| { | |
| "best_metric": 28.23428448830723, | |
| "best_model_checkpoint": "./output/small/yt-special-batch4/checkpoint-5000", | |
| "epoch": 3.965107057890563, | |
| "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": 1.5065, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.1712, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.3501, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.94e-06, | |
| "loss": 2.3945, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.42e-06, | |
| "loss": 4.8304, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.92e-06, | |
| "loss": 5.0208, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 8.8304, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 13.7868, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.4e-06, | |
| "loss": 17.6539, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 26.5598, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 5.36e-06, | |
| "loss": 45.766, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 5.86e-06, | |
| "loss": 35.7199, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 6.360000000000001e-06, | |
| "loss": 54.9605, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 6.860000000000001e-06, | |
| "loss": 55.4295, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.360000000000001e-06, | |
| "loss": 65.0878, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 69.5386, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.36e-06, | |
| "loss": 73.0867, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.84e-06, | |
| "loss": 94.3113, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 87.2808, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.84e-06, | |
| "loss": 98.8404, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.962222222222224e-06, | |
| "loss": 112.2076, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.906666666666668e-06, | |
| "loss": 103.5807, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.851111111111112e-06, | |
| "loss": 120.3094, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.795555555555556e-06, | |
| "loss": 123.5213, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.74e-06, | |
| "loss": 139.2144, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.684444444444446e-06, | |
| "loss": 115.0782, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 9.62888888888889e-06, | |
| "loss": 130.8006, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.573333333333334e-06, | |
| "loss": 137.604, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 9.517777777777778e-06, | |
| "loss": 145.0352, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 9.466666666666667e-06, | |
| "loss": 126.0557, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.415555555555556e-06, | |
| "loss": 105.6157, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.360000000000002e-06, | |
| "loss": 151.1266, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.304444444444444e-06, | |
| "loss": 142.0032, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 9.24888888888889e-06, | |
| "loss": 139.6682, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.193333333333334e-06, | |
| "loss": 140.1143, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.137777777777778e-06, | |
| "loss": 139.1344, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.082222222222224e-06, | |
| "loss": 133.7512, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.026666666666666e-06, | |
| "loss": 151.7043, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.971111111111112e-06, | |
| "loss": 135.3954, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.915555555555556e-06, | |
| "loss": 125.5137, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "eval_loss": 129.90090942382812, | |
| "eval_runtime": 2724.157, | |
| "eval_samples_per_second": 1.851, | |
| "eval_steps_per_second": 0.926, | |
| "eval_wer": 149.0002145462347, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.860000000000002e-06, | |
| "loss": 145.6737, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.804444444444446e-06, | |
| "loss": 144.7789, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.74888888888889e-06, | |
| "loss": 131.0249, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.693333333333334e-06, | |
| "loss": 143.5189, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.637777777777778e-06, | |
| "loss": 124.5124, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 8.582222222222223e-06, | |
| "loss": 119.1228, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.526666666666667e-06, | |
| "loss": 124.7488, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.471111111111112e-06, | |
| "loss": 122.202, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.415555555555556e-06, | |
| "loss": 116.7528, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.36e-06, | |
| "loss": 131.7794, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.304444444444445e-06, | |
| "loss": 124.1896, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.24888888888889e-06, | |
| "loss": 112.4703, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.193333333333335e-06, | |
| "loss": 114.8219, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.137777777777779e-06, | |
| "loss": 101.7731, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.082222222222223e-06, | |
| "loss": 98.6509, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.026666666666667e-06, | |
| "loss": 104.1523, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.971111111111111e-06, | |
| "loss": 92.43, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.915555555555557e-06, | |
| "loss": 94.9851, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 95.6963, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.804444444444445e-06, | |
| "loss": 87.4435, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.748888888888889e-06, | |
| "loss": 89.9063, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 7.693333333333333e-06, | |
| "loss": 94.5089, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.637777777777779e-06, | |
| "loss": 90.6531, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.582222222222223e-06, | |
| "loss": 78.8492, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.526666666666668e-06, | |
| "loss": 83.7677, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.471111111111111e-06, | |
| "loss": 81.6059, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 7.415555555555556e-06, | |
| "loss": 80.4736, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.360000000000001e-06, | |
| "loss": 85.1648, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.304444444444445e-06, | |
| "loss": 83.8457, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 7.24888888888889e-06, | |
| "loss": 85.8395, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.1933333333333345e-06, | |
| "loss": 78.4769, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.137777777777778e-06, | |
| "loss": 78.3302, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.0822222222222226e-06, | |
| "loss": 90.1422, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.0266666666666674e-06, | |
| "loss": 77.1199, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 6.9711111111111115e-06, | |
| "loss": 64.2029, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.915555555555556e-06, | |
| "loss": 71.5758, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.860000000000001e-06, | |
| "loss": 70.353, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 6.8044444444444444e-06, | |
| "loss": 71.4797, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.748888888888889e-06, | |
| "loss": 70.792, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.693333333333334e-06, | |
| "loss": 67.2464, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "eval_loss": 59.317176818847656, | |
| "eval_runtime": 3546.7342, | |
| "eval_samples_per_second": 1.422, | |
| "eval_steps_per_second": 0.711, | |
| "eval_wer": 298.8671958807123, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.637777777777778e-06, | |
| "loss": 66.5926, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.582222222222223e-06, | |
| "loss": 63.0575, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 6.526666666666666e-06, | |
| "loss": 64.5554, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.471111111111111e-06, | |
| "loss": 69.3081, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 6.415555555555556e-06, | |
| "loss": 59.7389, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 6.360000000000001e-06, | |
| "loss": 62.1109, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 6.304444444444445e-06, | |
| "loss": 70.6619, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 6.24888888888889e-06, | |
| "loss": 57.8578, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 6.193333333333333e-06, | |
| "loss": 62.461, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 6.137777777777778e-06, | |
| "loss": 57.0994, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 6.082222222222223e-06, | |
| "loss": 62.7162, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.026666666666668e-06, | |
| "loss": 54.8485, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 5.971111111111112e-06, | |
| "loss": 56.2333, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 5.915555555555556e-06, | |
| "loss": 56.7746, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 5.86e-06, | |
| "loss": 56.3926, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 5.804444444444445e-06, | |
| "loss": 55.1382, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 5.7488888888888896e-06, | |
| "loss": 57.5388, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 5.6933333333333344e-06, | |
| "loss": 49.7817, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 5.6377777777777785e-06, | |
| "loss": 51.4976, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 5.5822222222222225e-06, | |
| "loss": 59.6299, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.5266666666666666e-06, | |
| "loss": 55.4231, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 5.4711111111111114e-06, | |
| "loss": 47.1009, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 5.415555555555556e-06, | |
| "loss": 51.1057, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 5.36e-06, | |
| "loss": 41.7412, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.304444444444445e-06, | |
| "loss": 45.0973, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.248888888888889e-06, | |
| "loss": 43.8409, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.193333333333333e-06, | |
| "loss": 40.3858, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.137777777777778e-06, | |
| "loss": 41.3208, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.082222222222223e-06, | |
| "loss": 40.5827, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.026666666666667e-06, | |
| "loss": 34.6682, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 4.971111111111111e-06, | |
| "loss": 36.5551, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 4.915555555555556e-06, | |
| "loss": 35.1971, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 4.86e-06, | |
| "loss": 41.1468, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.804444444444445e-06, | |
| "loss": 37.0954, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.74888888888889e-06, | |
| "loss": 36.1367, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.693333333333334e-06, | |
| "loss": 37.777, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.637777777777778e-06, | |
| "loss": 36.123, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.582222222222223e-06, | |
| "loss": 38.9094, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.526666666666667e-06, | |
| "loss": 34.1404, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.471111111111112e-06, | |
| "loss": 34.7799, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "eval_loss": 28.52435302734375, | |
| "eval_runtime": 2528.7422, | |
| "eval_samples_per_second": 1.994, | |
| "eval_steps_per_second": 0.997, | |
| "eval_wer": 134.92598154902382, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.415555555555556e-06, | |
| "loss": 30.6602, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 4.360000000000001e-06, | |
| "loss": 30.5431, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 4.304444444444445e-06, | |
| "loss": 31.4769, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 4.248888888888889e-06, | |
| "loss": 31.7108, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 4.1933333333333336e-06, | |
| "loss": 32.7686, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.1377777777777784e-06, | |
| "loss": 30.9958, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.0822222222222225e-06, | |
| "loss": 31.8733, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 4.026666666666667e-06, | |
| "loss": 30.6928, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.971111111111111e-06, | |
| "loss": 32.7095, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 3.9155555555555554e-06, | |
| "loss": 29.5839, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 3.86e-06, | |
| "loss": 31.0112, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 3.8044444444444443e-06, | |
| "loss": 28.1384, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.7488888888888892e-06, | |
| "loss": 29.8798, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 3.6933333333333337e-06, | |
| "loss": 26.3054, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 3.6377777777777777e-06, | |
| "loss": 32.3702, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 3.5822222222222226e-06, | |
| "loss": 26.3976, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 3.526666666666667e-06, | |
| "loss": 28.478, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 3.471111111111111e-06, | |
| "loss": 29.9708, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.415555555555556e-06, | |
| "loss": 28.754, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.3600000000000004e-06, | |
| "loss": 26.5715, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.3044444444444445e-06, | |
| "loss": 25.6172, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.2488888888888894e-06, | |
| "loss": 25.3106, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 3.193333333333334e-06, | |
| "loss": 23.8701, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 3.137777777777778e-06, | |
| "loss": 24.291, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 3.0822222222222227e-06, | |
| "loss": 21.8997, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.0266666666666668e-06, | |
| "loss": 25.2987, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.9711111111111112e-06, | |
| "loss": 23.5505, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 2.915555555555556e-06, | |
| "loss": 25.5157, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 2.86e-06, | |
| "loss": 21.3761, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.8044444444444446e-06, | |
| "loss": 18.6819, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.748888888888889e-06, | |
| "loss": 23.4894, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 2.6933333333333335e-06, | |
| "loss": 18.3305, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 2.637777777777778e-06, | |
| "loss": 14.2677, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 2.5822222222222224e-06, | |
| "loss": 16.5562, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 2.526666666666667e-06, | |
| "loss": 13.7633, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 2.4711111111111114e-06, | |
| "loss": 17.7606, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 2.415555555555556e-06, | |
| "loss": 12.0998, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 2.3600000000000003e-06, | |
| "loss": 13.1156, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 2.3044444444444447e-06, | |
| "loss": 15.2432, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 2.248888888888889e-06, | |
| "loss": 13.5007, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "eval_loss": 12.516227722167969, | |
| "eval_runtime": 2000.0344, | |
| "eval_samples_per_second": 2.521, | |
| "eval_steps_per_second": 1.261, | |
| "eval_wer": 51.14567689337052, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 2.1933333333333332e-06, | |
| "loss": 15.0485, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 2.137777777777778e-06, | |
| "loss": 10.7393, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 2.0822222222222226e-06, | |
| "loss": 11.9499, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 2.0266666666666666e-06, | |
| "loss": 13.4629, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 1.971111111111111e-06, | |
| "loss": 13.1648, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.915555555555556e-06, | |
| "loss": 14.3279, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.8600000000000002e-06, | |
| "loss": 11.6005, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.8044444444444444e-06, | |
| "loss": 10.6341, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.7488888888888891e-06, | |
| "loss": 15.247, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.6933333333333336e-06, | |
| "loss": 10.8269, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.6377777777777778e-06, | |
| "loss": 13.0179, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.5822222222222223e-06, | |
| "loss": 9.0973, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.526666666666667e-06, | |
| "loss": 11.6001, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.4711111111111112e-06, | |
| "loss": 10.1239, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.4155555555555556e-06, | |
| "loss": 11.6063, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.3600000000000001e-06, | |
| "loss": 10.262, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 1.3044444444444446e-06, | |
| "loss": 12.105, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 1.248888888888889e-06, | |
| "loss": 9.206, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.1933333333333335e-06, | |
| "loss": 9.362, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 1.137777777777778e-06, | |
| "loss": 8.732, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 1.0822222222222222e-06, | |
| "loss": 8.9712, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 1.0266666666666669e-06, | |
| "loss": 9.5471, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 9.711111111111111e-07, | |
| "loss": 8.8118, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 9.155555555555557e-07, | |
| "loss": 7.7686, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 8.6e-07, | |
| "loss": 9.4581, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 8.044444444444445e-07, | |
| "loss": 7.6693, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 7.48888888888889e-07, | |
| "loss": 7.7522, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 6.933333333333334e-07, | |
| "loss": 7.9061, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 6.377777777777779e-07, | |
| "loss": 6.7074, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 5.822222222222223e-07, | |
| "loss": 10.0562, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 5.266666666666667e-07, | |
| "loss": 8.842, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 4.7111111111111113e-07, | |
| "loss": 6.8116, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 4.155555555555556e-07, | |
| "loss": 6.9662, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 3.6e-07, | |
| "loss": 7.8678, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 3.0444444444444445e-07, | |
| "loss": 6.2642, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 2.488888888888889e-07, | |
| "loss": 5.2553, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.9333333333333337e-07, | |
| "loss": 6.6561, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.3777777777777778e-07, | |
| "loss": 7.6037, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 8.222222222222223e-08, | |
| "loss": 6.5388, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 2.6666666666666667e-08, | |
| "loss": 7.3781, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "eval_loss": 3.6843667030334473, | |
| "eval_runtime": 1981.4795, | |
| "eval_samples_per_second": 2.545, | |
| "eval_steps_per_second": 1.273, | |
| "eval_wer": 28.23428448830723, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "step": 5000, | |
| "total_flos": 5.77084225019904e+18, | |
| "train_loss": 53.45060002822876, | |
| "train_runtime": 17302.3407, | |
| "train_samples_per_second": 1.156, | |
| "train_steps_per_second": 0.289 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 4, | |
| "total_flos": 5.77084225019904e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |