| { | |
| "best_metric": 31.750632620197837, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch8-tiny/checkpoint-4000", | |
| "epoch": 4.84027105517909, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.4e-07, | |
| "loss": 1.9132, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.8774, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.6394, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.517, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 1.356, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 1.1666, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.44e-06, | |
| "loss": 1.1098, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 3.94e-06, | |
| "loss": 1.0946, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.9803, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.9351, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.9698, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.9416, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.8972, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.8773, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.9044, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.8669, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.8283, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.8241, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.771, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.7903, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.8073, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.7234, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.7606, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.818, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.7483, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.6756, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.758, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.7298, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.807, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.6752, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.7351, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.6576, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.6682, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.7063, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.6425, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.6812, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.696, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.664, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.6812, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.6281, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_loss": 0.5817149877548218, | |
| "eval_runtime": 395.8607, | |
| "eval_samples_per_second": 9.15, | |
| "eval_steps_per_second": 2.289, | |
| "eval_wer": 37.69496204278813, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.7129, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.6053, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.52, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.4714, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.4623, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.52, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.5124, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.5038, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.5198, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.5137, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.4533, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.4812, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.4551, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.4878, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.5494, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.4786, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.4574, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.4675, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.4592, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.455, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.4434, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.4754, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.4655, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.4765, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.4637, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.452, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.4463, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.4431, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.4434, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.5051, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.4722, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.4514, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.434, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.4581, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.4511, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.4779, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.4567, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.4881, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.4939, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.4018, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_loss": 0.5157097578048706, | |
| "eval_runtime": 396.6463, | |
| "eval_samples_per_second": 9.132, | |
| "eval_steps_per_second": 2.284, | |
| "eval_wer": 34.21210029905682, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.4604, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.4541, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.4158, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.2983, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.3284, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.3222, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.3315, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.3001, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.3031, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.3073, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.2868, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.3437, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.2915, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.3414, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.3238, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.3229, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.304, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.3254, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.3052, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 5.5644444444444444e-06, | |
| "loss": 0.3131, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.508888888888889e-06, | |
| "loss": 0.3295, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 5.453333333333334e-06, | |
| "loss": 0.3132, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.397777777777778e-06, | |
| "loss": 0.3099, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 5.342222222222223e-06, | |
| "loss": 0.336, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 5.286666666666666e-06, | |
| "loss": 0.3083, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 5.231111111111111e-06, | |
| "loss": 0.3168, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 5.175555555555556e-06, | |
| "loss": 0.2933, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 5.12e-06, | |
| "loss": 0.298, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.064444444444445e-06, | |
| "loss": 0.3228, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.00888888888889e-06, | |
| "loss": 0.292, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.953333333333334e-06, | |
| "loss": 0.3284, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.897777777777778e-06, | |
| "loss": 0.319, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.842222222222223e-06, | |
| "loss": 0.3453, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.786666666666667e-06, | |
| "loss": 0.3213, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.731111111111112e-06, | |
| "loss": 0.3095, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.675555555555556e-06, | |
| "loss": 0.2698, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 4.620000000000001e-06, | |
| "loss": 0.3513, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 4.564444444444445e-06, | |
| "loss": 0.3191, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.50888888888889e-06, | |
| "loss": 0.2976, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 4.453333333333334e-06, | |
| "loss": 0.2914, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "eval_loss": 0.4979938864707947, | |
| "eval_runtime": 396.5928, | |
| "eval_samples_per_second": 9.133, | |
| "eval_steps_per_second": 2.284, | |
| "eval_wer": 32.49597423510467, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.397777777777778e-06, | |
| "loss": 0.2706, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.3422222222222225e-06, | |
| "loss": 0.3295, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.2866666666666666e-06, | |
| "loss": 0.3131, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.2311111111111114e-06, | |
| "loss": 0.3161, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 4.175555555555556e-06, | |
| "loss": 0.241, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 4.12e-06, | |
| "loss": 0.2271, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 4.064444444444444e-06, | |
| "loss": 0.2098, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.008888888888889e-06, | |
| "loss": 0.216, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 3.953333333333333e-06, | |
| "loss": 0.2378, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 3.897777777777778e-06, | |
| "loss": 0.2339, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.842222222222223e-06, | |
| "loss": 0.2331, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.7866666666666667e-06, | |
| "loss": 0.2315, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 3.7311111111111116e-06, | |
| "loss": 0.2328, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.675555555555556e-06, | |
| "loss": 0.2473, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.62e-06, | |
| "loss": 0.2203, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.564444444444445e-06, | |
| "loss": 0.2078, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.508888888888889e-06, | |
| "loss": 0.2314, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 3.4533333333333334e-06, | |
| "loss": 0.2414, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.3977777777777783e-06, | |
| "loss": 0.2341, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.3422222222222224e-06, | |
| "loss": 0.236, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.286666666666667e-06, | |
| "loss": 0.2171, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 3.2311111111111117e-06, | |
| "loss": 0.232, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.1755555555555557e-06, | |
| "loss": 0.232, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.12e-06, | |
| "loss": 0.2626, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 3.064444444444445e-06, | |
| "loss": 0.2207, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.008888888888889e-06, | |
| "loss": 0.2263, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.9533333333333336e-06, | |
| "loss": 0.2068, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 2.8977777777777785e-06, | |
| "loss": 0.2288, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 2.8422222222222225e-06, | |
| "loss": 0.2306, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 2.786666666666667e-06, | |
| "loss": 0.2123, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.7311111111111114e-06, | |
| "loss": 0.2077, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 2.675555555555556e-06, | |
| "loss": 0.2275, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 2.6200000000000003e-06, | |
| "loss": 0.2276, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 2.5644444444444444e-06, | |
| "loss": 0.2176, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.5088888888888892e-06, | |
| "loss": 0.223, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 2.4533333333333333e-06, | |
| "loss": 0.2283, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 2.397777777777778e-06, | |
| "loss": 0.2263, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 2.342222222222222e-06, | |
| "loss": 0.239, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 2.2866666666666667e-06, | |
| "loss": 0.2275, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 2.2311111111111115e-06, | |
| "loss": 0.2078, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "eval_loss": 0.49681562185287476, | |
| "eval_runtime": 400.1016, | |
| "eval_samples_per_second": 9.053, | |
| "eval_steps_per_second": 2.264, | |
| "eval_wer": 31.750632620197837, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.2278, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.2294, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.2142, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.2307, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.2052, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.1793, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.1794, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.1839, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.1691, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.176, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.1944, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.1979, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.1788, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.1828, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.1764, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.1625, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.1626, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.1917, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.196, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.1788, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.1875, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.2018, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.1695, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.1805, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.1805, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.157, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.171, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.1868, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.1767, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.1862, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.1507, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.1826, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.1754, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.1963, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.1856, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.2095, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.1874, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.1855, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.1767, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.1925, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "eval_loss": 0.49863627552986145, | |
| "eval_runtime": 284.9121, | |
| "eval_samples_per_second": 12.713, | |
| "eval_steps_per_second": 3.18, | |
| "eval_wer": 31.874856222682308, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "step": 5000, | |
| "total_flos": 9.8445997817856e+17, | |
| "train_loss": 0.42814777688980105, | |
| "train_runtime": 4071.3775, | |
| "train_samples_per_second": 9.825, | |
| "train_steps_per_second": 1.228 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 5, | |
| "total_flos": 9.8445997817856e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |