| { | |
| "best_metric": 0.1995279982836301, | |
| "best_model_checkpoint": "./output/small/special3/checkpoint-5000", | |
| "epoch": 11.876484560570072, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.4e-07, | |
| "loss": 1.4885, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.334, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.0739, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.7772, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.5313, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.4995, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.4633, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.4224, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.4044, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.4048, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.4277, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.3991, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.3876, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.3825, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.397, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.3506, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.347, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.2529, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.2574, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.2347, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.2218, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.2596, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.2479, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.218, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.2258, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.2482, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.2199, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.236, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.2305, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.2464, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.2507, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.2549, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.2249, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.1863, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.0872, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.0834, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.0901, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.0935, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.0903, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.0959, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "eval_loss": 0.07135879993438721, | |
| "eval_runtime": 1188.6362, | |
| "eval_samples_per_second": 4.243, | |
| "eval_steps_per_second": 0.708, | |
| "eval_wer": 2.8062647500536366, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.0971, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.0934, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.0937, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.0884, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.095, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.0959, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.0888, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.0968, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.1072, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.0849, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.0699, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.0404, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.0357, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.0312, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.0387, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.0358, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.0316, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.0382, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.0375, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.0441, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.0423, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.038, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.0404, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.0384, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.0388, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.039, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.0381, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.0252, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.0181, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.0177, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.0178, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.0187, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.0171, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.0161, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.0205, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.0159, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.0191, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.0161, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.0177, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.0152, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "eval_loss": 0.01105912309139967, | |
| "eval_runtime": 1201.4441, | |
| "eval_samples_per_second": 4.197, | |
| "eval_steps_per_second": 0.7, | |
| "eval_wer": 0.49131087749409996, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.0153, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.0139, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.019, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.0197, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.0125, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.0097, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.0084, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.0103, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.0102, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.0066, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.01, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.0065, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.0104, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.0069, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.0084, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.0099, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.0099, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.0059, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.0094, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.0079, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.0074, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.0063, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.0051, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.0038, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.0035, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.0045, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.0025, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.0029, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.0053, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.0042, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.0032, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 6.65, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.0035, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.0057, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.005, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.0034, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.0037, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 6.95, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.0058, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.0057, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.0024, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.0022, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "eval_loss": 0.002171080093830824, | |
| "eval_runtime": 1265.904, | |
| "eval_samples_per_second": 3.984, | |
| "eval_steps_per_second": 0.664, | |
| "eval_wer": 0.20596438532503755, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.002, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.0018, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.002, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.0018, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.002, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.002, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.0023, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.0018, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.0028, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.0021, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.0017, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.002, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.0017, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.0034, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.0017, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.0015, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.0024, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.0013, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.0024, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.0013, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.0013, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.0012, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.0012, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.0013, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.0012, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.0024, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.0012, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.0012, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.0014, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.0012, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.0027, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.0012, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.0013, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.0011, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.0011, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.001, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.0011, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 9.38, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.0016, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.001, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.0014, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "eval_loss": 0.0010555003536865115, | |
| "eval_runtime": 1374.5066, | |
| "eval_samples_per_second": 3.669, | |
| "eval_steps_per_second": 0.612, | |
| "eval_wer": 0.2188371594078524, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 2.1733333333333334e-06, | |
| "loss": 0.001, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.001, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.001, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.001, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.0011, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.001, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.001, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.001, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.001, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 10.1, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.0009, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.0009, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.0017, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 10.27, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.0009, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 10.33, | |
| "learning_rate": 1.4511111111111112e-06, | |
| "loss": 0.0009, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 10.39, | |
| "learning_rate": 1.3955555555555556e-06, | |
| "loss": 0.0009, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 10.45, | |
| "learning_rate": 1.34e-06, | |
| "loss": 0.0018, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 10.51, | |
| "learning_rate": 1.2844444444444445e-06, | |
| "loss": 0.001, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "learning_rate": 1.228888888888889e-06, | |
| "loss": 0.0009, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "learning_rate": 1.1733333333333335e-06, | |
| "loss": 0.0009, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 1.117777777777778e-06, | |
| "loss": 0.0009, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "learning_rate": 1.0622222222222222e-06, | |
| "loss": 0.0009, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 1.0066666666666668e-06, | |
| "loss": 0.0009, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 0.0009, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 10.93, | |
| "learning_rate": 8.955555555555557e-07, | |
| "loss": 0.0009, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 10.99, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 0.0009, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 7.844444444444445e-07, | |
| "loss": 0.0009, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 11.1, | |
| "learning_rate": 7.28888888888889e-07, | |
| "loss": 0.0008, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 11.16, | |
| "learning_rate": 6.733333333333334e-07, | |
| "loss": 0.0008, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 11.22, | |
| "learning_rate": 6.177777777777778e-07, | |
| "loss": 0.0009, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 5.622222222222223e-07, | |
| "loss": 0.0008, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 5.066666666666667e-07, | |
| "loss": 0.0008, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 4.511111111111111e-07, | |
| "loss": 0.0008, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 11.46, | |
| "learning_rate": 3.9555555555555557e-07, | |
| "loss": 0.0008, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 11.52, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 0.0009, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 11.58, | |
| "learning_rate": 2.844444444444445e-07, | |
| "loss": 0.0009, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 11.64, | |
| "learning_rate": 2.2888888888888892e-07, | |
| "loss": 0.0009, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 1.7333333333333335e-07, | |
| "loss": 0.0008, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 11.76, | |
| "learning_rate": 1.1777777777777778e-07, | |
| "loss": 0.0009, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 11.82, | |
| "learning_rate": 6.222222222222223e-08, | |
| "loss": 0.0009, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 6.666666666666667e-09, | |
| "loss": 0.0008, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "eval_loss": 0.0008248339290730655, | |
| "eval_runtime": 1392.7651, | |
| "eval_samples_per_second": 3.621, | |
| "eval_steps_per_second": 0.604, | |
| "eval_wer": 0.1995279982836301, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "step": 5000, | |
| "total_flos": 1.728655406456832e+19, | |
| "train_loss": 0.08451520547643304, | |
| "train_runtime": 17827.4181, | |
| "train_samples_per_second": 3.366, | |
| "train_steps_per_second": 0.28 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 12, | |
| "total_flos": 1.728655406456832e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |