| { | |
| "best_metric": 16.64483928516185, | |
| "best_model_checkpoint": "/mnt/e/whisper_trained_98h/checkpoint-9000", | |
| "epoch": 3.1174229303775545, | |
| "eval_steps": 1000, | |
| "global_step": 9000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 35.3004264831543, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 2.2775, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 23.328025817871094, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 1.8641, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 19.40353012084961, | |
| "learning_rate": 1.42e-06, | |
| "loss": 1.6219, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 16.085819244384766, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.2676, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 18.707691192626953, | |
| "learning_rate": 2.42e-06, | |
| "loss": 1.1709, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 12.572211265563965, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.9533, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 13.673868179321289, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.964, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 11.917014122009277, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.8704, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 11.749382019042969, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.8011, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 9.886258125305176, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.5665, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 11.503835678100586, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.5455, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 10.527493476867676, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.5435, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 11.787008285522461, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.4959, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 7.4050798416137695, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.5162, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 9.752595901489258, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.4915, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 8.89893627166748, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.4562, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 8.095244407653809, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.4571, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 10.477778434753418, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.4382, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 8.432029724121094, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.4468, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 11.235993385314941, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.4625, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 8.525556564331055, | |
| "learning_rate": 9.977894736842106e-06, | |
| "loss": 0.4673, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 7.134006500244141, | |
| "learning_rate": 9.951578947368423e-06, | |
| "loss": 0.5232, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 10.446462631225586, | |
| "learning_rate": 9.925263157894738e-06, | |
| "loss": 0.47, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 8.982714653015137, | |
| "learning_rate": 9.898947368421054e-06, | |
| "loss": 0.4395, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 8.84506607055664, | |
| "learning_rate": 9.87263157894737e-06, | |
| "loss": 0.462, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 9.220427513122559, | |
| "learning_rate": 9.846315789473684e-06, | |
| "loss": 0.4209, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 6.004112243652344, | |
| "learning_rate": 9.820000000000001e-06, | |
| "loss": 0.391, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 10.247870445251465, | |
| "learning_rate": 9.793684210526316e-06, | |
| "loss": 0.4134, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 8.254446029663086, | |
| "learning_rate": 9.767368421052632e-06, | |
| "loss": 0.4264, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 10.685470581054688, | |
| "learning_rate": 9.741052631578947e-06, | |
| "loss": 0.4347, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 11.406156539916992, | |
| "learning_rate": 9.714736842105264e-06, | |
| "loss": 0.4281, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 5.957642555236816, | |
| "learning_rate": 9.68842105263158e-06, | |
| "loss": 0.4132, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 8.0116548538208, | |
| "learning_rate": 9.662105263157896e-06, | |
| "loss": 0.3717, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 8.80966567993164, | |
| "learning_rate": 9.635789473684212e-06, | |
| "loss": 0.3659, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 8.115036964416504, | |
| "learning_rate": 9.609473684210527e-06, | |
| "loss": 0.4112, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 8.217317581176758, | |
| "learning_rate": 9.583157894736842e-06, | |
| "loss": 0.4415, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 11.775481224060059, | |
| "learning_rate": 9.556842105263159e-06, | |
| "loss": 0.4344, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 8.445450782775879, | |
| "learning_rate": 9.530526315789474e-06, | |
| "loss": 0.3313, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 9.221593856811523, | |
| "learning_rate": 9.50421052631579e-06, | |
| "loss": 0.4199, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 8.544211387634277, | |
| "learning_rate": 9.477894736842106e-06, | |
| "loss": 0.4042, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_loss": 0.40728408098220825, | |
| "eval_runtime": 1492.3248, | |
| "eval_samples_per_second": 2.731, | |
| "eval_steps_per_second": 0.342, | |
| "eval_wer": 28.237756901405547, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 7.7543230056762695, | |
| "learning_rate": 9.451578947368422e-06, | |
| "loss": 0.3941, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 8.227640151977539, | |
| "learning_rate": 9.425263157894737e-06, | |
| "loss": 0.3652, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 9.201411247253418, | |
| "learning_rate": 9.398947368421052e-06, | |
| "loss": 0.4321, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 8.538446426391602, | |
| "learning_rate": 9.372631578947369e-06, | |
| "loss": 0.4056, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 8.95832633972168, | |
| "learning_rate": 9.346315789473684e-06, | |
| "loss": 0.3858, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 6.771994590759277, | |
| "learning_rate": 9.32e-06, | |
| "loss": 0.3275, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 6.228338718414307, | |
| "learning_rate": 9.293684210526317e-06, | |
| "loss": 0.4061, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 7.886964321136475, | |
| "learning_rate": 9.267368421052632e-06, | |
| "loss": 0.3728, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 6.923808574676514, | |
| "learning_rate": 9.241052631578949e-06, | |
| "loss": 0.3279, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 8.336899757385254, | |
| "learning_rate": 9.214736842105264e-06, | |
| "loss": 0.3777, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 4.622466564178467, | |
| "learning_rate": 9.18842105263158e-06, | |
| "loss": 0.3346, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 6.4272332191467285, | |
| "learning_rate": 9.162105263157895e-06, | |
| "loss": 0.3065, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 7.080216884613037, | |
| "learning_rate": 9.13578947368421e-06, | |
| "loss": 0.3318, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 6.740713596343994, | |
| "learning_rate": 9.109473684210527e-06, | |
| "loss": 0.353, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 7.693573474884033, | |
| "learning_rate": 9.083157894736842e-06, | |
| "loss": 0.337, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 9.819792747497559, | |
| "learning_rate": 9.056842105263159e-06, | |
| "loss": 0.3563, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 8.117258071899414, | |
| "learning_rate": 9.030526315789474e-06, | |
| "loss": 0.3219, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 6.282182693481445, | |
| "learning_rate": 9.00421052631579e-06, | |
| "loss": 0.3688, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 8.659537315368652, | |
| "learning_rate": 8.977894736842107e-06, | |
| "loss": 0.3233, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 5.883866786956787, | |
| "learning_rate": 8.951578947368422e-06, | |
| "loss": 0.3716, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 7.345333576202393, | |
| "learning_rate": 8.925263157894739e-06, | |
| "loss": 0.3583, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 7.333982944488525, | |
| "learning_rate": 8.898947368421054e-06, | |
| "loss": 0.3352, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 7.652650833129883, | |
| "learning_rate": 8.872631578947369e-06, | |
| "loss": 0.314, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 8.704148292541504, | |
| "learning_rate": 8.846315789473685e-06, | |
| "loss": 0.3292, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 6.3192267417907715, | |
| "learning_rate": 8.82e-06, | |
| "loss": 0.3371, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 6.321137428283691, | |
| "learning_rate": 8.793684210526317e-06, | |
| "loss": 0.285, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 6.802587032318115, | |
| "learning_rate": 8.767368421052632e-06, | |
| "loss": 0.3416, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 9.17293930053711, | |
| "learning_rate": 8.741052631578949e-06, | |
| "loss": 0.3493, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 5.472736835479736, | |
| "learning_rate": 8.714736842105264e-06, | |
| "loss": 0.3337, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 8.298070907592773, | |
| "learning_rate": 8.688421052631579e-06, | |
| "loss": 0.2931, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 8.467511177062988, | |
| "learning_rate": 8.662105263157895e-06, | |
| "loss": 0.3279, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 6.326373100280762, | |
| "learning_rate": 8.63578947368421e-06, | |
| "loss": 0.3114, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 6.917250633239746, | |
| "learning_rate": 8.609473684210527e-06, | |
| "loss": 0.3143, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 5.873658180236816, | |
| "learning_rate": 8.583157894736843e-06, | |
| "loss": 0.2895, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 8.249701499938965, | |
| "learning_rate": 8.556842105263158e-06, | |
| "loss": 0.31, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 10.655940055847168, | |
| "learning_rate": 8.530526315789475e-06, | |
| "loss": 0.301, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 6.404816627502441, | |
| "learning_rate": 8.50421052631579e-06, | |
| "loss": 0.3183, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 7.184628486633301, | |
| "learning_rate": 8.477894736842107e-06, | |
| "loss": 0.3114, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 8.222028732299805, | |
| "learning_rate": 8.451578947368422e-06, | |
| "loss": 0.323, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 6.1425981521606445, | |
| "learning_rate": 8.425263157894737e-06, | |
| "loss": 0.2901, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "eval_loss": 0.357265442609787, | |
| "eval_runtime": 1354.7835, | |
| "eval_samples_per_second": 3.008, | |
| "eval_steps_per_second": 0.376, | |
| "eval_wer": 23.303393728151136, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 9.46043872833252, | |
| "learning_rate": 8.398947368421053e-06, | |
| "loss": 0.3116, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 8.215323448181152, | |
| "learning_rate": 8.372631578947368e-06, | |
| "loss": 0.3106, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 4.965717792510986, | |
| "learning_rate": 8.346315789473685e-06, | |
| "loss": 0.3135, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 6.425980567932129, | |
| "learning_rate": 8.32e-06, | |
| "loss": 0.2675, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 7.774525165557861, | |
| "learning_rate": 8.293684210526317e-06, | |
| "loss": 0.3058, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 4.193319797515869, | |
| "learning_rate": 8.267368421052632e-06, | |
| "loss": 0.2906, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 6.157843112945557, | |
| "learning_rate": 8.241052631578948e-06, | |
| "loss": 0.2695, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 6.667449474334717, | |
| "learning_rate": 8.214736842105265e-06, | |
| "loss": 0.2736, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 5.1777143478393555, | |
| "learning_rate": 8.18842105263158e-06, | |
| "loss": 0.3096, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 5.825611114501953, | |
| "learning_rate": 8.162105263157895e-06, | |
| "loss": 0.3117, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 8.687291145324707, | |
| "learning_rate": 8.135789473684212e-06, | |
| "loss": 0.2631, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 7.267210006713867, | |
| "learning_rate": 8.109473684210527e-06, | |
| "loss": 0.2446, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 7.924586296081543, | |
| "learning_rate": 8.083157894736843e-06, | |
| "loss": 0.285, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 5.925978183746338, | |
| "learning_rate": 8.056842105263158e-06, | |
| "loss": 0.2779, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 7.552637577056885, | |
| "learning_rate": 8.030526315789475e-06, | |
| "loss": 0.275, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 7.047172546386719, | |
| "learning_rate": 8.00421052631579e-06, | |
| "loss": 0.326, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 6.557082653045654, | |
| "learning_rate": 7.977894736842105e-06, | |
| "loss": 0.317, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 7.485312461853027, | |
| "learning_rate": 7.951578947368421e-06, | |
| "loss": 0.277, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 6.1593122482299805, | |
| "learning_rate": 7.925263157894736e-06, | |
| "loss": 0.2773, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 8.755908012390137, | |
| "learning_rate": 7.898947368421053e-06, | |
| "loss": 0.3042, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 7.109035968780518, | |
| "learning_rate": 7.87263157894737e-06, | |
| "loss": 0.2742, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 6.4371466636657715, | |
| "learning_rate": 7.846315789473685e-06, | |
| "loss": 0.2712, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 3.935817003250122, | |
| "learning_rate": 7.820000000000001e-06, | |
| "loss": 0.3128, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 6.779932022094727, | |
| "learning_rate": 7.793684210526316e-06, | |
| "loss": 0.312, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 6.750608921051025, | |
| "learning_rate": 7.767368421052633e-06, | |
| "loss": 0.3241, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 6.122084140777588, | |
| "learning_rate": 7.741052631578948e-06, | |
| "loss": 0.3062, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 8.169845581054688, | |
| "learning_rate": 7.714736842105263e-06, | |
| "loss": 0.2589, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 7.748149394989014, | |
| "learning_rate": 7.68842105263158e-06, | |
| "loss": 0.2647, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 4.9124321937561035, | |
| "learning_rate": 7.662105263157895e-06, | |
| "loss": 0.2586, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 6.429450511932373, | |
| "learning_rate": 7.635789473684211e-06, | |
| "loss": 0.2527, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 6.546786785125732, | |
| "learning_rate": 7.609473684210526e-06, | |
| "loss": 0.2399, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 7.384542465209961, | |
| "learning_rate": 7.583157894736842e-06, | |
| "loss": 0.315, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 5.672204494476318, | |
| "learning_rate": 7.556842105263158e-06, | |
| "loss": 0.2657, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 6.416686534881592, | |
| "learning_rate": 7.5305263157894745e-06, | |
| "loss": 0.2522, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 3.909656047821045, | |
| "learning_rate": 7.50421052631579e-06, | |
| "loss": 0.3207, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 5.621556282043457, | |
| "learning_rate": 7.477894736842106e-06, | |
| "loss": 0.1792, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "grad_norm": 6.261754989624023, | |
| "learning_rate": 7.451578947368422e-06, | |
| "loss": 0.163, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 4.041684150695801, | |
| "learning_rate": 7.425263157894738e-06, | |
| "loss": 0.1421, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "grad_norm": 4.004493236541748, | |
| "learning_rate": 7.398947368421054e-06, | |
| "loss": 0.1289, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 7.685792922973633, | |
| "learning_rate": 7.3726315789473694e-06, | |
| "loss": 0.1665, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_loss": 0.3465397357940674, | |
| "eval_runtime": 1333.1728, | |
| "eval_samples_per_second": 3.057, | |
| "eval_steps_per_second": 0.383, | |
| "eval_wer": 21.44411330989038, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "grad_norm": 5.127060413360596, | |
| "learning_rate": 7.346315789473684e-06, | |
| "loss": 0.1783, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 4.140434265136719, | |
| "learning_rate": 7.32e-06, | |
| "loss": 0.1392, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 7.5522942543029785, | |
| "learning_rate": 7.293684210526316e-06, | |
| "loss": 0.1326, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "grad_norm": 4.477723121643066, | |
| "learning_rate": 7.267368421052632e-06, | |
| "loss": 0.1638, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 4.424250602722168, | |
| "learning_rate": 7.241052631578948e-06, | |
| "loss": 0.1605, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "grad_norm": 6.303280353546143, | |
| "learning_rate": 7.2147368421052635e-06, | |
| "loss": 0.1181, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 4.989869594573975, | |
| "learning_rate": 7.18842105263158e-06, | |
| "loss": 0.1346, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "grad_norm": 5.7114763259887695, | |
| "learning_rate": 7.162105263157896e-06, | |
| "loss": 0.1501, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 5.886032581329346, | |
| "learning_rate": 7.135789473684212e-06, | |
| "loss": 0.1282, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 10.488823890686035, | |
| "learning_rate": 7.109473684210528e-06, | |
| "loss": 0.1787, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "grad_norm": 5.663763523101807, | |
| "learning_rate": 7.083157894736843e-06, | |
| "loss": 0.1483, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "grad_norm": 8.371013641357422, | |
| "learning_rate": 7.056842105263158e-06, | |
| "loss": 0.1458, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "grad_norm": 5.267305850982666, | |
| "learning_rate": 7.030526315789474e-06, | |
| "loss": 0.1435, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 6.178240776062012, | |
| "learning_rate": 7.00421052631579e-06, | |
| "loss": 0.1498, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "grad_norm": 5.349623680114746, | |
| "learning_rate": 6.977894736842106e-06, | |
| "loss": 0.1396, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 6.444940090179443, | |
| "learning_rate": 6.951578947368422e-06, | |
| "loss": 0.1414, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "grad_norm": 5.608889102935791, | |
| "learning_rate": 6.9252631578947375e-06, | |
| "loss": 0.1826, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 4.847830295562744, | |
| "learning_rate": 6.8989473684210525e-06, | |
| "loss": 0.1453, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 5.886878967285156, | |
| "learning_rate": 6.872631578947368e-06, | |
| "loss": 0.1619, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "grad_norm": 6.402673721313477, | |
| "learning_rate": 6.846315789473684e-06, | |
| "loss": 0.13, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 5.007193088531494, | |
| "learning_rate": 6.820000000000001e-06, | |
| "loss": 0.1579, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "grad_norm": 5.040092945098877, | |
| "learning_rate": 6.793684210526317e-06, | |
| "loss": 0.1287, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 6.6516242027282715, | |
| "learning_rate": 6.767368421052632e-06, | |
| "loss": 0.1547, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 5.129918575286865, | |
| "learning_rate": 6.741052631578948e-06, | |
| "loss": 0.1403, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 6.9857497215271, | |
| "learning_rate": 6.714736842105264e-06, | |
| "loss": 0.1396, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 3.786760091781616, | |
| "learning_rate": 6.68842105263158e-06, | |
| "loss": 0.1621, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "grad_norm": 3.897284746170044, | |
| "learning_rate": 6.662105263157896e-06, | |
| "loss": 0.1374, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 4.193057537078857, | |
| "learning_rate": 6.635789473684211e-06, | |
| "loss": 0.1192, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "grad_norm": 5.446070671081543, | |
| "learning_rate": 6.6094736842105265e-06, | |
| "loss": 0.1536, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 8.413087844848633, | |
| "learning_rate": 6.583157894736842e-06, | |
| "loss": 0.129, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "grad_norm": 2.960362672805786, | |
| "learning_rate": 6.556842105263158e-06, | |
| "loss": 0.1487, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 4.482843399047852, | |
| "learning_rate": 6.530526315789474e-06, | |
| "loss": 0.1193, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 4.52742862701416, | |
| "learning_rate": 6.50421052631579e-06, | |
| "loss": 0.1734, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "grad_norm": 5.126417636871338, | |
| "learning_rate": 6.477894736842106e-06, | |
| "loss": 0.1406, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 6.627448081970215, | |
| "learning_rate": 6.451578947368422e-06, | |
| "loss": 0.1354, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "grad_norm": 5.687572479248047, | |
| "learning_rate": 6.425263157894738e-06, | |
| "loss": 0.1179, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "grad_norm": 5.633472442626953, | |
| "learning_rate": 6.398947368421054e-06, | |
| "loss": 0.139, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "grad_norm": 3.897115468978882, | |
| "learning_rate": 6.372631578947369e-06, | |
| "loss": 0.1432, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 5.5287065505981445, | |
| "learning_rate": 6.346315789473685e-06, | |
| "loss": 0.136, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 7.957043647766113, | |
| "learning_rate": 6.3200000000000005e-06, | |
| "loss": 0.1451, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "eval_loss": 0.34090888500213623, | |
| "eval_runtime": 1313.1912, | |
| "eval_samples_per_second": 3.103, | |
| "eval_steps_per_second": 0.388, | |
| "eval_wer": 19.001165272807896, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "grad_norm": 5.235450267791748, | |
| "learning_rate": 6.293684210526316e-06, | |
| "loss": 0.1558, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 5.860323905944824, | |
| "learning_rate": 6.267368421052632e-06, | |
| "loss": 0.1345, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "grad_norm": 4.477651119232178, | |
| "learning_rate": 6.241052631578948e-06, | |
| "loss": 0.1924, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 3.619530439376831, | |
| "learning_rate": 6.214736842105264e-06, | |
| "loss": 0.1445, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "grad_norm": 3.86283802986145, | |
| "learning_rate": 6.188421052631579e-06, | |
| "loss": 0.1195, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 4.729098796844482, | |
| "learning_rate": 6.1621052631578945e-06, | |
| "loss": 0.1652, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 4.275981903076172, | |
| "learning_rate": 6.13578947368421e-06, | |
| "loss": 0.15, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "grad_norm": 4.429512023925781, | |
| "learning_rate": 6.109473684210527e-06, | |
| "loss": 0.1422, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 5.349700450897217, | |
| "learning_rate": 6.083157894736843e-06, | |
| "loss": 0.1455, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "grad_norm": 3.825136184692383, | |
| "learning_rate": 6.056842105263159e-06, | |
| "loss": 0.1636, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 5.376650810241699, | |
| "learning_rate": 6.0305263157894745e-06, | |
| "loss": 0.1404, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "grad_norm": 2.906466484069824, | |
| "learning_rate": 6.00421052631579e-06, | |
| "loss": 0.1105, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 3.4942939281463623, | |
| "learning_rate": 5.977894736842106e-06, | |
| "loss": 0.126, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "grad_norm": 4.702235698699951, | |
| "learning_rate": 5.951578947368422e-06, | |
| "loss": 0.1179, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 3.582550048828125, | |
| "learning_rate": 5.925263157894737e-06, | |
| "loss": 0.1384, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 1.9784924983978271, | |
| "learning_rate": 5.898947368421053e-06, | |
| "loss": 0.1422, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "grad_norm": 3.1378862857818604, | |
| "learning_rate": 5.8726315789473685e-06, | |
| "loss": 0.1157, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 4.853507995605469, | |
| "learning_rate": 5.846315789473684e-06, | |
| "loss": 0.1398, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "grad_norm": 2.9173786640167236, | |
| "learning_rate": 5.82e-06, | |
| "loss": 0.1556, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 6.456027984619141, | |
| "learning_rate": 5.793684210526316e-06, | |
| "loss": 0.1696, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "grad_norm": 3.9393770694732666, | |
| "learning_rate": 5.767368421052632e-06, | |
| "loss": 0.1811, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 4.720993518829346, | |
| "learning_rate": 5.7410526315789485e-06, | |
| "loss": 0.114, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 5.864256858825684, | |
| "learning_rate": 5.714736842105264e-06, | |
| "loss": 0.1377, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "grad_norm": 4.679260730743408, | |
| "learning_rate": 5.68842105263158e-06, | |
| "loss": 0.1292, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 2.880730628967285, | |
| "learning_rate": 5.662105263157895e-06, | |
| "loss": 0.1195, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "grad_norm": 4.698614120483398, | |
| "learning_rate": 5.635789473684211e-06, | |
| "loss": 0.1504, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "grad_norm": 5.099815368652344, | |
| "learning_rate": 5.609473684210527e-06, | |
| "loss": 0.1413, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "grad_norm": 5.136727333068848, | |
| "learning_rate": 5.5831578947368425e-06, | |
| "loss": 0.1597, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "grad_norm": 6.930360794067383, | |
| "learning_rate": 5.556842105263158e-06, | |
| "loss": 0.1118, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 5.311463832855225, | |
| "learning_rate": 5.530526315789474e-06, | |
| "loss": 0.1273, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "grad_norm": 4.134125232696533, | |
| "learning_rate": 5.50421052631579e-06, | |
| "loss": 0.1358, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "grad_norm": 3.4871582984924316, | |
| "learning_rate": 5.477894736842105e-06, | |
| "loss": 0.1436, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "grad_norm": 3.543318748474121, | |
| "learning_rate": 5.451578947368421e-06, | |
| "loss": 0.1198, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "grad_norm": 5.746144771575928, | |
| "learning_rate": 5.425263157894737e-06, | |
| "loss": 0.1086, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "grad_norm": 5.34747838973999, | |
| "learning_rate": 5.398947368421053e-06, | |
| "loss": 0.1709, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 4.202004909515381, | |
| "learning_rate": 5.372631578947369e-06, | |
| "loss": 0.1442, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 5.171213150024414, | |
| "learning_rate": 5.346315789473685e-06, | |
| "loss": 0.1459, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "grad_norm": 3.257408857345581, | |
| "learning_rate": 5.320000000000001e-06, | |
| "loss": 0.1311, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 6.485959529876709, | |
| "learning_rate": 5.2936842105263166e-06, | |
| "loss": 0.1275, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "grad_norm": 4.531818866729736, | |
| "learning_rate": 5.267368421052632e-06, | |
| "loss": 0.137, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "eval_loss": 0.32671791315078735, | |
| "eval_runtime": 1382.3945, | |
| "eval_samples_per_second": 2.948, | |
| "eval_steps_per_second": 0.369, | |
| "eval_wer": 19.96431996535118, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "grad_norm": 5.3848876953125, | |
| "learning_rate": 5.241052631578948e-06, | |
| "loss": 0.1089, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "grad_norm": 9.49464225769043, | |
| "learning_rate": 5.214736842105263e-06, | |
| "loss": 0.1431, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 5.698088645935059, | |
| "learning_rate": 5.188421052631579e-06, | |
| "loss": 0.1276, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "grad_norm": 4.742617607116699, | |
| "learning_rate": 5.162105263157895e-06, | |
| "loss": 0.1519, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 4.096518039703369, | |
| "learning_rate": 5.135789473684211e-06, | |
| "loss": 0.1073, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 4.13607931137085, | |
| "learning_rate": 5.1094736842105264e-06, | |
| "loss": 0.1501, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "grad_norm": 4.111109733581543, | |
| "learning_rate": 5.083157894736842e-06, | |
| "loss": 0.126, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 5.25567102432251, | |
| "learning_rate": 5.056842105263158e-06, | |
| "loss": 0.1153, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "grad_norm": 4.265854358673096, | |
| "learning_rate": 5.030526315789475e-06, | |
| "loss": 0.1177, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "grad_norm": 7.380852699279785, | |
| "learning_rate": 5.0042105263157906e-06, | |
| "loss": 0.1454, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "grad_norm": 4.927216529846191, | |
| "learning_rate": 4.9778947368421055e-06, | |
| "loss": 0.1184, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 6.93438720703125, | |
| "learning_rate": 4.951578947368421e-06, | |
| "loss": 0.1308, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "grad_norm": 5.152746677398682, | |
| "learning_rate": 4.925263157894737e-06, | |
| "loss": 0.1191, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "grad_norm": 5.116291522979736, | |
| "learning_rate": 4.898947368421053e-06, | |
| "loss": 0.1239, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "grad_norm": 4.750244140625, | |
| "learning_rate": 4.872631578947369e-06, | |
| "loss": 0.1644, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "grad_norm": 4.14854097366333, | |
| "learning_rate": 4.846315789473685e-06, | |
| "loss": 0.0952, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 5.03564977645874, | |
| "learning_rate": 4.8200000000000004e-06, | |
| "loss": 0.1179, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "grad_norm": 4.313379764556885, | |
| "learning_rate": 4.793684210526316e-06, | |
| "loss": 0.117, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 3.6667168140411377, | |
| "learning_rate": 4.767368421052632e-06, | |
| "loss": 0.1144, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "grad_norm": 5.04818058013916, | |
| "learning_rate": 4.741052631578948e-06, | |
| "loss": 0.1176, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "grad_norm": 4.215138912200928, | |
| "learning_rate": 4.714736842105264e-06, | |
| "loss": 0.1246, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 3.1632046699523926, | |
| "learning_rate": 4.6884210526315795e-06, | |
| "loss": 0.108, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "grad_norm": 6.4082560539245605, | |
| "learning_rate": 4.662105263157895e-06, | |
| "loss": 0.1121, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 7.097683429718018, | |
| "learning_rate": 4.63578947368421e-06, | |
| "loss": 0.1137, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "grad_norm": 6.186515808105469, | |
| "learning_rate": 4.609473684210526e-06, | |
| "loss": 0.1277, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 4.960139274597168, | |
| "learning_rate": 4.583157894736843e-06, | |
| "loss": 0.081, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 5.584286212921143, | |
| "learning_rate": 4.556842105263159e-06, | |
| "loss": 0.1108, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "grad_norm": 6.04296875, | |
| "learning_rate": 4.5305263157894744e-06, | |
| "loss": 0.1261, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 4.827033996582031, | |
| "learning_rate": 4.504210526315789e-06, | |
| "loss": 0.0954, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "grad_norm": 5.5975422859191895, | |
| "learning_rate": 4.477894736842105e-06, | |
| "loss": 0.1475, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 4.030209541320801, | |
| "learning_rate": 4.451578947368421e-06, | |
| "loss": 0.1244, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "grad_norm": 3.0414323806762695, | |
| "learning_rate": 4.425263157894737e-06, | |
| "loss": 0.0383, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 3.9159984588623047, | |
| "learning_rate": 4.3989473684210535e-06, | |
| "loss": 0.0606, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "grad_norm": 3.7275309562683105, | |
| "learning_rate": 4.3726315789473685e-06, | |
| "loss": 0.0426, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "grad_norm": 2.282072067260742, | |
| "learning_rate": 4.346315789473684e-06, | |
| "loss": 0.051, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 3.405268669128418, | |
| "learning_rate": 4.32e-06, | |
| "loss": 0.0747, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "grad_norm": 2.372917413711548, | |
| "learning_rate": 4.293684210526316e-06, | |
| "loss": 0.042, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 2.6286582946777344, | |
| "learning_rate": 4.267368421052632e-06, | |
| "loss": 0.0391, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "grad_norm": 5.5469841957092285, | |
| "learning_rate": 4.241052631578948e-06, | |
| "loss": 0.0476, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 6.648013591766357, | |
| "learning_rate": 4.214736842105263e-06, | |
| "loss": 0.0544, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_loss": 0.35065552592277527, | |
| "eval_runtime": 1299.6934, | |
| "eval_samples_per_second": 3.135, | |
| "eval_steps_per_second": 0.392, | |
| "eval_wer": 18.08750889422829, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "grad_norm": 4.275888442993164, | |
| "learning_rate": 4.188421052631579e-06, | |
| "loss": 0.0497, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 3.1425840854644775, | |
| "learning_rate": 4.162105263157895e-06, | |
| "loss": 0.0511, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 7.669829845428467, | |
| "learning_rate": 4.135789473684211e-06, | |
| "loss": 0.05, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "grad_norm": 3.776144504547119, | |
| "learning_rate": 4.109473684210527e-06, | |
| "loss": 0.0378, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 3.3173623085021973, | |
| "learning_rate": 4.0831578947368425e-06, | |
| "loss": 0.0466, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "grad_norm": 3.4908010959625244, | |
| "learning_rate": 4.056842105263158e-06, | |
| "loss": 0.0288, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 1.7697815895080566, | |
| "learning_rate": 4.030526315789474e-06, | |
| "loss": 0.0328, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "grad_norm": 3.193303346633911, | |
| "learning_rate": 4.00421052631579e-06, | |
| "loss": 0.0563, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 3.5942752361297607, | |
| "learning_rate": 3.977894736842106e-06, | |
| "loss": 0.0475, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 4.8846917152404785, | |
| "learning_rate": 3.951578947368422e-06, | |
| "loss": 0.0511, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "grad_norm": 3.5842273235321045, | |
| "learning_rate": 3.9252631578947366e-06, | |
| "loss": 0.0535, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 4.6863250732421875, | |
| "learning_rate": 3.898947368421052e-06, | |
| "loss": 0.0393, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "grad_norm": 2.1344845294952393, | |
| "learning_rate": 3.872631578947369e-06, | |
| "loss": 0.0442, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 2.1561779975891113, | |
| "learning_rate": 3.846315789473685e-06, | |
| "loss": 0.0433, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "grad_norm": 3.0101683139801025, | |
| "learning_rate": 3.820000000000001e-06, | |
| "loss": 0.0514, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 2.804795026779175, | |
| "learning_rate": 3.793684210526316e-06, | |
| "loss": 0.0546, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "grad_norm": 1.121903896331787, | |
| "learning_rate": 3.767368421052632e-06, | |
| "loss": 0.0501, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "grad_norm": 3.670987606048584, | |
| "learning_rate": 3.7410526315789473e-06, | |
| "loss": 0.0581, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 3.9386682510375977, | |
| "learning_rate": 3.714736842105263e-06, | |
| "loss": 0.0399, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "grad_norm": 3.8390467166900635, | |
| "learning_rate": 3.6884210526315794e-06, | |
| "loss": 0.0466, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 0.33132675290107727, | |
| "learning_rate": 3.662105263157895e-06, | |
| "loss": 0.0536, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "grad_norm": 4.512747287750244, | |
| "learning_rate": 3.635789473684211e-06, | |
| "loss": 0.0386, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "grad_norm": 2.1268844604492188, | |
| "learning_rate": 3.6094736842105264e-06, | |
| "loss": 0.0545, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 4.197777271270752, | |
| "learning_rate": 3.5831578947368422e-06, | |
| "loss": 0.0502, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "grad_norm": 3.005239725112915, | |
| "learning_rate": 3.556842105263158e-06, | |
| "loss": 0.04, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "grad_norm": 2.6406736373901367, | |
| "learning_rate": 3.5305263157894743e-06, | |
| "loss": 0.0508, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "grad_norm": 3.452228307723999, | |
| "learning_rate": 3.50421052631579e-06, | |
| "loss": 0.0409, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 8.10651683807373, | |
| "learning_rate": 3.4778947368421055e-06, | |
| "loss": 0.055, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "grad_norm": 3.4862449169158936, | |
| "learning_rate": 3.4515789473684213e-06, | |
| "loss": 0.0482, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 1.1421183347702026, | |
| "learning_rate": 3.425263157894737e-06, | |
| "loss": 0.0413, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "grad_norm": 2.27363657951355, | |
| "learning_rate": 3.398947368421053e-06, | |
| "loss": 0.0468, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 6.514892101287842, | |
| "learning_rate": 3.3726315789473683e-06, | |
| "loss": 0.0562, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 4.252927780151367, | |
| "learning_rate": 3.3463157894736846e-06, | |
| "loss": 0.0392, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "grad_norm": 3.4321846961975098, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 0.0567, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "grad_norm": 2.55214786529541, | |
| "learning_rate": 3.2936842105263162e-06, | |
| "loss": 0.0466, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "grad_norm": 4.45693302154541, | |
| "learning_rate": 3.267368421052632e-06, | |
| "loss": 0.0432, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 5.478282928466797, | |
| "learning_rate": 3.2410526315789474e-06, | |
| "loss": 0.0527, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "grad_norm": 6.43654727935791, | |
| "learning_rate": 3.2147368421052633e-06, | |
| "loss": 0.0717, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 2.998990058898926, | |
| "learning_rate": 3.188421052631579e-06, | |
| "loss": 0.0433, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 2.361015796661377, | |
| "learning_rate": 3.1621052631578953e-06, | |
| "loss": 0.0379, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_loss": 0.3512289226055145, | |
| "eval_runtime": 1291.9018, | |
| "eval_samples_per_second": 3.154, | |
| "eval_steps_per_second": 0.395, | |
| "eval_wer": 17.68430387839914, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "grad_norm": 1.4288549423217773, | |
| "learning_rate": 3.135789473684211e-06, | |
| "loss": 0.0448, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 5.668064117431641, | |
| "learning_rate": 3.1094736842105265e-06, | |
| "loss": 0.0372, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "grad_norm": 2.740320920944214, | |
| "learning_rate": 3.0831578947368423e-06, | |
| "loss": 0.0477, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "grad_norm": 2.9186103343963623, | |
| "learning_rate": 3.056842105263158e-06, | |
| "loss": 0.037, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "grad_norm": 2.684190511703491, | |
| "learning_rate": 3.0305263157894736e-06, | |
| "loss": 0.0418, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 2.112553358078003, | |
| "learning_rate": 3.0042105263157894e-06, | |
| "loss": 0.0424, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "grad_norm": 2.508024215698242, | |
| "learning_rate": 2.9778947368421056e-06, | |
| "loss": 0.0496, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "grad_norm": 2.080995559692383, | |
| "learning_rate": 2.9515789473684214e-06, | |
| "loss": 0.0506, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 3.4865529537200928, | |
| "learning_rate": 2.9252631578947373e-06, | |
| "loss": 0.0542, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "grad_norm": 0.8756410479545593, | |
| "learning_rate": 2.8989473684210526e-06, | |
| "loss": 0.0503, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "grad_norm": 1.8365063667297363, | |
| "learning_rate": 2.8726315789473685e-06, | |
| "loss": 0.0357, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "grad_norm": 4.022027969360352, | |
| "learning_rate": 2.8463157894736843e-06, | |
| "loss": 0.0319, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "grad_norm": 5.825380325317383, | |
| "learning_rate": 2.82e-06, | |
| "loss": 0.0589, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 3.63252592086792, | |
| "learning_rate": 2.7936842105263163e-06, | |
| "loss": 0.0447, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "grad_norm": 3.397965669631958, | |
| "learning_rate": 2.7673684210526317e-06, | |
| "loss": 0.0465, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 1.4917527437210083, | |
| "learning_rate": 2.7410526315789476e-06, | |
| "loss": 0.0407, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "grad_norm": 1.1915754079818726, | |
| "learning_rate": 2.7147368421052634e-06, | |
| "loss": 0.039, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 1.5035796165466309, | |
| "learning_rate": 2.688421052631579e-06, | |
| "loss": 0.0413, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "grad_norm": 2.2426724433898926, | |
| "learning_rate": 2.6621052631578946e-06, | |
| "loss": 0.0498, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 3.2812421321868896, | |
| "learning_rate": 2.635789473684211e-06, | |
| "loss": 0.0426, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "grad_norm": 2.2315256595611572, | |
| "learning_rate": 2.6094736842105267e-06, | |
| "loss": 0.0455, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 2.0561962127685547, | |
| "learning_rate": 2.5831578947368425e-06, | |
| "loss": 0.047, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 0.6745893359184265, | |
| "learning_rate": 2.5568421052631583e-06, | |
| "loss": 0.0369, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "grad_norm": 5.30702543258667, | |
| "learning_rate": 2.5305263157894737e-06, | |
| "loss": 0.0483, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 3.2221291065216064, | |
| "learning_rate": 2.5042105263157895e-06, | |
| "loss": 0.0447, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "grad_norm": 2.8724985122680664, | |
| "learning_rate": 2.4778947368421053e-06, | |
| "loss": 0.0425, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 8.444465637207031, | |
| "learning_rate": 2.451578947368421e-06, | |
| "loss": 0.0375, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "grad_norm": 2.455996036529541, | |
| "learning_rate": 2.425263157894737e-06, | |
| "loss": 0.042, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 2.7106616497039795, | |
| "learning_rate": 2.3989473684210528e-06, | |
| "loss": 0.0479, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 4.59207820892334, | |
| "learning_rate": 2.3726315789473686e-06, | |
| "loss": 0.0618, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "grad_norm": 4.093899726867676, | |
| "learning_rate": 2.3463157894736844e-06, | |
| "loss": 0.0429, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "grad_norm": 5.356892108917236, | |
| "learning_rate": 2.3200000000000002e-06, | |
| "loss": 0.0443, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "grad_norm": 5.630926609039307, | |
| "learning_rate": 2.293684210526316e-06, | |
| "loss": 0.0461, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "grad_norm": 1.5302308797836304, | |
| "learning_rate": 2.267368421052632e-06, | |
| "loss": 0.0619, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "grad_norm": 4.4012131690979, | |
| "learning_rate": 2.2410526315789473e-06, | |
| "loss": 0.0469, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 2.787708282470703, | |
| "learning_rate": 2.2147368421052635e-06, | |
| "loss": 0.048, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 4.563180923461914, | |
| "learning_rate": 2.188421052631579e-06, | |
| "loss": 0.0542, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 3.1255428791046143, | |
| "learning_rate": 2.1621052631578947e-06, | |
| "loss": 0.0409, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 3.2584402561187744, | |
| "learning_rate": 2.135789473684211e-06, | |
| "loss": 0.0412, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "grad_norm": 3.4598073959350586, | |
| "learning_rate": 2.1094736842105264e-06, | |
| "loss": 0.0381, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "eval_loss": 0.35260751843452454, | |
| "eval_runtime": 1280.7304, | |
| "eval_samples_per_second": 3.182, | |
| "eval_steps_per_second": 0.398, | |
| "eval_wer": 17.009889350643995, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "grad_norm": Infinity, | |
| "learning_rate": 2.0842105263157897e-06, | |
| "loss": 0.0415, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "grad_norm": 1.2554646730422974, | |
| "learning_rate": 2.0578947368421055e-06, | |
| "loss": 0.0487, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 3.3572278022766113, | |
| "learning_rate": 2.031578947368421e-06, | |
| "loss": 0.0314, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "grad_norm": 2.819499969482422, | |
| "learning_rate": 2.005263157894737e-06, | |
| "loss": 0.028, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "grad_norm": 3.161909341812134, | |
| "learning_rate": 1.978947368421053e-06, | |
| "loss": 0.0371, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 3.9492976665496826, | |
| "learning_rate": 1.9526315789473684e-06, | |
| "loss": 0.0446, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "grad_norm": 3.000467538833618, | |
| "learning_rate": 1.9263157894736846e-06, | |
| "loss": 0.045, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 7.680356502532959, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.0388, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "grad_norm": 2.7837448120117188, | |
| "learning_rate": 1.8736842105263158e-06, | |
| "loss": 0.0458, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 5.497086524963379, | |
| "learning_rate": 1.8473684210526319e-06, | |
| "loss": 0.0292, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "grad_norm": 3.6240792274475098, | |
| "learning_rate": 1.8210526315789475e-06, | |
| "loss": 0.0463, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "grad_norm": 4.298013210296631, | |
| "learning_rate": 1.7947368421052633e-06, | |
| "loss": 0.0571, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 1.796762228012085, | |
| "learning_rate": 1.768421052631579e-06, | |
| "loss": 0.0356, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "grad_norm": 3.264356851577759, | |
| "learning_rate": 1.742105263157895e-06, | |
| "loss": 0.0413, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 3.744009256362915, | |
| "learning_rate": 1.7157894736842107e-06, | |
| "loss": 0.0439, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "grad_norm": 1.785539984703064, | |
| "learning_rate": 1.6894736842105264e-06, | |
| "loss": 0.0284, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 2.177640199661255, | |
| "learning_rate": 1.6631578947368424e-06, | |
| "loss": 0.0378, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "grad_norm": 3.3337082862854004, | |
| "learning_rate": 1.636842105263158e-06, | |
| "loss": 0.046, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 4.369349479675293, | |
| "learning_rate": 1.6105263157894738e-06, | |
| "loss": 0.0345, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 4.415827751159668, | |
| "learning_rate": 1.5842105263157894e-06, | |
| "loss": 0.0376, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "grad_norm": 3.198944091796875, | |
| "learning_rate": 1.5578947368421054e-06, | |
| "loss": 0.04, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 2.9279515743255615, | |
| "learning_rate": 1.5315789473684213e-06, | |
| "loss": 0.0413, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "grad_norm": 3.4797816276550293, | |
| "learning_rate": 1.5052631578947369e-06, | |
| "loss": 0.037, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 2.2802135944366455, | |
| "learning_rate": 1.478947368421053e-06, | |
| "loss": 0.0436, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "grad_norm": 2.8511545658111572, | |
| "learning_rate": 1.4526315789473685e-06, | |
| "loss": 0.0459, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 2.9255013465881348, | |
| "learning_rate": 1.4263157894736843e-06, | |
| "loss": 0.033, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.6602519750595093, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.0326, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "grad_norm": 1.8406025171279907, | |
| "learning_rate": 1.373684210526316e-06, | |
| "loss": 0.0241, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "grad_norm": 0.500821590423584, | |
| "learning_rate": 1.3473684210526316e-06, | |
| "loss": 0.0088, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "grad_norm": 1.8509939908981323, | |
| "learning_rate": 1.3210526315789474e-06, | |
| "loss": 0.0146, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "grad_norm": 0.14848609268665314, | |
| "learning_rate": 1.2947368421052634e-06, | |
| "loss": 0.0111, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "grad_norm": 1.1834708452224731, | |
| "learning_rate": 1.268421052631579e-06, | |
| "loss": 0.0113, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "grad_norm": 1.8275666236877441, | |
| "learning_rate": 1.2421052631578948e-06, | |
| "loss": 0.011, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 2.32273006439209, | |
| "learning_rate": 1.2157894736842107e-06, | |
| "loss": 0.0127, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "grad_norm": 1.478613257408142, | |
| "learning_rate": 1.1894736842105265e-06, | |
| "loss": 0.0088, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "grad_norm": 0.323777973651886, | |
| "learning_rate": 1.163157894736842e-06, | |
| "loss": 0.016, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "grad_norm": 1.2564371824264526, | |
| "learning_rate": 1.136842105263158e-06, | |
| "loss": 0.011, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "grad_norm": 2.7100419998168945, | |
| "learning_rate": 1.110526315789474e-06, | |
| "loss": 0.011, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "grad_norm": 3.0302305221557617, | |
| "learning_rate": 1.0842105263157895e-06, | |
| "loss": 0.0118, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "grad_norm": 1.5977123975753784, | |
| "learning_rate": 1.0578947368421054e-06, | |
| "loss": 0.0128, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "eval_loss": 0.36985671520233154, | |
| "eval_runtime": 1284.7105, | |
| "eval_samples_per_second": 3.172, | |
| "eval_steps_per_second": 0.397, | |
| "eval_wer": 16.64483928516185, | |
| "step": 9000 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 1000, | |
| "total_flos": 7.346518245310464e+19, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |