| { | |
| "best_metric": 76.37636988522145, | |
| "best_model_checkpoint": "./output/small/yt-special-batch8-tiny/checkpoint-5000", | |
| "epoch": 0.8588114050154586, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 3.0375, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 2.8516, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 2.8996, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 2.413, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 2.2213, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9e-06, | |
| "loss": 2.0832, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.7823, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 1.7617, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.8501, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 1.7876, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 1.7035, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 5.9e-06, | |
| "loss": 1.7522, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.584, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 6.9e-06, | |
| "loss": 1.6716, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 7.4e-06, | |
| "loss": 1.796, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.9e-06, | |
| "loss": 1.5802, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.4449, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 1.6765, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.6156, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.5887, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.955555555555556e-06, | |
| "loss": 1.5356, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.9e-06, | |
| "loss": 1.5713, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.844444444444446e-06, | |
| "loss": 1.5336, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.78888888888889e-06, | |
| "loss": 1.6768, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 1.4941, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 1.5387, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 1.5816, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 1.4096, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 1.4932, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 1.5254, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.4e-06, | |
| "loss": 1.3965, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.344444444444446e-06, | |
| "loss": 1.4093, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.28888888888889e-06, | |
| "loss": 1.3946, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.233333333333334e-06, | |
| "loss": 1.3491, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.17777777777778e-06, | |
| "loss": 1.4431, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.122222222222223e-06, | |
| "loss": 1.4492, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 1.4466, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.011111111111111e-06, | |
| "loss": 1.3713, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.955555555555555e-06, | |
| "loss": 1.4049, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 1.4295, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "eval_loss": 1.4093937873840332, | |
| "eval_runtime": 278.5863, | |
| "eval_samples_per_second": 8.798, | |
| "eval_steps_per_second": 2.2, | |
| "eval_wer": 110.15498346410739, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.844444444444445e-06, | |
| "loss": 1.5387, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.788888888888891e-06, | |
| "loss": 1.3363, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 1.4644, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.677777777777779e-06, | |
| "loss": 1.3795, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.622222222222223e-06, | |
| "loss": 1.3536, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.566666666666667e-06, | |
| "loss": 1.5356, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.511111111111113e-06, | |
| "loss": 1.5368, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.455555555555555e-06, | |
| "loss": 1.3584, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 1.3795, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.344444444444445e-06, | |
| "loss": 1.5068, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 1.2671, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 1.4027, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 1.3407, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 1.3717, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 1.4011, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 1.3397, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 1.4166, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.9e-06, | |
| "loss": 1.3022, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 1.3544, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 1.3412, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 1.5112, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 1.3038, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 1.3752, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 1.3446, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 1.2387, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 1.3236, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.4e-06, | |
| "loss": 1.3304, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 1.2388, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 1.3252, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 1.2723, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 1.3858, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 1.2257, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 1.3428, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 1.3284, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 1.2452, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.9e-06, | |
| "loss": 1.3218, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 1.3243, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 1.3978, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 1.2782, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 1.3042, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "eval_loss": 1.2886050939559937, | |
| "eval_runtime": 281.1051, | |
| "eval_samples_per_second": 8.719, | |
| "eval_steps_per_second": 2.181, | |
| "eval_wer": 86.29142078983205, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 1.2557, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 1.3275, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 1.3171, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 1.2166, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 1.1869, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 1.3423, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 1.3287, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 1.2774, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 1.2855, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 1.2227, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 1.198, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 1.3709, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 1.3011, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.9e-06, | |
| "loss": 1.3681, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 1.3086, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 1.2167, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 1.2282, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 1.266, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 1.2449, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 1.2406, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 1.2414, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 1.3149, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 1.2321, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 1.1998, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 1.2292, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 1.3701, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 1.2846, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 1.2457, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 1.2912, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 1.2478, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 1.3023, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 1.1222, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 1.1982, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 1.2535, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 1.307, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 1.1476, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 1.3246, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 1.3723, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 1.192, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 1.2212, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_loss": 1.2206178903579712, | |
| "eval_runtime": 254.8252, | |
| "eval_samples_per_second": 9.618, | |
| "eval_steps_per_second": 2.406, | |
| "eval_wer": 84.11905842682057, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.2387, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 1.1781, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 1.0837, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 1.2201, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 1.3045, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 1.2479, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 1.1863, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 1.3211, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 1.2037, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 1.153, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 1.252, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 1.1224, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 1.1686, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 1.1249, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 1.1661, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 1.1425, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 1.3145, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 1.212, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.3946, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 1.1685, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 1.253, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 1.0476, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 1.2093, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 1.1977, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 1.2082, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 1.1935, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 1.2793, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.9e-06, | |
| "loss": 1.2551, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 1.0723, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 1.2802, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 1.2181, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 1.1853, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 1.1862, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 1.2022, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 1.261, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 1.2105, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.0427, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 1.293, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 1.1512, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 1.1306, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "eval_loss": 1.1814459562301636, | |
| "eval_runtime": 194.4571, | |
| "eval_samples_per_second": 12.604, | |
| "eval_steps_per_second": 3.152, | |
| "eval_wer": 78.15316775825173, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 1.2084, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 1.2137, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 1.195, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 1.0943, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 1.2158, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 1.1431, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 1.2699, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 1.1721, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 1.2271, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 1.1629, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 1.3203, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 1.1082, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 1.1323, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 1.335, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 1.0591, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 1.3213, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 1.2386, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 1.1389, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 1.3361, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 1.1915, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 1.067, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 1.1499, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 1.2431, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 1.2343, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 1.152, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 1.2086, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 1.229, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 1.2757, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 1.1664, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 1.0384, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 1.2596, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 1.1237, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 1.0471, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 1.0879, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 1.1853, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 1.1763, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 1.1031, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 1.1508, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 1.0186, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 1.1333, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "eval_loss": 1.1651121377944946, | |
| "eval_runtime": 180.3205, | |
| "eval_samples_per_second": 13.592, | |
| "eval_steps_per_second": 3.4, | |
| "eval_wer": 76.37636988522145, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "step": 5000, | |
| "total_flos": 9.847554048e+17, | |
| "train_loss": 1.343574110031128, | |
| "train_runtime": 2849.8517, | |
| "train_samples_per_second": 14.036, | |
| "train_steps_per_second": 1.754 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 1, | |
| "total_flos": 9.847554048e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |