483, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 3.374870197300104, | |
| "grad_norm": 1.7954652309417725, | |
| "learning_rate": 2.6315789473684213e-07, | |
| "loss": 0.0441, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 3.3835237106265144, | |
| "grad_norm": 1.4561618566513062, | |
| "learning_rate": 2.3684210526315792e-07, | |
| "loss": 0.0477, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 3.392177223952925, | |
| "grad_norm": 1.0472415685653687, | |
| "learning_rate": 2.105263157894737e-07, | |
| "loss": 0.0466, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 3.4008307372793354, | |
| "grad_norm": 1.8670166730880737, | |
| "learning_rate": 1.8421052631578948e-07, | |
| "loss": 0.047, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 3.409484250605746, | |
| "grad_norm": 2.0903148651123047, | |
| "learning_rate": 1.5789473684210527e-07, | |
| "loss": 0.0501, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 3.4181377639321564, | |
| "grad_norm": 1.7378038167953491, | |
| "learning_rate": 1.3157894736842107e-07, | |
| "loss": 0.0499, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 3.426791277258567, | |
| "grad_norm": 1.3576021194458008, | |
| "learning_rate": 1.0526315789473685e-07, | |
| "loss": 0.0485, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 3.4354447905849774, | |
| "grad_norm": 1.5308172702789307, | |
| "learning_rate": 7.894736842105264e-08, | |
| "loss": 0.054, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 3.444098303911388, | |
| "grad_norm": 1.8448562622070312, | |
| "learning_rate": 5.263157894736842e-08, | |
| "loss": 0.0525, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 3.4527518172377984, | |
| "grad_norm": 2.0659971237182617, | |
| "learning_rate": 2.631578947368421e-08, | |
| "loss": 0.0476, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 3.461405330564209, | |
| "grad_norm": 1.5937784910202026, | |
| "learning_rate": 0.0, | |
| "loss": 0.0471, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.461405330564209, | |
| "eval_loss": 0.21720539033412933, | |
| "eval_runtime": 6798.3915, | |
| "eval_samples_per_second": 1.511, | |
| "eval_steps_per_second": 0.094, | |
| "eval_wer": 16.571172166229463, | |
| "step": 10000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.265935704064e+20, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |