| { | |
| "best_metric": 93.07875894988067, | |
| "best_model_checkpoint": "ASFEMA_SEV/checkpoint-6000", | |
| "epoch": 77.91318327974277, | |
| "global_step": 6000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 4.0894, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 3.7, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.46e-06, | |
| "loss": 3.0759, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 2.5729, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.46e-06, | |
| "loss": 2.1885, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.96e-06, | |
| "loss": 1.9431, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 3.46e-06, | |
| "loss": 1.5753, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 3.96e-06, | |
| "loss": 1.3319, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 1.2076, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.9211, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.7332, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.6601, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.4824, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.3455, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.3741, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.2448, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "eval_loss": 0.9975764751434326, | |
| "eval_runtime": 24.147, | |
| "eval_samples_per_second": 5.756, | |
| "eval_steps_per_second": 0.745, | |
| "eval_wer": 132.57756563245823, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.1713, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.1906, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.1479, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.103, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 9.975789473684211e-06, | |
| "loss": 0.1271, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 9.949473684210526e-06, | |
| "loss": 0.1033, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 9.923157894736844e-06, | |
| "loss": 0.0799, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 9.89684210526316e-06, | |
| "loss": 0.0814, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 9.870526315789474e-06, | |
| "loss": 0.069, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 9.84421052631579e-06, | |
| "loss": 0.0546, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 9.817894736842106e-06, | |
| "loss": 0.0627, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 9.791578947368422e-06, | |
| "loss": 0.0459, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 9.765263157894737e-06, | |
| "loss": 0.034, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 9.738947368421054e-06, | |
| "loss": 0.036, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 9.712631578947369e-06, | |
| "loss": 0.0367, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 9.686315789473684e-06, | |
| "loss": 0.0248, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "eval_loss": 0.8046127557754517, | |
| "eval_runtime": 19.9036, | |
| "eval_samples_per_second": 6.984, | |
| "eval_steps_per_second": 0.904, | |
| "eval_wer": 108.59188544152745, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.0324, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 9.633684210526316e-06, | |
| "loss": 0.0374, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 11.36, | |
| "learning_rate": 9.607368421052632e-06, | |
| "loss": 0.0253, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 11.68, | |
| "learning_rate": 9.581052631578947e-06, | |
| "loss": 0.0262, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 12.01, | |
| "learning_rate": 9.554736842105264e-06, | |
| "loss": 0.0374, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 12.33, | |
| "learning_rate": 9.52842105263158e-06, | |
| "loss": 0.0179, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 12.66, | |
| "learning_rate": 9.502105263157896e-06, | |
| "loss": 0.0236, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 12.98, | |
| "learning_rate": 9.475789473684212e-06, | |
| "loss": 0.0187, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 13.31, | |
| "learning_rate": 9.449473684210527e-06, | |
| "loss": 0.0186, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 9.423157894736842e-06, | |
| "loss": 0.0217, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 13.95, | |
| "learning_rate": 9.396842105263159e-06, | |
| "loss": 0.0206, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 14.28, | |
| "learning_rate": 9.370526315789474e-06, | |
| "loss": 0.0103, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 14.6, | |
| "learning_rate": 9.34421052631579e-06, | |
| "loss": 0.0168, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 14.93, | |
| "learning_rate": 9.317894736842105e-06, | |
| "loss": 0.0214, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 9.291578947368422e-06, | |
| "loss": 0.0115, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 15.58, | |
| "learning_rate": 9.265263157894737e-06, | |
| "loss": 0.0138, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 15.58, | |
| "eval_loss": 0.8237143158912659, | |
| "eval_runtime": 22.295, | |
| "eval_samples_per_second": 6.235, | |
| "eval_steps_per_second": 0.807, | |
| "eval_wer": 117.89976133651552, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 15.9, | |
| "learning_rate": 9.238947368421052e-06, | |
| "loss": 0.0216, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 16.23, | |
| "learning_rate": 9.21263157894737e-06, | |
| "loss": 0.014, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 16.55, | |
| "learning_rate": 9.186315789473685e-06, | |
| "loss": 0.0117, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 16.87, | |
| "learning_rate": 9.16e-06, | |
| "loss": 0.0104, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 17.21, | |
| "learning_rate": 9.133684210526317e-06, | |
| "loss": 0.0086, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 17.53, | |
| "learning_rate": 9.107368421052632e-06, | |
| "loss": 0.0092, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 9.081052631578949e-06, | |
| "loss": 0.0098, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 18.18, | |
| "learning_rate": 9.054736842105264e-06, | |
| "loss": 0.0125, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 18.5, | |
| "learning_rate": 9.02842105263158e-06, | |
| "loss": 0.0055, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 18.82, | |
| "learning_rate": 9.002105263157895e-06, | |
| "loss": 0.011, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 19.15, | |
| "learning_rate": 8.97578947368421e-06, | |
| "loss": 0.0088, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 19.48, | |
| "learning_rate": 8.949473684210527e-06, | |
| "loss": 0.0065, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 19.8, | |
| "learning_rate": 8.923157894736842e-06, | |
| "loss": 0.0098, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 20.13, | |
| "learning_rate": 8.896842105263159e-06, | |
| "loss": 0.0038, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 20.45, | |
| "learning_rate": 8.870526315789474e-06, | |
| "loss": 0.0053, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 20.77, | |
| "learning_rate": 8.84421052631579e-06, | |
| "loss": 0.0075, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 20.77, | |
| "eval_loss": 0.8213781118392944, | |
| "eval_runtime": 18.9322, | |
| "eval_samples_per_second": 7.342, | |
| "eval_steps_per_second": 0.951, | |
| "eval_wer": 103.5799522673031, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 21.1, | |
| "learning_rate": 8.817894736842107e-06, | |
| "loss": 0.0111, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 21.42, | |
| "learning_rate": 8.791578947368422e-06, | |
| "loss": 0.0083, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 21.75, | |
| "learning_rate": 8.765263157894739e-06, | |
| "loss": 0.0056, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 22.08, | |
| "learning_rate": 8.738947368421053e-06, | |
| "loss": 0.0055, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 22.4, | |
| "learning_rate": 8.712631578947368e-06, | |
| "loss": 0.0031, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 22.72, | |
| "learning_rate": 8.686315789473685e-06, | |
| "loss": 0.0075, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 23.05, | |
| "learning_rate": 8.66e-06, | |
| "loss": 0.0078, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 23.37, | |
| "learning_rate": 8.633684210526317e-06, | |
| "loss": 0.0063, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 23.69, | |
| "learning_rate": 8.607368421052632e-06, | |
| "loss": 0.0012, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 24.03, | |
| "learning_rate": 8.581052631578948e-06, | |
| "loss": 0.0045, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 24.35, | |
| "learning_rate": 8.554736842105263e-06, | |
| "loss": 0.004, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 24.67, | |
| "learning_rate": 8.528421052631578e-06, | |
| "loss": 0.006, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 24.99, | |
| "learning_rate": 8.502105263157897e-06, | |
| "loss": 0.0034, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 25.32, | |
| "learning_rate": 8.475789473684212e-06, | |
| "loss": 0.0024, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 25.64, | |
| "learning_rate": 8.449473684210527e-06, | |
| "loss": 0.0032, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 25.96, | |
| "learning_rate": 8.423157894736843e-06, | |
| "loss": 0.0021, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 25.96, | |
| "eval_loss": 0.7943580746650696, | |
| "eval_runtime": 20.6652, | |
| "eval_samples_per_second": 6.726, | |
| "eval_steps_per_second": 0.871, | |
| "eval_wer": 106.92124105011933, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 26.3, | |
| "learning_rate": 8.396842105263158e-06, | |
| "loss": 0.0006, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 26.62, | |
| "learning_rate": 8.370526315789475e-06, | |
| "loss": 0.0018, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 26.94, | |
| "learning_rate": 8.34421052631579e-06, | |
| "loss": 0.0032, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 27.27, | |
| "learning_rate": 8.317894736842107e-06, | |
| "loss": 0.0034, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 27.59, | |
| "learning_rate": 8.291578947368422e-06, | |
| "loss": 0.0021, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 27.91, | |
| "learning_rate": 8.265263157894737e-06, | |
| "loss": 0.005, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 28.24, | |
| "learning_rate": 8.238947368421053e-06, | |
| "loss": 0.002, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 28.57, | |
| "learning_rate": 8.212631578947368e-06, | |
| "loss": 0.0023, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 28.89, | |
| "learning_rate": 8.186315789473685e-06, | |
| "loss": 0.0126, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 29.22, | |
| "learning_rate": 8.16e-06, | |
| "loss": 0.0064, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 29.54, | |
| "learning_rate": 8.133684210526316e-06, | |
| "loss": 0.004, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 29.86, | |
| "learning_rate": 8.107368421052633e-06, | |
| "loss": 0.003, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 30.19, | |
| "learning_rate": 8.081052631578948e-06, | |
| "loss": 0.0064, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 30.51, | |
| "learning_rate": 8.054736842105265e-06, | |
| "loss": 0.0038, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 30.84, | |
| "learning_rate": 8.02842105263158e-06, | |
| "loss": 0.006, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 31.17, | |
| "learning_rate": 8.002105263157895e-06, | |
| "loss": 0.003, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 31.17, | |
| "eval_loss": 0.811363160610199, | |
| "eval_runtime": 17.6674, | |
| "eval_samples_per_second": 7.868, | |
| "eval_steps_per_second": 1.019, | |
| "eval_wer": 98.68735083532219, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 31.49, | |
| "learning_rate": 7.975789473684211e-06, | |
| "loss": 0.0021, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 31.81, | |
| "learning_rate": 7.949473684210526e-06, | |
| "loss": 0.0045, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 32.14, | |
| "learning_rate": 7.923157894736843e-06, | |
| "loss": 0.0028, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 32.46, | |
| "learning_rate": 7.896842105263158e-06, | |
| "loss": 0.0008, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 32.78, | |
| "learning_rate": 7.870526315789475e-06, | |
| "loss": 0.0006, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 33.12, | |
| "learning_rate": 7.84421052631579e-06, | |
| "loss": 0.0023, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 33.44, | |
| "learning_rate": 7.817894736842105e-06, | |
| "loss": 0.0023, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 33.76, | |
| "learning_rate": 7.791578947368423e-06, | |
| "loss": 0.0068, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 34.09, | |
| "learning_rate": 7.765263157894738e-06, | |
| "loss": 0.0009, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 34.41, | |
| "learning_rate": 7.738947368421053e-06, | |
| "loss": 0.0011, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 34.73, | |
| "learning_rate": 7.71263157894737e-06, | |
| "loss": 0.0027, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 35.06, | |
| "learning_rate": 7.686315789473685e-06, | |
| "loss": 0.0008, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 35.39, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 0.0021, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 35.71, | |
| "learning_rate": 7.633684210526316e-06, | |
| "loss": 0.0028, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 36.04, | |
| "learning_rate": 7.607368421052632e-06, | |
| "loss": 0.0018, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 36.36, | |
| "learning_rate": 7.581052631578948e-06, | |
| "loss": 0.0032, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 36.36, | |
| "eval_loss": 0.7548059821128845, | |
| "eval_runtime": 19.1115, | |
| "eval_samples_per_second": 7.273, | |
| "eval_steps_per_second": 0.942, | |
| "eval_wer": 105.48926014319808, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 36.68, | |
| "learning_rate": 7.554736842105264e-06, | |
| "loss": 0.003, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 37.01, | |
| "learning_rate": 7.5284210526315794e-06, | |
| "loss": 0.0005, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 37.33, | |
| "learning_rate": 7.502105263157895e-06, | |
| "loss": 0.0038, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 37.66, | |
| "learning_rate": 7.475789473684211e-06, | |
| "loss": 0.0021, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 37.98, | |
| "learning_rate": 7.449473684210526e-06, | |
| "loss": 0.0011, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 38.31, | |
| "learning_rate": 7.4231578947368436e-06, | |
| "loss": 0.0023, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 38.63, | |
| "learning_rate": 7.3968421052631585e-06, | |
| "loss": 0.003, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 38.95, | |
| "learning_rate": 7.370526315789474e-06, | |
| "loss": 0.0046, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 39.28, | |
| "learning_rate": 7.34421052631579e-06, | |
| "loss": 0.001, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 39.6, | |
| "learning_rate": 7.317894736842106e-06, | |
| "loss": 0.0003, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 39.93, | |
| "learning_rate": 7.291578947368422e-06, | |
| "loss": 0.003, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 40.26, | |
| "learning_rate": 7.265263157894738e-06, | |
| "loss": 0.0013, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 40.58, | |
| "learning_rate": 7.2389473684210534e-06, | |
| "loss": 0.0016, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 40.9, | |
| "learning_rate": 7.212631578947369e-06, | |
| "loss": 0.0009, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 41.23, | |
| "learning_rate": 7.186315789473684e-06, | |
| "loss": 0.0004, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 41.55, | |
| "learning_rate": 7.16e-06, | |
| "loss": 0.0002, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 41.55, | |
| "eval_loss": 0.7998279929161072, | |
| "eval_runtime": 17.346, | |
| "eval_samples_per_second": 8.013, | |
| "eval_steps_per_second": 1.038, | |
| "eval_wer": 99.88066825775657, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 41.87, | |
| "learning_rate": 7.133684210526316e-06, | |
| "loss": 0.0008, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 42.21, | |
| "learning_rate": 7.107368421052632e-06, | |
| "loss": 0.0014, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 42.53, | |
| "learning_rate": 7.0810526315789475e-06, | |
| "loss": 0.0016, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 42.85, | |
| "learning_rate": 7.054736842105264e-06, | |
| "loss": 0.0015, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 43.18, | |
| "learning_rate": 7.02842105263158e-06, | |
| "loss": 0.0032, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 43.5, | |
| "learning_rate": 7.002105263157896e-06, | |
| "loss": 0.0027, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 43.82, | |
| "learning_rate": 6.975789473684212e-06, | |
| "loss": 0.0027, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 44.15, | |
| "learning_rate": 6.9494736842105275e-06, | |
| "loss": 0.0052, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 44.48, | |
| "learning_rate": 6.9231578947368424e-06, | |
| "loss": 0.0081, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 44.8, | |
| "learning_rate": 6.896842105263158e-06, | |
| "loss": 0.0076, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 45.13, | |
| "learning_rate": 6.870526315789474e-06, | |
| "loss": 0.0074, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 45.45, | |
| "learning_rate": 6.84421052631579e-06, | |
| "loss": 0.003, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 45.77, | |
| "learning_rate": 6.817894736842106e-06, | |
| "loss": 0.0058, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 46.1, | |
| "learning_rate": 6.7915789473684215e-06, | |
| "loss": 0.007, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 46.42, | |
| "learning_rate": 6.765263157894737e-06, | |
| "loss": 0.0098, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 46.75, | |
| "learning_rate": 6.738947368421052e-06, | |
| "loss": 0.0077, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 46.75, | |
| "eval_loss": 0.8303264379501343, | |
| "eval_runtime": 22.148, | |
| "eval_samples_per_second": 6.276, | |
| "eval_steps_per_second": 0.813, | |
| "eval_wer": 123.74701670644393, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 47.08, | |
| "learning_rate": 6.71263157894737e-06, | |
| "loss": 0.0072, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 47.4, | |
| "learning_rate": 6.686315789473685e-06, | |
| "loss": 0.0067, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 47.72, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 0.0031, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 48.05, | |
| "learning_rate": 6.6336842105263164e-06, | |
| "loss": 0.0041, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 48.37, | |
| "learning_rate": 6.607368421052632e-06, | |
| "loss": 0.0018, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 48.69, | |
| "learning_rate": 6.581052631578948e-06, | |
| "loss": 0.0032, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 49.03, | |
| "learning_rate": 6.554736842105264e-06, | |
| "loss": 0.0046, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 49.35, | |
| "learning_rate": 6.52842105263158e-06, | |
| "loss": 0.0025, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 49.67, | |
| "learning_rate": 6.5021052631578955e-06, | |
| "loss": 0.0042, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 49.99, | |
| "learning_rate": 6.4757894736842105e-06, | |
| "loss": 0.0044, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 50.32, | |
| "learning_rate": 6.449473684210526e-06, | |
| "loss": 0.003, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 50.64, | |
| "learning_rate": 6.423157894736842e-06, | |
| "loss": 0.0023, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 50.96, | |
| "learning_rate": 6.396842105263158e-06, | |
| "loss": 0.0019, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 51.3, | |
| "learning_rate": 6.370526315789474e-06, | |
| "loss": 0.0037, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 51.62, | |
| "learning_rate": 6.3442105263157904e-06, | |
| "loss": 0.0027, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 51.94, | |
| "learning_rate": 6.317894736842106e-06, | |
| "loss": 0.0013, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 51.94, | |
| "eval_loss": 0.7767909169197083, | |
| "eval_runtime": 21.9995, | |
| "eval_samples_per_second": 6.318, | |
| "eval_steps_per_second": 0.818, | |
| "eval_wer": 118.49642004773268, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 52.27, | |
| "learning_rate": 6.291578947368422e-06, | |
| "loss": 0.0022, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 52.59, | |
| "learning_rate": 6.265263157894738e-06, | |
| "loss": 0.0019, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 52.91, | |
| "learning_rate": 6.238947368421054e-06, | |
| "loss": 0.0006, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 53.24, | |
| "learning_rate": 6.212631578947369e-06, | |
| "loss": 0.0018, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 53.57, | |
| "learning_rate": 6.1863157894736845e-06, | |
| "loss": 0.0027, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 53.89, | |
| "learning_rate": 6.16e-06, | |
| "loss": 0.0007, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 54.22, | |
| "learning_rate": 6.133684210526316e-06, | |
| "loss": 0.0005, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 54.54, | |
| "learning_rate": 6.107368421052632e-06, | |
| "loss": 0.0005, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 54.86, | |
| "learning_rate": 6.081052631578948e-06, | |
| "loss": 0.0002, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 55.19, | |
| "learning_rate": 6.054736842105264e-06, | |
| "loss": 0.0004, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 55.51, | |
| "learning_rate": 6.0284210526315786e-06, | |
| "loss": 0.0001, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 55.84, | |
| "learning_rate": 6.002105263157896e-06, | |
| "loss": 0.0001, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 56.17, | |
| "learning_rate": 5.975789473684212e-06, | |
| "loss": 0.0001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 56.49, | |
| "learning_rate": 5.949473684210527e-06, | |
| "loss": 0.0001, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 56.81, | |
| "learning_rate": 5.923157894736843e-06, | |
| "loss": 0.0001, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 57.14, | |
| "learning_rate": 5.8968421052631585e-06, | |
| "loss": 0.0001, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 57.14, | |
| "eval_loss": 0.772087812423706, | |
| "eval_runtime": 20.4241, | |
| "eval_samples_per_second": 6.806, | |
| "eval_steps_per_second": 0.881, | |
| "eval_wer": 97.01670644391407, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 57.46, | |
| "learning_rate": 5.870526315789474e-06, | |
| "loss": 0.0001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 57.78, | |
| "learning_rate": 5.84421052631579e-06, | |
| "loss": 0.0001, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 58.12, | |
| "learning_rate": 5.817894736842106e-06, | |
| "loss": 0.0001, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 58.44, | |
| "learning_rate": 5.791578947368422e-06, | |
| "loss": 0.0001, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 58.76, | |
| "learning_rate": 5.765263157894737e-06, | |
| "loss": 0.0001, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 59.09, | |
| "learning_rate": 5.7389473684210526e-06, | |
| "loss": 0.0001, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 59.41, | |
| "learning_rate": 5.712631578947368e-06, | |
| "loss": 0.0, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 59.73, | |
| "learning_rate": 5.686315789473684e-06, | |
| "loss": 0.0001, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 60.06, | |
| "learning_rate": 5.66e-06, | |
| "loss": 0.0001, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 60.39, | |
| "learning_rate": 5.633684210526317e-06, | |
| "loss": 0.0, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 60.71, | |
| "learning_rate": 5.6073684210526325e-06, | |
| "loss": 0.0, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 61.04, | |
| "learning_rate": 5.581052631578948e-06, | |
| "loss": 0.0, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 61.36, | |
| "learning_rate": 5.554736842105264e-06, | |
| "loss": 0.0, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 61.68, | |
| "learning_rate": 5.52842105263158e-06, | |
| "loss": 0.0, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 62.01, | |
| "learning_rate": 5.502105263157895e-06, | |
| "loss": 0.0, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 62.33, | |
| "learning_rate": 5.475789473684211e-06, | |
| "loss": 0.0, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 62.33, | |
| "eval_loss": 0.7726908922195435, | |
| "eval_runtime": 20.4469, | |
| "eval_samples_per_second": 6.798, | |
| "eval_steps_per_second": 0.88, | |
| "eval_wer": 97.49403341288783, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 62.66, | |
| "learning_rate": 5.4494736842105266e-06, | |
| "loss": 0.0, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 62.98, | |
| "learning_rate": 5.423157894736842e-06, | |
| "loss": 0.0, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 63.31, | |
| "learning_rate": 5.396842105263158e-06, | |
| "loss": 0.0, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 63.63, | |
| "learning_rate": 5.370526315789474e-06, | |
| "loss": 0.0, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 63.95, | |
| "learning_rate": 5.34421052631579e-06, | |
| "loss": 0.0, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 64.28, | |
| "learning_rate": 5.317894736842105e-06, | |
| "loss": 0.0, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 64.6, | |
| "learning_rate": 5.291578947368422e-06, | |
| "loss": 0.0, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 64.93, | |
| "learning_rate": 5.265263157894738e-06, | |
| "loss": 0.0, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 65.26, | |
| "learning_rate": 5.238947368421053e-06, | |
| "loss": 0.0, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 65.58, | |
| "learning_rate": 5.212631578947369e-06, | |
| "loss": 0.0, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 65.9, | |
| "learning_rate": 5.186315789473685e-06, | |
| "loss": 0.0, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 66.23, | |
| "learning_rate": 5.1600000000000006e-06, | |
| "loss": 0.0, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 66.55, | |
| "learning_rate": 5.133684210526316e-06, | |
| "loss": 0.0, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 66.87, | |
| "learning_rate": 5.107368421052632e-06, | |
| "loss": 0.0, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 67.21, | |
| "learning_rate": 5.081052631578948e-06, | |
| "loss": 0.0, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 67.53, | |
| "learning_rate": 5.054736842105263e-06, | |
| "loss": 0.0, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 67.53, | |
| "eval_loss": 0.7741473317146301, | |
| "eval_runtime": 20.4587, | |
| "eval_samples_per_second": 6.794, | |
| "eval_steps_per_second": 0.88, | |
| "eval_wer": 98.21002386634845, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 67.85, | |
| "learning_rate": 5.028421052631579e-06, | |
| "loss": 0.0, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 68.18, | |
| "learning_rate": 5.002105263157895e-06, | |
| "loss": 0.0, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 68.5, | |
| "learning_rate": 4.975789473684211e-06, | |
| "loss": 0.0, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 68.82, | |
| "learning_rate": 4.949473684210527e-06, | |
| "loss": 0.0, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 69.15, | |
| "learning_rate": 4.923157894736842e-06, | |
| "loss": 0.0, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 69.48, | |
| "learning_rate": 4.896842105263158e-06, | |
| "loss": 0.0, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 69.8, | |
| "learning_rate": 4.870526315789474e-06, | |
| "loss": 0.0, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 70.13, | |
| "learning_rate": 4.84421052631579e-06, | |
| "loss": 0.0, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 70.45, | |
| "learning_rate": 4.817894736842106e-06, | |
| "loss": 0.0, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 70.77, | |
| "learning_rate": 4.791578947368421e-06, | |
| "loss": 0.0, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 71.1, | |
| "learning_rate": 4.765263157894737e-06, | |
| "loss": 0.0, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 71.42, | |
| "learning_rate": 4.738947368421053e-06, | |
| "loss": 0.0, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 71.75, | |
| "learning_rate": 4.712631578947369e-06, | |
| "loss": 0.0, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 72.08, | |
| "learning_rate": 4.6863157894736845e-06, | |
| "loss": 0.0, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 72.4, | |
| "learning_rate": 4.66e-06, | |
| "loss": 0.0, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 72.72, | |
| "learning_rate": 4.633684210526316e-06, | |
| "loss": 0.0, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 72.72, | |
| "eval_loss": 0.7758468389511108, | |
| "eval_runtime": 19.2002, | |
| "eval_samples_per_second": 7.239, | |
| "eval_steps_per_second": 0.937, | |
| "eval_wer": 93.67541766109785, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 73.05, | |
| "learning_rate": 4.607368421052632e-06, | |
| "loss": 0.0, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 73.37, | |
| "learning_rate": 4.581052631578948e-06, | |
| "loss": 0.0, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 73.69, | |
| "learning_rate": 4.5547368421052636e-06, | |
| "loss": 0.0, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 74.03, | |
| "learning_rate": 4.528421052631579e-06, | |
| "loss": 0.0, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 74.35, | |
| "learning_rate": 4.502105263157895e-06, | |
| "loss": 0.0, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 74.67, | |
| "learning_rate": 4.475789473684211e-06, | |
| "loss": 0.0, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 74.99, | |
| "learning_rate": 4.449473684210527e-06, | |
| "loss": 0.0, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 75.32, | |
| "learning_rate": 4.423157894736843e-06, | |
| "loss": 0.0, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 75.64, | |
| "learning_rate": 4.3968421052631585e-06, | |
| "loss": 0.0, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 75.96, | |
| "learning_rate": 4.370526315789474e-06, | |
| "loss": 0.0, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 76.3, | |
| "learning_rate": 4.344210526315789e-06, | |
| "loss": 0.0, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 76.62, | |
| "learning_rate": 4.317894736842105e-06, | |
| "loss": 0.0, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 76.94, | |
| "learning_rate": 4.291578947368422e-06, | |
| "loss": 0.0, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 77.27, | |
| "learning_rate": 4.2652631578947376e-06, | |
| "loss": 0.0, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 77.59, | |
| "learning_rate": 4.238947368421053e-06, | |
| "loss": 0.0, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 77.91, | |
| "learning_rate": 4.212631578947368e-06, | |
| "loss": 0.0, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 77.91, | |
| "eval_loss": 0.7771462202072144, | |
| "eval_runtime": 19.1772, | |
| "eval_samples_per_second": 7.248, | |
| "eval_steps_per_second": 0.939, | |
| "eval_wer": 93.07875894988067, | |
| "step": 6000 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 130, | |
| "total_flos": 9.892111005646848e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |