| { | |
| "best_metric": 11.4438961596224, | |
| "best_model_checkpoint": "./output/small/yt-special-batch8-base/checkpoint-5000", | |
| "epoch": 7.9239302694136295, | |
| "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": 1.7491, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.4857, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.1541, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.0995, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 2.2131, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 3.4199, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 4.7849, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 9.5528, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.42e-06, | |
| "loss": 12.497, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 12.4801, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.380000000000001e-06, | |
| "loss": 14.1177, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.86e-06, | |
| "loss": 15.3981, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.360000000000001e-06, | |
| "loss": 18.7705, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 6.860000000000001e-06, | |
| "loss": 23.5459, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.360000000000001e-06, | |
| "loss": 31.1971, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 33.355, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.36e-06, | |
| "loss": 45.9219, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.860000000000002e-06, | |
| "loss": 46.7898, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.360000000000002e-06, | |
| "loss": 63.6532, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.86e-06, | |
| "loss": 59.3715, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 61.0177, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 9.904444444444445e-06, | |
| "loss": 67.4881, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.851111111111112e-06, | |
| "loss": 68.3865, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.795555555555556e-06, | |
| "loss": 79.6374, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.74e-06, | |
| "loss": 64.3938, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.684444444444446e-06, | |
| "loss": 70.0958, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 9.62888888888889e-06, | |
| "loss": 69.7785, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 9.584444444444445e-06, | |
| "loss": 64.8585, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 9.535555555555557e-06, | |
| "loss": 61.1897, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.482222222222223e-06, | |
| "loss": 64.2252, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.426666666666667e-06, | |
| "loss": 61.7694, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 9.371111111111111e-06, | |
| "loss": 60.8694, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.315555555555557e-06, | |
| "loss": 59.7583, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.260000000000001e-06, | |
| "loss": 55.3815, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 9.204444444444445e-06, | |
| "loss": 51.8172, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 9.14888888888889e-06, | |
| "loss": 48.4673, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.093333333333333e-06, | |
| "loss": 45.7032, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.037777777777779e-06, | |
| "loss": 54.2089, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 8.982222222222223e-06, | |
| "loss": 47.5647, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 8.926666666666669e-06, | |
| "loss": 41.113, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "eval_loss": 42.97588348388672, | |
| "eval_runtime": 1037.3254, | |
| "eval_samples_per_second": 4.862, | |
| "eval_steps_per_second": 1.216, | |
| "eval_wer": 107.56275477365371, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 8.871111111111111e-06, | |
| "loss": 47.9835, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.815555555555557e-06, | |
| "loss": 42.2165, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.76e-06, | |
| "loss": 43.1054, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.704444444444445e-06, | |
| "loss": 40.2444, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.64888888888889e-06, | |
| "loss": 38.3884, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 8.593333333333333e-06, | |
| "loss": 41.0448, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 8.537777777777779e-06, | |
| "loss": 37.6895, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.482222222222223e-06, | |
| "loss": 39.4216, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.426666666666667e-06, | |
| "loss": 43.0472, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.371111111111112e-06, | |
| "loss": 37.6302, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.315555555555557e-06, | |
| "loss": 33.0809, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 8.26e-06, | |
| "loss": 38.3013, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 8.204444444444445e-06, | |
| "loss": 32.9067, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 8.14888888888889e-06, | |
| "loss": 34.4728, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.093333333333334e-06, | |
| "loss": 34.53, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 8.037777777777778e-06, | |
| "loss": 33.0867, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.982222222222224e-06, | |
| "loss": 30.1566, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 7.926666666666666e-06, | |
| "loss": 32.9418, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.871111111111112e-06, | |
| "loss": 31.0219, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 7.815555555555556e-06, | |
| "loss": 29.2539, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.76e-06, | |
| "loss": 30.4172, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 7.704444444444446e-06, | |
| "loss": 30.1107, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 7.648888888888888e-06, | |
| "loss": 30.2495, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.593333333333334e-06, | |
| "loss": 28.5465, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.537777777777778e-06, | |
| "loss": 29.2619, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.482222222222223e-06, | |
| "loss": 29.9081, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 7.426666666666668e-06, | |
| "loss": 27.4152, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 7.371111111111112e-06, | |
| "loss": 31.0304, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 7.315555555555556e-06, | |
| "loss": 28.8473, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 7.260000000000001e-06, | |
| "loss": 24.3152, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 7.204444444444445e-06, | |
| "loss": 30.2247, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.14888888888889e-06, | |
| "loss": 26.9633, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.093333333333335e-06, | |
| "loss": 24.6396, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 7.037777777777778e-06, | |
| "loss": 30.4137, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 6.982222222222223e-06, | |
| "loss": 26.6009, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.926666666666667e-06, | |
| "loss": 24.5352, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.871111111111112e-06, | |
| "loss": 23.6272, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 6.8155555555555565e-06, | |
| "loss": 21.0893, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.760000000000001e-06, | |
| "loss": 21.3604, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 6.7044444444444445e-06, | |
| "loss": 17.3442, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "eval_loss": 18.70366668701172, | |
| "eval_runtime": 1280.3674, | |
| "eval_samples_per_second": 3.939, | |
| "eval_steps_per_second": 0.985, | |
| "eval_wer": 144.10641493241795, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 6.648888888888889e-06, | |
| "loss": 21.8499, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 6.5933333333333335e-06, | |
| "loss": 20.0117, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 6.537777777777778e-06, | |
| "loss": 21.6871, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 6.482222222222223e-06, | |
| "loss": 18.9966, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.426666666666668e-06, | |
| "loss": 17.4701, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 6.371111111111111e-06, | |
| "loss": 22.0492, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 6.317777777777778e-06, | |
| "loss": 16.6105, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 6.262222222222223e-06, | |
| "loss": 17.567, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 6.211111111111111e-06, | |
| "loss": 17.675, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 6.155555555555556e-06, | |
| "loss": 20.1354, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 6.1e-06, | |
| "loss": 18.9197, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 6.044444444444445e-06, | |
| "loss": 17.748, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 5.98888888888889e-06, | |
| "loss": 19.3784, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 5.933333333333335e-06, | |
| "loss": 18.7225, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 5.877777777777778e-06, | |
| "loss": 18.112, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 5.822222222222223e-06, | |
| "loss": 16.86, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 5.766666666666667e-06, | |
| "loss": 20.3351, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 5.711111111111112e-06, | |
| "loss": 17.9965, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 5.6555555555555566e-06, | |
| "loss": 19.2315, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 16.2915, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.544444444444445e-06, | |
| "loss": 15.6132, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 5.4888888888888895e-06, | |
| "loss": 12.7693, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.4333333333333335e-06, | |
| "loss": 12.2693, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 5.3777777777777784e-06, | |
| "loss": 10.6931, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 5.322222222222223e-06, | |
| "loss": 12.7438, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 5.2666666666666665e-06, | |
| "loss": 11.1154, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 5.211111111111111e-06, | |
| "loss": 12.2588, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 5.155555555555556e-06, | |
| "loss": 11.8185, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 5.1e-06, | |
| "loss": 11.9872, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 5.044444444444445e-06, | |
| "loss": 10.9318, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 4.988888888888889e-06, | |
| "loss": 10.5515, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 4.933333333333334e-06, | |
| "loss": 10.555, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 12.2332, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 4.824444444444445e-06, | |
| "loss": 11.013, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 4.768888888888889e-06, | |
| "loss": 10.5057, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 4.713333333333334e-06, | |
| "loss": 11.5147, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 4.6577777777777785e-06, | |
| "loss": 10.1689, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 4.602222222222223e-06, | |
| "loss": 10.7969, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 4.546666666666667e-06, | |
| "loss": 10.6932, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 4.4911111111111115e-06, | |
| "loss": 10.8061, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "eval_loss": 7.153076648712158, | |
| "eval_runtime": 931.4057, | |
| "eval_samples_per_second": 5.414, | |
| "eval_steps_per_second": 1.354, | |
| "eval_wer": 52.55095473074447, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 4.4355555555555555e-06, | |
| "loss": 10.9146, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 4.38e-06, | |
| "loss": 9.9721, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 4.324444444444445e-06, | |
| "loss": 11.8253, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 4.268888888888889e-06, | |
| "loss": 9.9704, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 4.213333333333333e-06, | |
| "loss": 10.1412, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 4.157777777777778e-06, | |
| "loss": 9.0687, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 4.102222222222222e-06, | |
| "loss": 6.6716, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 4.046666666666667e-06, | |
| "loss": 7.9673, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 3.991111111111112e-06, | |
| "loss": 5.9019, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.935555555555556e-06, | |
| "loss": 6.3131, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 3.88e-06, | |
| "loss": 6.243, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 3.824444444444444e-06, | |
| "loss": 6.7379, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 3.768888888888889e-06, | |
| "loss": 6.1734, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.713333333333334e-06, | |
| "loss": 5.9483, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 3.657777777777778e-06, | |
| "loss": 5.7135, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 3.6022222222222224e-06, | |
| "loss": 6.0274, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 3.5466666666666673e-06, | |
| "loss": 5.4355, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 3.4911111111111113e-06, | |
| "loss": 6.2385, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 3.435555555555556e-06, | |
| "loss": 5.4162, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 3.3800000000000007e-06, | |
| "loss": 5.6809, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 3.3244444444444447e-06, | |
| "loss": 6.5519, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 3.268888888888889e-06, | |
| "loss": 6.0984, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 3.213333333333334e-06, | |
| "loss": 5.8196, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 3.157777777777778e-06, | |
| "loss": 6.6902, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 3.1022222222222225e-06, | |
| "loss": 5.5926, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.0466666666666666e-06, | |
| "loss": 5.16, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 2.9911111111111115e-06, | |
| "loss": 5.036, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 2.935555555555556e-06, | |
| "loss": 5.5856, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 2.88e-06, | |
| "loss": 4.9832, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 2.824444444444445e-06, | |
| "loss": 5.1377, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 2.7688888888888893e-06, | |
| "loss": 4.9745, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.7133333333333333e-06, | |
| "loss": 4.0952, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 2.6577777777777782e-06, | |
| "loss": 2.9451, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2.6022222222222227e-06, | |
| "loss": 2.7328, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 2.5466666666666667e-06, | |
| "loss": 3.0824, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 2.491111111111111e-06, | |
| "loss": 2.7726, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 2.4355555555555556e-06, | |
| "loss": 3.2206, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 2.38e-06, | |
| "loss": 3.4037, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 2.3244444444444445e-06, | |
| "loss": 3.4738, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 2.268888888888889e-06, | |
| "loss": 3.3269, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "eval_loss": 3.1034955978393555, | |
| "eval_runtime": 930.155, | |
| "eval_samples_per_second": 5.422, | |
| "eval_steps_per_second": 1.356, | |
| "eval_wer": 47.058571122076806, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 2.2133333333333335e-06, | |
| "loss": 3.2691, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 2.157777777777778e-06, | |
| "loss": 3.1905, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 2.1022222222222224e-06, | |
| "loss": 3.0545, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 2.046666666666667e-06, | |
| "loss": 2.8343, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 1.9911111111111113e-06, | |
| "loss": 3.0692, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 1.9355555555555558e-06, | |
| "loss": 2.4305, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 1.8800000000000002e-06, | |
| "loss": 2.9358, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 1.8244444444444445e-06, | |
| "loss": 3.1364, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 1.7688888888888891e-06, | |
| "loss": 2.2167, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 1.7133333333333336e-06, | |
| "loss": 2.2972, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 1.6577777777777778e-06, | |
| "loss": 2.4747, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.6022222222222223e-06, | |
| "loss": 1.9043, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 1.546666666666667e-06, | |
| "loss": 2.297, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 1.4911111111111112e-06, | |
| "loss": 2.6366, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 1.4355555555555557e-06, | |
| "loss": 2.4637, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 3.0088, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.3244444444444446e-06, | |
| "loss": 1.8821, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 1.268888888888889e-06, | |
| "loss": 1.2078, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 1.2133333333333335e-06, | |
| "loss": 1.1169, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 1.1577777777777778e-06, | |
| "loss": 1.0951, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 1.1022222222222222e-06, | |
| "loss": 1.2286, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 1.0466666666666669e-06, | |
| "loss": 1.3975, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 9.911111111111111e-07, | |
| "loss": 1.5158, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 9.355555555555557e-07, | |
| "loss": 1.1447, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 8.8e-07, | |
| "loss": 0.9339, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 8.244444444444445e-07, | |
| "loss": 0.9653, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 7.688888888888891e-07, | |
| "loss": 0.9988, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 7.45, | |
| "learning_rate": 7.133333333333334e-07, | |
| "loss": 0.935, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 6.577777777777779e-07, | |
| "loss": 1.0832, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 6.022222222222223e-07, | |
| "loss": 1.079, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 5.466666666666667e-07, | |
| "loss": 0.8524, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 4.911111111111112e-07, | |
| "loss": 0.8658, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 4.355555555555556e-07, | |
| "loss": 0.8488, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 3.8e-07, | |
| "loss": 0.8372, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.73, | |
| "learning_rate": 3.2444444444444447e-07, | |
| "loss": 0.7349, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 2.6888888888888893e-07, | |
| "loss": 0.9287, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 2.1333333333333334e-07, | |
| "loss": 0.6612, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 1.577777777777778e-07, | |
| "loss": 0.7809, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 1.0222222222222224e-07, | |
| "loss": 0.794, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 4.6666666666666674e-08, | |
| "loss": 0.7405, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "eval_loss": 0.41551142930984497, | |
| "eval_runtime": 890.3201, | |
| "eval_samples_per_second": 5.664, | |
| "eval_steps_per_second": 1.416, | |
| "eval_wer": 11.4438961596224, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "step": 5000, | |
| "total_flos": 2.5921297465344e+18, | |
| "train_loss": 18.966414898300172, | |
| "train_runtime": 10244.3749, | |
| "train_samples_per_second": 3.905, | |
| "train_steps_per_second": 0.488 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 8, | |
| "total_flos": 2.5921297465344e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |