| { | |
| "best_metric": 23.40004600874166, | |
| "best_model_checkpoint": "./output/small/cv9-special-batch4-base/checkpoint-5000", | |
| "epoch": 2.420135527589545, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.4e-07, | |
| "loss": 1.6477, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.4589, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.4109, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.2925, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 1.055, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.9288, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.8256, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.8196, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.7644, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.7946, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.7244, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.658, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.6782, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.6831, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.6938, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.6656, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.6021, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.598, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.5946, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.5774, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.6129, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.654, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.842222222222223e-06, | |
| "loss": 0.654, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.786666666666667e-06, | |
| "loss": 0.5921, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.733333333333334e-06, | |
| "loss": 0.6112, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.677777777777778e-06, | |
| "loss": 0.525, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.622222222222222e-06, | |
| "loss": 0.5711, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 9.566666666666668e-06, | |
| "loss": 0.5876, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.511111111111112e-06, | |
| "loss": 0.6233, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.455555555555557e-06, | |
| "loss": 0.6066, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.5517, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.344444444444446e-06, | |
| "loss": 0.6004, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.28888888888889e-06, | |
| "loss": 0.5741, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.233333333333334e-06, | |
| "loss": 0.5758, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.17777777777778e-06, | |
| "loss": 0.5497, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.122222222222223e-06, | |
| "loss": 0.5894, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.066666666666667e-06, | |
| "loss": 0.5201, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.011111111111111e-06, | |
| "loss": 0.4977, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.955555555555555e-06, | |
| "loss": 0.5727, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.5013, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_loss": 0.4523269236087799, | |
| "eval_runtime": 378.1435, | |
| "eval_samples_per_second": 9.578, | |
| "eval_steps_per_second": 4.789, | |
| "eval_wer": 28.599033816425123, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.844444444444445e-06, | |
| "loss": 0.594, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.788888888888891e-06, | |
| "loss": 0.5337, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.733333333333333e-06, | |
| "loss": 0.4364, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.677777777777779e-06, | |
| "loss": 0.5458, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.622222222222223e-06, | |
| "loss": 0.5614, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.566666666666667e-06, | |
| "loss": 0.5171, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.511111111111113e-06, | |
| "loss": 0.6269, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.455555555555555e-06, | |
| "loss": 0.521, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.4816, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.344444444444445e-06, | |
| "loss": 0.5121, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 8.288888888888889e-06, | |
| "loss": 0.4647, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.233333333333335e-06, | |
| "loss": 0.4602, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.177777777777779e-06, | |
| "loss": 0.5051, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.122222222222223e-06, | |
| "loss": 0.5201, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.066666666666667e-06, | |
| "loss": 0.543, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.011111111111113e-06, | |
| "loss": 0.5342, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.955555555555557e-06, | |
| "loss": 0.5787, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.5514, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.844444444444446e-06, | |
| "loss": 0.4387, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.788888888888889e-06, | |
| "loss": 0.5233, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.733333333333334e-06, | |
| "loss": 0.5037, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.677777777777778e-06, | |
| "loss": 0.534, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.622222222222223e-06, | |
| "loss": 0.4463, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.566666666666667e-06, | |
| "loss": 0.4531, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 7.511111111111111e-06, | |
| "loss": 0.5009, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 7.455555555555556e-06, | |
| "loss": 0.4674, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.4772, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.344444444444445e-06, | |
| "loss": 0.5021, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.28888888888889e-06, | |
| "loss": 0.4403, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.233333333333334e-06, | |
| "loss": 0.467, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 7.177777777777778e-06, | |
| "loss": 0.4857, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 7.122222222222222e-06, | |
| "loss": 0.4724, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 7.066666666666667e-06, | |
| "loss": 0.5039, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 7.011111111111112e-06, | |
| "loss": 0.4806, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.955555555555557e-06, | |
| "loss": 0.442, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.4887, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.844444444444445e-06, | |
| "loss": 0.4449, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.788888888888889e-06, | |
| "loss": 0.5259, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.733333333333334e-06, | |
| "loss": 0.4716, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.677777777777779e-06, | |
| "loss": 0.4145, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "eval_loss": 0.406696617603302, | |
| "eval_runtime": 681.4801, | |
| "eval_samples_per_second": 5.315, | |
| "eval_steps_per_second": 2.657, | |
| "eval_wer": 25.81090407177364, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.6222222222222236e-06, | |
| "loss": 0.5639, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.566666666666667e-06, | |
| "loss": 0.4991, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.511111111111112e-06, | |
| "loss": 0.4578, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 6.455555555555556e-06, | |
| "loss": 0.2633, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.2449, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 6.3444444444444454e-06, | |
| "loss": 0.2683, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 6.28888888888889e-06, | |
| "loss": 0.3193, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 6.2333333333333335e-06, | |
| "loss": 0.3325, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 6.177777777777778e-06, | |
| "loss": 0.2915, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.1222222222222224e-06, | |
| "loss": 0.2436, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 6.066666666666667e-06, | |
| "loss": 0.266, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.011111111111112e-06, | |
| "loss": 0.2699, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.955555555555555e-06, | |
| "loss": 0.3019, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.243, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.844444444444445e-06, | |
| "loss": 0.244, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 5.788888888888889e-06, | |
| "loss": 0.2796, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 5.733333333333334e-06, | |
| "loss": 0.312, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 5.677777777777779e-06, | |
| "loss": 0.2963, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 5.622222222222222e-06, | |
| "loss": 0.2605, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 5.566666666666667e-06, | |
| "loss": 0.2747, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.511111111111112e-06, | |
| "loss": 0.2872, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.455555555555556e-06, | |
| "loss": 0.2951, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.2757, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 5.344444444444446e-06, | |
| "loss": 0.2917, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 5.288888888888889e-06, | |
| "loss": 0.2663, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 5.233333333333334e-06, | |
| "loss": 0.2537, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 5.177777777777779e-06, | |
| "loss": 0.2905, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 5.122222222222223e-06, | |
| "loss": 0.2493, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 5.0666666666666676e-06, | |
| "loss": 0.2665, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 5.011111111111111e-06, | |
| "loss": 0.2604, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.9555555555555565e-06, | |
| "loss": 0.2883, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.2455, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 4.8444444444444446e-06, | |
| "loss": 0.2266, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 4.7888888888888894e-06, | |
| "loss": 0.2811, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 4.7333333333333335e-06, | |
| "loss": 0.2659, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 4.677777777777778e-06, | |
| "loss": 0.2514, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 4.622222222222222e-06, | |
| "loss": 0.2826, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 4.566666666666667e-06, | |
| "loss": 0.2396, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 4.511111111111111e-06, | |
| "loss": 0.2654, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.455555555555555e-06, | |
| "loss": 0.2437, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "eval_loss": 0.3820730447769165, | |
| "eval_runtime": 565.6444, | |
| "eval_samples_per_second": 6.403, | |
| "eval_steps_per_second": 3.202, | |
| "eval_wer": 24.38003220611916, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.2481, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 4.344444444444445e-06, | |
| "loss": 0.3334, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 4.288888888888889e-06, | |
| "loss": 0.2478, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 4.233333333333334e-06, | |
| "loss": 0.2906, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 4.177777777777778e-06, | |
| "loss": 0.2694, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 4.122222222222222e-06, | |
| "loss": 0.2475, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 4.066666666666667e-06, | |
| "loss": 0.2726, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 4.011111111111111e-06, | |
| "loss": 0.2255, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.955555555555556e-06, | |
| "loss": 0.3179, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.2579, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 3.844444444444445e-06, | |
| "loss": 0.3085, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.7888888888888893e-06, | |
| "loss": 0.2959, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.7333333333333337e-06, | |
| "loss": 0.2888, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 3.6777777777777778e-06, | |
| "loss": 0.2779, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.6222222222222226e-06, | |
| "loss": 0.2676, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.566666666666667e-06, | |
| "loss": 0.2689, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 3.511111111111111e-06, | |
| "loss": 0.2797, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.455555555555556e-06, | |
| "loss": 0.2329, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.2284, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.3444444444444445e-06, | |
| "loss": 0.277, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.2888888888888894e-06, | |
| "loss": 0.2117, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.2333333333333334e-06, | |
| "loss": 0.2259, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.177777777777778e-06, | |
| "loss": 0.2746, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 3.1222222222222228e-06, | |
| "loss": 0.2484, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.066666666666667e-06, | |
| "loss": 0.2516, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.0111111111111113e-06, | |
| "loss": 0.2158, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.955555555555556e-06, | |
| "loss": 0.2516, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.2316, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.8444444444444446e-06, | |
| "loss": 0.2505, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.788888888888889e-06, | |
| "loss": 0.2461, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 2.7333333333333336e-06, | |
| "loss": 0.2254, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 2.677777777777778e-06, | |
| "loss": 0.2987, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 2.6222222222222225e-06, | |
| "loss": 0.3086, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.566666666666667e-06, | |
| "loss": 0.2593, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.5111111111111114e-06, | |
| "loss": 0.2399, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 2.455555555555556e-06, | |
| "loss": 0.2339, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.2315, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.3444444444444448e-06, | |
| "loss": 0.2651, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 2.2888888888888892e-06, | |
| "loss": 0.251, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.2333333333333333e-06, | |
| "loss": 0.2566, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "eval_loss": 0.36946794390678406, | |
| "eval_runtime": 554.8411, | |
| "eval_samples_per_second": 6.528, | |
| "eval_steps_per_second": 3.264, | |
| "eval_wer": 23.979756153669197, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.1777777777777777e-06, | |
| "loss": 0.2733, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 2.1222222222222226e-06, | |
| "loss": 0.2981, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 2.0666666666666666e-06, | |
| "loss": 0.2824, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 2.011111111111111e-06, | |
| "loss": 0.2428, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.955555555555556e-06, | |
| "loss": 0.2112, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.1964, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.8444444444444445e-06, | |
| "loss": 0.1408, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.788888888888889e-06, | |
| "loss": 0.1748, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.7333333333333336e-06, | |
| "loss": 0.143, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.6777777777777779e-06, | |
| "loss": 0.1605, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.6222222222222223e-06, | |
| "loss": 0.1626, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.566666666666667e-06, | |
| "loss": 0.1456, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5111111111111112e-06, | |
| "loss": 0.1871, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4555555555555557e-06, | |
| "loss": 0.1666, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.1368, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.3444444444444446e-06, | |
| "loss": 0.1974, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.288888888888889e-06, | |
| "loss": 0.1703, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.2333333333333335e-06, | |
| "loss": 0.176, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.1777777777777778e-06, | |
| "loss": 0.167, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.1222222222222222e-06, | |
| "loss": 0.1454, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.066666666666667e-06, | |
| "loss": 0.1593, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.0111111111111111e-06, | |
| "loss": 0.1554, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 9.555555555555556e-07, | |
| "loss": 0.1632, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.1542, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 8.444444444444445e-07, | |
| "loss": 0.1936, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 7.888888888888889e-07, | |
| "loss": 0.1256, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 7.333333333333334e-07, | |
| "loss": 0.1519, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 6.777777777777779e-07, | |
| "loss": 0.1406, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 6.222222222222223e-07, | |
| "loss": 0.1598, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.666666666666667e-07, | |
| "loss": 0.184, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.111111111111112e-07, | |
| "loss": 0.1584, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.5555555555555563e-07, | |
| "loss": 0.1453, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.1655, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 3.444444444444445e-07, | |
| "loss": 0.1646, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.9111111111111116e-07, | |
| "loss": 0.1422, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.3555555555555556e-07, | |
| "loss": 0.1781, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.8e-07, | |
| "loss": 0.1361, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.2444444444444446e-07, | |
| "loss": 0.1535, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 6.888888888888889e-08, | |
| "loss": 0.1522, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 1.3333333333333334e-08, | |
| "loss": 0.1161, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "eval_loss": 0.36965835094451904, | |
| "eval_runtime": 558.4119, | |
| "eval_samples_per_second": 6.486, | |
| "eval_steps_per_second": 3.243, | |
| "eval_wer": 23.40004600874166, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "step": 5000, | |
| "total_flos": 1.29681076322304e+18, | |
| "train_loss": 0.3893299392700195, | |
| "train_runtime": 4181.8476, | |
| "train_samples_per_second": 4.783, | |
| "train_steps_per_second": 1.196 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.29681076322304e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |