| { | |
| "best_metric": 9.032982817995986, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 21.275715800636267, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 0.7151, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 0.4297, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.42e-06, | |
| "loss": 0.2399, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 0.1871, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.42e-06, | |
| "loss": 0.161, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.1511, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.1481, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.1392, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.1328, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.1152, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.0998, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.0936, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.0953, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.0962, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.0971, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.1003, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.0931, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.0925, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.092, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.0509, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 9.955555555555556e-06, | |
| "loss": 0.0563, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.0563, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.844444444444446e-06, | |
| "loss": 0.0581, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 9.78888888888889e-06, | |
| "loss": 0.0557, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 0.0561, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 0.0548, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 0.056, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 0.0582, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 0.0341, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 0.0256, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.0263, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 9.344444444444446e-06, | |
| "loss": 0.0275, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 9.28888888888889e-06, | |
| "loss": 0.0289, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 9.233333333333334e-06, | |
| "loss": 0.0274, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 9.17777777777778e-06, | |
| "loss": 0.0309, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 9.122222222222223e-06, | |
| "loss": 0.0307, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 0.0304, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 9.011111111111111e-06, | |
| "loss": 0.026, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 8.955555555555555e-06, | |
| "loss": 0.0139, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.0149, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "eval_loss": 0.16215506196022034, | |
| "eval_runtime": 17422.9164, | |
| "eval_samples_per_second": 0.443, | |
| "eval_steps_per_second": 0.055, | |
| "eval_wer": 10.040330748973316, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 8.844444444444445e-06, | |
| "loss": 0.0141, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 8.788888888888891e-06, | |
| "loss": 0.0159, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 0.0165, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 8.677777777777779e-06, | |
| "loss": 0.0134, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 8.622222222222223e-06, | |
| "loss": 0.0174, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 8.566666666666667e-06, | |
| "loss": 0.0147, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 8.511111111111113e-06, | |
| "loss": 0.0179, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 8.455555555555555e-06, | |
| "loss": 0.01, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.0094, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 8.344444444444445e-06, | |
| "loss": 0.0087, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 0.0087, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 0.0113, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 0.0126, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 0.0107, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.0105, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 0.0104, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 0.0082, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.0056, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 0.0056, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 0.007, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 0.0062, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 0.0058, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 0.0061, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 0.0055, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 0.006, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 0.0063, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.0043, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 0.0035, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.0035, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 0.004, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 0.0034, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 0.0042, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 7.76, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 0.0042, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 0.0042, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 7.98, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 0.0043, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.0034, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 0.0029, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 0.0034, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 0.0033, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 8.51, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.0027, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.51, | |
| "eval_loss": 0.18483087420463562, | |
| "eval_runtime": 21193.0144, | |
| "eval_samples_per_second": 0.364, | |
| "eval_steps_per_second": 0.046, | |
| "eval_wer": 9.51363694959577, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.0042, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.0042, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.0046, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 0.004, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.004, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 0.0029, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 0.0027, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.003, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 0.003, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 0.0031, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 0.0028, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 0.0029, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 9.89, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 0.003, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.0035, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 0.0023, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 0.0019, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 0.002, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 0.0024, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 0.0021, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 10.64, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 0.0021, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 0.0019, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 0.002, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 10.95, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.0018, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 0.0017, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 0.0024, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.0019, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 11.38, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 0.0018, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 11.49, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 0.0017, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 0.0014, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 0.0014, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 0.0015, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.0017, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 12.02, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 0.0016, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 0.001, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 12.23, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.001, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 12.34, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.0018, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 12.45, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 0.0014, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 12.55, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.0012, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 12.66, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.0012, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 12.76, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 0.0008, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.76, | |
| "eval_loss": 0.19296467304229736, | |
| "eval_runtime": 16619.4936, | |
| "eval_samples_per_second": 0.464, | |
| "eval_steps_per_second": 0.058, | |
| "eval_wer": 9.31658717150697, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.0009, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 12.98, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 0.001, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 13.08, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 0.0007, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 0.0007, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 13.3, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 0.0009, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 13.4, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 0.0006, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 13.51, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 0.0007, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 13.62, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 0.0006, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 0.0008, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 13.83, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.0007, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 0.0007, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 0.0006, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 0.0005, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 14.25, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 14.36, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 0.0005, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 0.0005, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 0.0005, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.0011, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 0.0005, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 0.0008, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 0.0008, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 15.21, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 0.0005, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 0.0007, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 15.42, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 15.53, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 0.0005, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 0.0009, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 15.74, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.0005, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 0.0005, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 15.95, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 0.0006, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 16.06, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 0.0005, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 16.17, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 16.28, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 0.0007, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 16.38, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 16.49, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 0.0005, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 16.59, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.0005, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 0.0004, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 16.91, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 0.0004, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 0.0004, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "eval_loss": 0.20616495609283447, | |
| "eval_runtime": 12625.6787, | |
| "eval_samples_per_second": 0.611, | |
| "eval_steps_per_second": 0.076, | |
| "eval_wer": 9.032982817995986, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 17.13, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 0.0004, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 0.0004, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 17.34, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 0.0004, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 17.45, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 0.0004, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 17.55, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 17.66, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.0004, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 17.76, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.0004, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 17.87, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.0004, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 17.98, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.0004, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 18.08, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.0004, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 18.19, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.0004, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 18.3, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.0004, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 18.4, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.0003, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 18.51, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.0003, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 18.62, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.0004, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 18.72, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.0003, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 18.83, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 18.93, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.0004, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.0003, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 19.15, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.0003, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 19.25, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 19.36, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.0003, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 19.47, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.0003, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 19.57, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.0003, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 19.68, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.0003, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 19.78, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.0003, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 19.89, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.0003, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.0003, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 20.11, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.0003, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 20.21, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.0003, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 20.32, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.0003, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 20.42, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.0003, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 20.53, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.0003, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 20.64, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.0003, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 20.74, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.0003, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 20.85, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.0003, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 20.95, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.0003, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 21.06, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.0003, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 21.17, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.0003, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 21.28, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.0003, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.28, | |
| "eval_loss": 0.2131042629480362, | |
| "eval_runtime": 12843.4698, | |
| "eval_samples_per_second": 0.601, | |
| "eval_steps_per_second": 0.075, | |
| "eval_wer": 9.044032338262648, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 21.28, | |
| "step": 5000, | |
| "total_flos": 6.814657098620928e+20, | |
| "train_loss": 0.02271956424098462, | |
| "train_runtime": 245218.2405, | |
| "train_samples_per_second": 1.305, | |
| "train_steps_per_second": 0.02 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 22, | |
| "total_flos": 6.814657098620928e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |