| { | |
| "best_metric": 55.89780169898191, | |
| "best_model_checkpoint": "./output/small/yt-special-batch12-base/checkpoint-4000", | |
| "epoch": 1.2883277505797475, | |
| "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": 2.6974, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 2.7075, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.44e-06, | |
| "loss": 2.2727, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.94e-06, | |
| "loss": 2.0922, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 1.7334, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 1.6666, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.44e-06, | |
| "loss": 1.4382, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 3.94e-06, | |
| "loss": 1.5157, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 1.3801, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.94e-06, | |
| "loss": 1.4242, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 1.2648, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 5.94e-06, | |
| "loss": 1.3495, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 1.3184, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 1.382, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 1.2905, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 7.94e-06, | |
| "loss": 1.3657, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 8.44e-06, | |
| "loss": 1.3017, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 8.94e-06, | |
| "loss": 1.2553, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 1.1857, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 1.261, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 1.1911, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 1.1363, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.84e-06, | |
| "loss": 1.1426, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 1.2282, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 1.1897, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 1.1078, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 1.1843, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 1.184, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 1.1953, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 1.1384, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 1.2715, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 1.2485, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 1.1931, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 1.1123, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 1.2028, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 1.1147, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 1.1369, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 1.1365, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 1.0916, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 1.0995, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_loss": 1.1249420642852783, | |
| "eval_runtime": 290.3743, | |
| "eval_samples_per_second": 8.441, | |
| "eval_steps_per_second": 1.409, | |
| "eval_wer": 91.35594319434537, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.84e-06, | |
| "loss": 1.1713, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 1.1435, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 1.0325, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 1.1324, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 1.106, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 1.0426, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 1.0912, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 1.0399, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 1.1191, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.34e-06, | |
| "loss": 1.0492, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 1.1337, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 1.119, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 1.0759, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 1.0667, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 1.1291, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 1.043, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.9725, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 1.1375, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 1.0954, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.9691, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 1.0073, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 1.0875, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 1.1436, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 1.0335, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 1.0381, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 1.0247, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 1.0112, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 1.0773, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 1.0352, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.9993, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 1.0549, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 1.1329, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 1.0539, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 1.0886, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.9541, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 1.0453, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 1.0315, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 1.0545, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 1.117, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.9995, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_loss": 1.0126415491104126, | |
| "eval_runtime": 199.186, | |
| "eval_samples_per_second": 12.305, | |
| "eval_steps_per_second": 2.053, | |
| "eval_wer": 68.1343622333182, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 1.0156, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.9016, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 1.0186, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 1.076, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 1.0318, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 6.34e-06, | |
| "loss": 1.0431, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.9964, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.9557, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.9485, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.9851, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.9713, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 1.0508, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 1.0921, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 1.0123, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.939, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.9636, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.9706, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 1.012, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 1.0342, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.9621, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.9963, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.9884, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.9775, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 1.0338, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.9153, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.948, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.9778, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 1.0181, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.953, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.9678, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.9637, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 1.0114, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.9786, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 1.0951, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.916, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 1.012, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.9977, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 1.0063, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 1.0191, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.9872, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "eval_loss": 0.9619873762130737, | |
| "eval_runtime": 250.345, | |
| "eval_samples_per_second": 9.79, | |
| "eval_steps_per_second": 1.634, | |
| "eval_wer": 65.94254587899617, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.8916, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.34e-06, | |
| "loss": 1.0239, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.9655, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 1.0034, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.9691, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 1.026, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.8627, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 1.0219, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.8905, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.8927, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.9561, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.904, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.8444, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.9512, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.9364, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.9826, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.9423, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.9588, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.9318, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.9167, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.9583, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.9108, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.9248, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.9327, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.9788, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.9412, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.9422, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.7868, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.9272, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.959, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.908, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.8578, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.9653, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.9333, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 1.0096, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.7768, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.7146, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.7899, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.7086, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.7043, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "eval_loss": 0.9330242872238159, | |
| "eval_runtime": 144.1119, | |
| "eval_samples_per_second": 17.008, | |
| "eval_steps_per_second": 2.838, | |
| "eval_wer": 55.89780169898191, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.1755555555555556e-06, | |
| "loss": 0.7524, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.12e-06, | |
| "loss": 0.7031, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 2.064444444444445e-06, | |
| "loss": 0.7542, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2.008888888888889e-06, | |
| "loss": 0.7392, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.9533333333333334e-06, | |
| "loss": 0.7672, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.8977777777777779e-06, | |
| "loss": 0.7571, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.8422222222222225e-06, | |
| "loss": 0.6917, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.7866666666666668e-06, | |
| "loss": 0.7342, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.7311111111111112e-06, | |
| "loss": 0.8147, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.675555555555556e-06, | |
| "loss": 0.7602, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.6200000000000002e-06, | |
| "loss": 0.7838, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.5644444444444446e-06, | |
| "loss": 0.8098, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.5088888888888889e-06, | |
| "loss": 0.8006, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.4533333333333335e-06, | |
| "loss": 0.7597, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.397777777777778e-06, | |
| "loss": 0.757, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.3422222222222222e-06, | |
| "loss": 0.7504, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.286666666666667e-06, | |
| "loss": 0.665, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.2311111111111112e-06, | |
| "loss": 0.7536, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.1755555555555556e-06, | |
| "loss": 0.802, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.12e-06, | |
| "loss": 0.7649, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.0644444444444445e-06, | |
| "loss": 0.7574, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.008888888888889e-06, | |
| "loss": 0.7421, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 9.533333333333335e-07, | |
| "loss": 0.7787, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.977777777777778e-07, | |
| "loss": 0.722, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.422222222222224e-07, | |
| "loss": 0.7026, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.866666666666667e-07, | |
| "loss": 0.845, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 7.311111111111112e-07, | |
| "loss": 0.7323, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 6.755555555555555e-07, | |
| "loss": 0.7269, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 6.200000000000001e-07, | |
| "loss": 0.7942, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 5.644444444444445e-07, | |
| "loss": 0.7383, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 5.088888888888889e-07, | |
| "loss": 0.6216, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.533333333333334e-07, | |
| "loss": 0.7405, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.9777777777777783e-07, | |
| "loss": 0.7197, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.422222222222223e-07, | |
| "loss": 0.7528, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.866666666666667e-07, | |
| "loss": 0.8744, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.3111111111111112e-07, | |
| "loss": 0.7853, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.7555555555555558e-07, | |
| "loss": 0.875, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.2000000000000002e-07, | |
| "loss": 0.8089, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 6.444444444444445e-08, | |
| "loss": 0.7222, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.88888888888889e-09, | |
| "loss": 0.7292, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "eval_loss": 0.9223517179489136, | |
| "eval_runtime": 158.0146, | |
| "eval_samples_per_second": 15.511, | |
| "eval_steps_per_second": 2.588, | |
| "eval_wer": 62.50567408079891, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "step": 5000, | |
| "total_flos": 3.89147004960768e+18, | |
| "train_loss": 1.0220142887115478, | |
| "train_runtime": 4418.3338, | |
| "train_samples_per_second": 13.58, | |
| "train_steps_per_second": 1.132 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 2, | |
| "total_flos": 3.89147004960768e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |