| { | |
| "best_metric": 12.666206579250058, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch8-grad2-small/checkpoint-3000", | |
| "epoch": 9.68054211035818, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 1.3439, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 1.2498, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.9629, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.675, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.4729, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.4846, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.4205, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.4407, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.4055, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.3558, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.3636, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.3824, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.3401, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.3665, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.3671, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.3398, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.332, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.3269, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.3379, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.3343, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.94888888888889e-06, | |
| "loss": 0.3209, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.893333333333334e-06, | |
| "loss": 0.2115, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.837777777777778e-06, | |
| "loss": 0.2083, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 9.782222222222222e-06, | |
| "loss": 0.2037, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 9.726666666666668e-06, | |
| "loss": 0.207, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.671111111111112e-06, | |
| "loss": 0.2431, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.615555555555558e-06, | |
| "loss": 0.2192, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 9.56e-06, | |
| "loss": 0.2025, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.504444444444446e-06, | |
| "loss": 0.2025, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 9.44888888888889e-06, | |
| "loss": 0.215, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.393333333333334e-06, | |
| "loss": 0.2139, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.33777777777778e-06, | |
| "loss": 0.209, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.282222222222222e-06, | |
| "loss": 0.1885, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.226666666666668e-06, | |
| "loss": 0.2045, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.171111111111112e-06, | |
| "loss": 0.2011, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 9.115555555555556e-06, | |
| "loss": 0.2032, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.060000000000001e-06, | |
| "loss": 0.2006, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.004444444444445e-06, | |
| "loss": 0.185, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 8.94888888888889e-06, | |
| "loss": 0.1881, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.893333333333333e-06, | |
| "loss": 0.2062, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_loss": 0.24846424162387848, | |
| "eval_runtime": 590.5214, | |
| "eval_samples_per_second": 6.134, | |
| "eval_steps_per_second": 1.534, | |
| "eval_wer": 14.363929146537844, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.83777777777778e-06, | |
| "loss": 0.1929, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.782222222222223e-06, | |
| "loss": 0.1334, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 8.726666666666667e-06, | |
| "loss": 0.0826, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.671111111111113e-06, | |
| "loss": 0.0782, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.615555555555555e-06, | |
| "loss": 0.0781, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.560000000000001e-06, | |
| "loss": 0.0746, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 8.504444444444445e-06, | |
| "loss": 0.0782, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 8.448888888888889e-06, | |
| "loss": 0.0934, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 8.393333333333335e-06, | |
| "loss": 0.0855, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 8.337777777777777e-06, | |
| "loss": 0.0826, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.282222222222223e-06, | |
| "loss": 0.0817, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.226666666666667e-06, | |
| "loss": 0.0822, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 8.171111111111113e-06, | |
| "loss": 0.086, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 8.115555555555557e-06, | |
| "loss": 0.1015, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.0828, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 8.004444444444445e-06, | |
| "loss": 0.0858, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 7.948888888888889e-06, | |
| "loss": 0.0834, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.893333333333335e-06, | |
| "loss": 0.0859, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 7.837777777777779e-06, | |
| "loss": 0.0858, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 7.782222222222223e-06, | |
| "loss": 0.0893, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 7.726666666666667e-06, | |
| "loss": 0.0838, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.67111111111111e-06, | |
| "loss": 0.0903, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 7.6155555555555564e-06, | |
| "loss": 0.0339, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 0.0418, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 7.504444444444445e-06, | |
| "loss": 0.0379, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 7.44888888888889e-06, | |
| "loss": 0.0407, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 7.393333333333333e-06, | |
| "loss": 0.0348, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 7.337777777777778e-06, | |
| "loss": 0.0361, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 7.282222222222222e-06, | |
| "loss": 0.0314, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 7.226666666666667e-06, | |
| "loss": 0.0368, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 7.171111111111112e-06, | |
| "loss": 0.0333, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 7.115555555555557e-06, | |
| "loss": 0.0316, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.0396, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 7.004444444444445e-06, | |
| "loss": 0.0352, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 6.948888888888889e-06, | |
| "loss": 0.0357, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 6.893333333333334e-06, | |
| "loss": 0.0392, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 6.837777777777779e-06, | |
| "loss": 0.0369, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 6.782222222222222e-06, | |
| "loss": 0.0373, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 6.726666666666667e-06, | |
| "loss": 0.0344, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.671111111111112e-06, | |
| "loss": 0.0373, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "eval_loss": 0.2621954083442688, | |
| "eval_runtime": 597.9047, | |
| "eval_samples_per_second": 6.058, | |
| "eval_steps_per_second": 1.515, | |
| "eval_wer": 13.41154819415689, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 6.615555555555556e-06, | |
| "loss": 0.0373, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 0.0416, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 6.504444444444446e-06, | |
| "loss": 0.0265, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 6.448888888888889e-06, | |
| "loss": 0.0181, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 6.393333333333334e-06, | |
| "loss": 0.0178, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 6.3377777777777786e-06, | |
| "loss": 0.0179, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 6.282222222222223e-06, | |
| "loss": 0.0211, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 6.2266666666666675e-06, | |
| "loss": 0.0156, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 6.171111111111112e-06, | |
| "loss": 0.0175, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 6.1155555555555555e-06, | |
| "loss": 0.0161, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.0172, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 6.004444444444445e-06, | |
| "loss": 0.0166, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 5.948888888888889e-06, | |
| "loss": 0.0169, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 5.893333333333334e-06, | |
| "loss": 0.0174, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 5.837777777777777e-06, | |
| "loss": 0.0188, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 5.782222222222222e-06, | |
| "loss": 0.0139, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 5.726666666666667e-06, | |
| "loss": 0.0198, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.671111111111112e-06, | |
| "loss": 0.0157, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 5.615555555555556e-06, | |
| "loss": 0.0167, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 5.560000000000001e-06, | |
| "loss": 0.016, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 5.504444444444444e-06, | |
| "loss": 0.0129, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.448888888888889e-06, | |
| "loss": 0.013, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 5.393333333333334e-06, | |
| "loss": 0.0145, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 5.337777777777779e-06, | |
| "loss": 0.0098, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 5.282222222222223e-06, | |
| "loss": 0.0086, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 5.226666666666667e-06, | |
| "loss": 0.0087, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 5.171111111111111e-06, | |
| "loss": 0.0056, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 5.115555555555556e-06, | |
| "loss": 0.0083, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 0.0067, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 5.004444444444445e-06, | |
| "loss": 0.0079, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 4.94888888888889e-06, | |
| "loss": 0.0069, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 4.893333333333334e-06, | |
| "loss": 0.0097, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 4.837777777777778e-06, | |
| "loss": 0.0111, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 4.7822222222222226e-06, | |
| "loss": 0.0093, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.57, | |
| "learning_rate": 4.7266666666666674e-06, | |
| "loss": 0.0071, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 4.6711111111111115e-06, | |
| "loss": 0.0064, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 4.6155555555555555e-06, | |
| "loss": 0.0061, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 4.56e-06, | |
| "loss": 0.0078, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 4.504444444444444e-06, | |
| "loss": 0.0066, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 4.448888888888889e-06, | |
| "loss": 0.0077, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "eval_loss": 0.3042357265949249, | |
| "eval_runtime": 592.9763, | |
| "eval_samples_per_second": 6.108, | |
| "eval_steps_per_second": 1.528, | |
| "eval_wer": 12.666206579250058, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 4.393333333333334e-06, | |
| "loss": 0.0076, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 4.337777777777778e-06, | |
| "loss": 0.0099, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 4.282222222222222e-06, | |
| "loss": 0.0064, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 4.226666666666667e-06, | |
| "loss": 0.006, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 4.171111111111111e-06, | |
| "loss": 0.0047, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 4.115555555555556e-06, | |
| "loss": 0.0034, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 0.0032, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 4.004444444444445e-06, | |
| "loss": 0.0034, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 3.948888888888889e-06, | |
| "loss": 0.0035, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 3.893333333333333e-06, | |
| "loss": 0.0039, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 3.837777777777778e-06, | |
| "loss": 0.0038, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 3.782222222222223e-06, | |
| "loss": 0.0043, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 3.726666666666667e-06, | |
| "loss": 0.0042, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 3.6711111111111113e-06, | |
| "loss": 0.0046, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 3.615555555555556e-06, | |
| "loss": 0.0043, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 3.5600000000000002e-06, | |
| "loss": 0.0063, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 3.5044444444444447e-06, | |
| "loss": 0.0029, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 3.4488888888888896e-06, | |
| "loss": 0.0028, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "learning_rate": 3.3933333333333336e-06, | |
| "loss": 0.004, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 3.337777777777778e-06, | |
| "loss": 0.0028, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 3.282222222222223e-06, | |
| "loss": 0.0038, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 3.226666666666667e-06, | |
| "loss": 0.0039, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 3.1711111111111114e-06, | |
| "loss": 0.0033, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 3.1155555555555555e-06, | |
| "loss": 0.003, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 0.0028, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 3.004444444444445e-06, | |
| "loss": 0.0023, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 2.948888888888889e-06, | |
| "loss": 0.0021, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 2.8933333333333337e-06, | |
| "loss": 0.0021, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 2.837777777777778e-06, | |
| "loss": 0.0034, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 2.7822222222222222e-06, | |
| "loss": 0.002, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 2.726666666666667e-06, | |
| "loss": 0.003, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 2.6711111111111116e-06, | |
| "loss": 0.0027, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 2.6155555555555556e-06, | |
| "loss": 0.0023, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 2.56e-06, | |
| "loss": 0.0022, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 2.504444444444445e-06, | |
| "loss": 0.0019, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 2.448888888888889e-06, | |
| "loss": 0.0022, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 2.3933333333333334e-06, | |
| "loss": 0.0018, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 2.337777777777778e-06, | |
| "loss": 0.0021, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 2.2822222222222223e-06, | |
| "loss": 0.0018, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 2.226666666666667e-06, | |
| "loss": 0.0019, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "eval_loss": 0.3232622444629669, | |
| "eval_runtime": 593.0232, | |
| "eval_samples_per_second": 6.108, | |
| "eval_steps_per_second": 1.528, | |
| "eval_wer": 12.804232804232804, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 2.1711111111111113e-06, | |
| "loss": 0.0025, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 2.1155555555555557e-06, | |
| "loss": 0.0018, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 2.06e-06, | |
| "loss": 0.0018, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 2.0044444444444446e-06, | |
| "loss": 0.0034, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 1.948888888888889e-06, | |
| "loss": 0.0019, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 1.8933333333333333e-06, | |
| "loss": 0.0016, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 1.837777777777778e-06, | |
| "loss": 0.0017, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 8.13, | |
| "learning_rate": 1.7822222222222225e-06, | |
| "loss": 0.0016, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 1.7266666666666667e-06, | |
| "loss": 0.0015, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 8.23, | |
| "learning_rate": 1.6711111111111112e-06, | |
| "loss": 0.0015, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 1.6155555555555559e-06, | |
| "loss": 0.0021, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 1.56e-06, | |
| "loss": 0.0015, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 1.5044444444444446e-06, | |
| "loss": 0.0015, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "learning_rate": 1.4488888888888892e-06, | |
| "loss": 0.0015, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 1.3933333333333335e-06, | |
| "loss": 0.0015, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 1.337777777777778e-06, | |
| "loss": 0.0016, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 1.2822222222222222e-06, | |
| "loss": 0.0016, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 1.2266666666666666e-06, | |
| "loss": 0.0015, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 8.66, | |
| "learning_rate": 1.171111111111111e-06, | |
| "loss": 0.0016, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 1.1155555555555558e-06, | |
| "loss": 0.0017, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 1.06e-06, | |
| "loss": 0.0018, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 1.0044444444444445e-06, | |
| "loss": 0.0015, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 9.488888888888889e-07, | |
| "loss": 0.0021, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 8.933333333333334e-07, | |
| "loss": 0.0017, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 8.37777777777778e-07, | |
| "loss": 0.002, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 7.822222222222223e-07, | |
| "loss": 0.0015, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 7.266666666666668e-07, | |
| "loss": 0.0014, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "learning_rate": 6.711111111111111e-07, | |
| "loss": 0.0014, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 6.155555555555556e-07, | |
| "loss": 0.0015, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 5.6e-07, | |
| "loss": 0.0014, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 5.044444444444445e-07, | |
| "loss": 0.0015, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 9.29, | |
| "learning_rate": 4.488888888888889e-07, | |
| "loss": 0.0013, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 3.9333333333333336e-07, | |
| "loss": 0.0014, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "learning_rate": 3.3777777777777777e-07, | |
| "loss": 0.0014, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.822222222222222e-07, | |
| "loss": 0.0014, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 9.49, | |
| "learning_rate": 2.266666666666667e-07, | |
| "loss": 0.0014, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 1.7111111111111114e-07, | |
| "loss": 0.0014, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 9.58, | |
| "learning_rate": 1.1555555555555556e-07, | |
| "loss": 0.0014, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 9.63, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 0.0017, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 4.444444444444445e-09, | |
| "loss": 0.0013, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "eval_loss": 0.33058518171310425, | |
| "eval_runtime": 588.7842, | |
| "eval_samples_per_second": 6.152, | |
| "eval_steps_per_second": 1.539, | |
| "eval_wer": 12.919254658385093, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "step": 5000, | |
| "total_flos": 2.307904021979136e+19, | |
| "train_loss": 0.0899478263899684, | |
| "train_runtime": 12142.2545, | |
| "train_samples_per_second": 6.589, | |
| "train_steps_per_second": 0.412 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.307904021979136e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |