| { | |
| "best_metric": 5.397983265393693, | |
| "best_model_checkpoint": "./output/small/yt-special-batch8-tiny/checkpoint-5000", | |
| "epoch": 7.9239302694136295, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4e-07, | |
| "loss": 2.0472, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.8247, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.4094, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.2571, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 1.4853, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 2.9638, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 5.2628, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 7.8449, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.38e-06, | |
| "loss": 15.3839, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 17.3879, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.380000000000001e-06, | |
| "loss": 11.4254, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 8.3453, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.380000000000001e-06, | |
| "loss": 6.8998, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 6.88e-06, | |
| "loss": 7.6137, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.3800000000000005e-06, | |
| "loss": 6.4318, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.88e-06, | |
| "loss": 4.926, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.380000000000001e-06, | |
| "loss": 5.7242, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.880000000000001e-06, | |
| "loss": 4.7901, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.38e-06, | |
| "loss": 5.1702, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.88e-06, | |
| "loss": 5.1131, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.957777777777779e-06, | |
| "loss": 5.3323, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.902222222222223e-06, | |
| "loss": 4.4564, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.846666666666668e-06, | |
| "loss": 4.3714, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.791111111111112e-06, | |
| "loss": 4.2409, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.735555555555556e-06, | |
| "loss": 4.4811, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.68e-06, | |
| "loss": 4.0169, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.624444444444445e-06, | |
| "loss": 3.9582, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.56888888888889e-06, | |
| "loss": 4.1751, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.513333333333334e-06, | |
| "loss": 3.5919, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.457777777777778e-06, | |
| "loss": 3.9336, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.402222222222222e-06, | |
| "loss": 3.7917, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.346666666666666e-06, | |
| "loss": 3.503, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.291111111111112e-06, | |
| "loss": 3.7914, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.235555555555556e-06, | |
| "loss": 4.126, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.180000000000002e-06, | |
| "loss": 3.6003, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.124444444444444e-06, | |
| "loss": 3.7748, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.06888888888889e-06, | |
| "loss": 3.6516, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.013333333333334e-06, | |
| "loss": 3.6103, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 8.957777777777778e-06, | |
| "loss": 3.4672, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.902222222222224e-06, | |
| "loss": 3.4292, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_loss": 2.7892673015594482, | |
| "eval_runtime": 1660.416, | |
| "eval_samples_per_second": 3.037, | |
| "eval_steps_per_second": 0.759, | |
| "eval_wer": 302.9156833297576, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 8.846666666666668e-06, | |
| "loss": 3.3502, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.791111111111112e-06, | |
| "loss": 3.2234, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.735555555555556e-06, | |
| "loss": 3.4342, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.68e-06, | |
| "loss": 3.3114, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.624444444444446e-06, | |
| "loss": 2.9772, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 8.56888888888889e-06, | |
| "loss": 3.1963, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 8.515555555555556e-06, | |
| "loss": 3.1974, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.46e-06, | |
| "loss": 3.5305, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.404444444444444e-06, | |
| "loss": 3.2849, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.34888888888889e-06, | |
| "loss": 3.0515, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.293333333333334e-06, | |
| "loss": 2.668, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.237777777777778e-06, | |
| "loss": 2.4469, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 8.182222222222222e-06, | |
| "loss": 2.6318, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 8.126666666666668e-06, | |
| "loss": 2.441, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.071111111111112e-06, | |
| "loss": 2.6253, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.015555555555556e-06, | |
| "loss": 2.3484, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 2.6521, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 7.904444444444444e-06, | |
| "loss": 2.5311, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.84888888888889e-06, | |
| "loss": 2.5159, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.793333333333334e-06, | |
| "loss": 2.6463, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.737777777777778e-06, | |
| "loss": 2.454, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 7.682222222222224e-06, | |
| "loss": 2.6442, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.626666666666668e-06, | |
| "loss": 2.5474, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.571111111111112e-06, | |
| "loss": 2.5901, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.515555555555556e-06, | |
| "loss": 2.2717, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 2.2699, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 7.4044444444444455e-06, | |
| "loss": 2.4251, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.3488888888888895e-06, | |
| "loss": 2.3752, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 7.2933333333333335e-06, | |
| "loss": 2.4347, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 7.237777777777778e-06, | |
| "loss": 2.36, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.1822222222222224e-06, | |
| "loss": 2.2226, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.126666666666667e-06, | |
| "loss": 2.4829, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.071111111111112e-06, | |
| "loss": 2.6744, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 7.015555555555556e-06, | |
| "loss": 2.4284, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 6.96e-06, | |
| "loss": 2.1663, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.904444444444444e-06, | |
| "loss": 2.0312, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.848888888888889e-06, | |
| "loss": 1.6857, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.793333333333334e-06, | |
| "loss": 1.6288, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.737777777777779e-06, | |
| "loss": 1.7115, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 6.682222222222223e-06, | |
| "loss": 1.7988, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "eval_loss": 1.5462772846221924, | |
| "eval_runtime": 1006.8394, | |
| "eval_samples_per_second": 5.009, | |
| "eval_steps_per_second": 1.252, | |
| "eval_wer": 110.16520060072945, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 6.626666666666667e-06, | |
| "loss": 1.8027, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 6.571111111111111e-06, | |
| "loss": 1.7406, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 6.515555555555556e-06, | |
| "loss": 1.8264, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 1.7339, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.404444444444446e-06, | |
| "loss": 1.5614, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 6.348888888888889e-06, | |
| "loss": 1.7282, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.293333333333334e-06, | |
| "loss": 1.6852, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 6.237777777777778e-06, | |
| "loss": 1.6439, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 6.182222222222223e-06, | |
| "loss": 1.8005, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.126666666666668e-06, | |
| "loss": 1.7866, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 6.0711111111111125e-06, | |
| "loss": 1.6142, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 6.015555555555556e-06, | |
| "loss": 1.5877, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 1.7717, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 5.9044444444444446e-06, | |
| "loss": 1.7162, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 5.8488888888888895e-06, | |
| "loss": 1.6105, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 5.793333333333334e-06, | |
| "loss": 1.6877, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 5.737777777777778e-06, | |
| "loss": 1.5392, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.682222222222222e-06, | |
| "loss": 1.6446, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 5.626666666666667e-06, | |
| "loss": 1.6507, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.571111111111111e-06, | |
| "loss": 1.6794, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.515555555555556e-06, | |
| "loss": 1.5191, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 1.0089, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.404444444444444e-06, | |
| "loss": 1.0366, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 5.348888888888889e-06, | |
| "loss": 1.1613, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 5.293333333333334e-06, | |
| "loss": 1.1485, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 5.237777777777778e-06, | |
| "loss": 1.1352, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 5.182222222222223e-06, | |
| "loss": 1.2326, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 5.126666666666668e-06, | |
| "loss": 1.1992, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 5.071111111111111e-06, | |
| "loss": 1.0852, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 5.015555555555556e-06, | |
| "loss": 1.0308, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 1.0451, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 4.904444444444445e-06, | |
| "loss": 1.2291, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 4.848888888888889e-06, | |
| "loss": 1.0855, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 4.793333333333334e-06, | |
| "loss": 1.1677, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 4.737777777777779e-06, | |
| "loss": 1.1195, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 4.682222222222223e-06, | |
| "loss": 1.0037, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 4.626666666666667e-06, | |
| "loss": 1.002, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 4.571111111111112e-06, | |
| "loss": 1.0344, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 4.515555555555556e-06, | |
| "loss": 1.0506, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 1.083, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "eval_loss": 0.7804726362228394, | |
| "eval_runtime": 860.0655, | |
| "eval_samples_per_second": 5.864, | |
| "eval_steps_per_second": 1.466, | |
| "eval_wer": 76.9319888435958, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 4.404444444444445e-06, | |
| "loss": 1.0825, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 4.348888888888889e-06, | |
| "loss": 0.9834, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 4.2933333333333334e-06, | |
| "loss": 1.105, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 4.2377777777777775e-06, | |
| "loss": 1.0449, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 4.182222222222222e-06, | |
| "loss": 0.9722, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 4.126666666666667e-06, | |
| "loss": 1.0611, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 4.071111111111111e-06, | |
| "loss": 0.7303, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 4.015555555555556e-06, | |
| "loss": 0.7109, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.7129, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.904444444444444e-06, | |
| "loss": 0.6991, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.848888888888889e-06, | |
| "loss": 0.6753, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.793333333333334e-06, | |
| "loss": 0.6539, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 3.737777777777778e-06, | |
| "loss": 0.7438, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.6822222222222225e-06, | |
| "loss": 0.6727, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.6266666666666674e-06, | |
| "loss": 0.7528, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.5711111111111114e-06, | |
| "loss": 0.7318, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 3.515555555555556e-06, | |
| "loss": 0.7009, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.7471, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.404444444444445e-06, | |
| "loss": 0.6263, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 3.3488888888888892e-06, | |
| "loss": 0.6796, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 3.2933333333333333e-06, | |
| "loss": 0.6472, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 3.237777777777778e-06, | |
| "loss": 0.7417, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.1822222222222226e-06, | |
| "loss": 0.9388, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 3.1266666666666667e-06, | |
| "loss": 0.6601, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 3.0711111111111115e-06, | |
| "loss": 0.7833, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.015555555555556e-06, | |
| "loss": 0.6362, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.5105, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 2.904444444444445e-06, | |
| "loss": 0.6755, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 2.8488888888888894e-06, | |
| "loss": 0.6358, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 2.7933333333333334e-06, | |
| "loss": 0.6796, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 2.7377777777777783e-06, | |
| "loss": 0.5597, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.6822222222222223e-06, | |
| "loss": 0.4393, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.6266666666666668e-06, | |
| "loss": 0.4255, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2.5711111111111112e-06, | |
| "loss": 0.3777, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 2.5155555555555557e-06, | |
| "loss": 0.37, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.3804, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 2.4044444444444446e-06, | |
| "loss": 0.4035, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 2.348888888888889e-06, | |
| "loss": 0.359, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 2.2933333333333335e-06, | |
| "loss": 0.3744, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 2.237777777777778e-06, | |
| "loss": 0.3718, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "eval_loss": 0.31917494535446167, | |
| "eval_runtime": 741.7781, | |
| "eval_samples_per_second": 6.799, | |
| "eval_steps_per_second": 1.7, | |
| "eval_wer": 20.596438532503754, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 2.1822222222222225e-06, | |
| "loss": 0.3978, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 2.126666666666667e-06, | |
| "loss": 0.3506, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 2.0711111111111114e-06, | |
| "loss": 0.3445, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 2.0155555555555554e-06, | |
| "loss": 0.356, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.3703, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 1.9044444444444445e-06, | |
| "loss": 0.3256, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 1.848888888888889e-06, | |
| "loss": 0.2961, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 1.7933333333333337e-06, | |
| "loss": 0.3836, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 1.737777777777778e-06, | |
| "loss": 0.3737, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 1.6822222222222224e-06, | |
| "loss": 0.3581, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 1.6266666666666666e-06, | |
| "loss": 0.3401, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.5711111111111113e-06, | |
| "loss": 0.3376, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 1.5155555555555558e-06, | |
| "loss": 0.3044, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.3738, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 1.4044444444444447e-06, | |
| "loss": 0.3131, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 1.3488888888888891e-06, | |
| "loss": 0.297, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.2933333333333334e-06, | |
| "loss": 0.2354, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.2377777777777778e-06, | |
| "loss": 0.1903, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 1.1822222222222223e-06, | |
| "loss": 0.1756, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 1.1266666666666667e-06, | |
| "loss": 0.1705, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 1.0711111111111112e-06, | |
| "loss": 0.1619, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 1.0155555555555557e-06, | |
| "loss": 0.1681, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 0.171, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 9.044444444444445e-07, | |
| "loss": 0.1677, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 8.488888888888889e-07, | |
| "loss": 0.1992, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 7.933333333333335e-07, | |
| "loss": 0.1542, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 7.377777777777779e-07, | |
| "loss": 0.1572, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 6.822222222222223e-07, | |
| "loss": 0.1511, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 6.266666666666667e-07, | |
| "loss": 0.1958, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 5.711111111111111e-07, | |
| "loss": 0.1586, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 5.155555555555556e-07, | |
| "loss": 0.1713, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 0.1402, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 4.0444444444444445e-07, | |
| "loss": 0.1408, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 3.488888888888889e-07, | |
| "loss": 0.1561, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 2.9333333333333337e-07, | |
| "loss": 0.1437, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 2.3777777777777777e-07, | |
| "loss": 0.138, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 1.8222222222222226e-07, | |
| "loss": 0.1529, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 1.2666666666666666e-07, | |
| "loss": 0.1303, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 7.111111111111112e-08, | |
| "loss": 0.1428, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 1.5555555555555557e-08, | |
| "loss": 0.1292, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "eval_loss": 0.08825201541185379, | |
| "eval_runtime": 742.7307, | |
| "eval_samples_per_second": 6.79, | |
| "eval_steps_per_second": 1.698, | |
| "eval_wer": 5.397983265393693, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "step": 5000, | |
| "total_flos": 9.838937438208e+17, | |
| "train_loss": 1.992731960439682, | |
| "train_runtime": 9973.9472, | |
| "train_samples_per_second": 4.01, | |
| "train_steps_per_second": 0.501 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 8, | |
| "total_flos": 9.838937438208e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |