| { | |
| "best_metric": 32.100299056820795, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch12-tiny/checkpoint-3000", | |
| "epoch": 7.256894049346879, | |
| "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": 1.9375, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.8381, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.6524, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.468, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 1.3267, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 1.1383, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.44e-06, | |
| "loss": 1.053, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.94e-06, | |
| "loss": 1.0537, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.9684, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.9787, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.9279, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.892, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.8551, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.8492, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.8154, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.8398, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.7837, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.7798, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.812, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.7708, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.7603, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.7113, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.7249, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.7157, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.717, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.7148, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.7042, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.6626, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.594, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.5845, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.5614, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.5359, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.5744, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.5481, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.5169, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.5265, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.5621, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.5436, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.5334, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.5121, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_loss": 0.5645238757133484, | |
| "eval_runtime": 298.983, | |
| "eval_samples_per_second": 12.114, | |
| "eval_steps_per_second": 2.02, | |
| "eval_wer": 36.259489302967566, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.4907, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.5562, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.502, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.5082, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.5139, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.5065, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.5009, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.4819, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.494, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.4816, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.4931, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.4944, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.496, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.5112, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.4972, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.3688, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.3728, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.3301, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.3493, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.3443, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.3687, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.373, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.3352, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.354, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.363, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.3635, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.3909, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.3417, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.3542, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.3202, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.3559, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.3712, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.3603, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.3679, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.3402, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.3283, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.3731, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.3282, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.3375, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.3401, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "eval_loss": 0.5052144527435303, | |
| "eval_runtime": 300.7527, | |
| "eval_samples_per_second": 12.043, | |
| "eval_steps_per_second": 2.008, | |
| "eval_wer": 32.93765815504946, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.3272, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.3313, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.2996, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.2741, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.2425, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.2394, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.2429, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.2367, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.2294, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.2449, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.255, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.257, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.2258, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.2372, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.2456, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.2521, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.2625, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.2459, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.2658, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.2582, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.2466, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.2473, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.2564, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.2852, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.245, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.2644, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.2563, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.238, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.2675, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.2459, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.206, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.1704, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.1793, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.1819, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.1865, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.1648, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.1824, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.1887, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.1817, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.1855, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "eval_loss": 0.5086480975151062, | |
| "eval_runtime": 302.0955, | |
| "eval_samples_per_second": 11.99, | |
| "eval_steps_per_second": 1.999, | |
| "eval_wer": 32.100299056820795, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.1797, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.183, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.1892, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.1744, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.1977, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.207, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.1869, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.2087, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.1898, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.1738, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.1851, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.1976, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.1907, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.1899, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.2002, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.1716, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.1981, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.1662, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.1411, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.15, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.1417, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.1423, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.1263, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.1442, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.1385, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.1588, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.1541, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.1532, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.1395, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.1469, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.1507, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.1438, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.1359, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.1526, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.1476, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.1449, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.1383, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.1324, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.1424, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.1638, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "eval_loss": 0.5087841153144836, | |
| "eval_runtime": 313.9894, | |
| "eval_samples_per_second": 11.535, | |
| "eval_steps_per_second": 1.924, | |
| "eval_wer": 32.183114791810446, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.1445, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.134, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.1387, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.1377, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.1524, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.1314, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.1138, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.1195, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.1261, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.119, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.1172, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.1077, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.1215, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.1217, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.1247, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.1175, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.1234, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.1069, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.1283, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.127, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.1142, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.1234, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.1176, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.1216, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.1076, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.1134, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 6.79, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.1292, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.1235, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.1293, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.1179, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.1255, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.1195, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.1145, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.1045, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 2.888888888888889e-07, | |
| "loss": 0.11, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 2.3333333333333336e-07, | |
| "loss": 0.1096, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 1.777777777777778e-07, | |
| "loss": 0.111, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 1.2222222222222225e-07, | |
| "loss": 0.1102, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 6.666666666666668e-08, | |
| "loss": 0.115, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 1.1111111111111112e-08, | |
| "loss": 0.1047, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "eval_loss": 0.5143000483512878, | |
| "eval_runtime": 151.0485, | |
| "eval_samples_per_second": 23.979, | |
| "eval_steps_per_second": 3.999, | |
| "eval_wer": 32.16011042097999, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "step": 5000, | |
| "total_flos": 1.47592678182912e+18, | |
| "train_loss": 0.356853648853302, | |
| "train_runtime": 4131.9082, | |
| "train_samples_per_second": 14.521, | |
| "train_steps_per_second": 1.21 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 8, | |
| "total_flos": 1.47592678182912e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |