| { | |
| "best_metric": 5006.62100456621, | |
| "best_model_checkpoint": "./20230912-output\\checkpoint-2000", | |
| "epoch": 25.17042475091767, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.8e-07, | |
| "loss": 18.9312, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.8e-07, | |
| "loss": 17.0566, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 13.4189, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.8800000000000002e-06, | |
| "loss": 10.4803, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 2.38e-06, | |
| "loss": 9.3304, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.88e-06, | |
| "loss": 8.3203, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.3600000000000004e-06, | |
| "loss": 8.9822, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.86e-06, | |
| "loss": 10.7581, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 4.360000000000001e-06, | |
| "loss": 13.2164, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 4.86e-06, | |
| "loss": 12.7052, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.36e-06, | |
| "loss": 10.3904, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 5.86e-06, | |
| "loss": 8.2617, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 6.360000000000001e-06, | |
| "loss": 6.8613, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 6.860000000000001e-06, | |
| "loss": 6.1093, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 7.360000000000001e-06, | |
| "loss": 5.8769, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 5.6272, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.36e-06, | |
| "loss": 5.674, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 8.860000000000002e-06, | |
| "loss": 5.5667, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 9.360000000000002e-06, | |
| "loss": 5.4705, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 9.86e-06, | |
| "loss": 5.1703, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 9.967272727272728e-06, | |
| "loss": 5.1044, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 9.921818181818183e-06, | |
| "loss": 4.9244, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 9.876363636363637e-06, | |
| "loss": 4.8602, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 9.830909090909092e-06, | |
| "loss": 4.6437, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 9.785454545454546e-06, | |
| "loss": 4.0014, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 9.74e-06, | |
| "loss": 3.8653, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 9.694545454545455e-06, | |
| "loss": 3.8299, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 9.64909090909091e-06, | |
| "loss": 3.6928, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 9.603636363636365e-06, | |
| "loss": 3.1672, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 9.558181818181818e-06, | |
| "loss": 2.5309, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 9.512727272727274e-06, | |
| "loss": 2.4801, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 9.467272727272727e-06, | |
| "loss": 2.2236, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 9.421818181818183e-06, | |
| "loss": 2.3091, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 7.13, | |
| "learning_rate": 9.376363636363637e-06, | |
| "loss": 1.9126, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 9.330909090909092e-06, | |
| "loss": 1.4071, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 9.285454545454546e-06, | |
| "loss": 1.5367, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 7.76, | |
| "learning_rate": 9.240000000000001e-06, | |
| "loss": 1.3686, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 9.194545454545455e-06, | |
| "loss": 1.3177, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 9.14909090909091e-06, | |
| "loss": 0.9167, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 8.39, | |
| "learning_rate": 9.103636363636364e-06, | |
| "loss": 0.9502, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 8.39, | |
| "eval_loss": 3.2547426223754883, | |
| "eval_runtime": 507.4588, | |
| "eval_samples_per_second": 1.255, | |
| "eval_steps_per_second": 0.158, | |
| "eval_wer": 5215.41095890411, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 9.05818181818182e-06, | |
| "loss": 0.774, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 9.012727272727273e-06, | |
| "loss": 0.8792, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 8.967272727272727e-06, | |
| "loss": 0.8182, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 8.921818181818183e-06, | |
| "loss": 0.5773, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 8.876363636363638e-06, | |
| "loss": 0.5618, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 8.830909090909092e-06, | |
| "loss": 0.5499, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 8.785454545454546e-06, | |
| "loss": 0.5145, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 10.07, | |
| "learning_rate": 8.740000000000001e-06, | |
| "loss": 0.4436, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 8.694545454545455e-06, | |
| "loss": 0.3293, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 10.49, | |
| "learning_rate": 8.64909090909091e-06, | |
| "loss": 0.3603, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 10.7, | |
| "learning_rate": 8.603636363636364e-06, | |
| "loss": 0.3796, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "learning_rate": 8.55818181818182e-06, | |
| "loss": 0.347, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 8.512727272727273e-06, | |
| "loss": 0.263, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 11.33, | |
| "learning_rate": 8.467272727272729e-06, | |
| "loss": 0.272, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 11.54, | |
| "learning_rate": 8.421818181818182e-06, | |
| "loss": 0.2657, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 11.75, | |
| "learning_rate": 8.376363636363636e-06, | |
| "loss": 0.2299, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 11.96, | |
| "learning_rate": 8.330909090909092e-06, | |
| "loss": 0.2563, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 12.17, | |
| "learning_rate": 8.285454545454547e-06, | |
| "loss": 0.1685, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 12.38, | |
| "learning_rate": 8.24e-06, | |
| "loss": 0.2457, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 8.194545454545454e-06, | |
| "loss": 0.1859, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 12.79, | |
| "learning_rate": 8.14909090909091e-06, | |
| "loss": 0.1883, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 8.103636363636365e-06, | |
| "loss": 0.1484, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "learning_rate": 8.058181818181819e-06, | |
| "loss": 0.1322, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 13.42, | |
| "learning_rate": 8.012727272727273e-06, | |
| "loss": 0.1099, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 7.967272727272728e-06, | |
| "loss": 0.1455, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 13.84, | |
| "learning_rate": 7.921818181818182e-06, | |
| "loss": 0.1534, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 14.05, | |
| "learning_rate": 7.876363636363637e-06, | |
| "loss": 0.1459, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 7.830909090909091e-06, | |
| "loss": 0.1286, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 7.785454545454547e-06, | |
| "loss": 0.0906, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "learning_rate": 7.74e-06, | |
| "loss": 0.1186, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "learning_rate": 7.694545454545456e-06, | |
| "loss": 0.0855, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "learning_rate": 7.64909090909091e-06, | |
| "loss": 0.1135, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 15.31, | |
| "learning_rate": 7.603636363636364e-06, | |
| "loss": 0.0613, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 15.52, | |
| "learning_rate": 7.558181818181819e-06, | |
| "loss": 0.0682, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 15.73, | |
| "learning_rate": 7.512727272727273e-06, | |
| "loss": 0.0667, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 15.94, | |
| "learning_rate": 7.467272727272727e-06, | |
| "loss": 0.0815, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 16.15, | |
| "learning_rate": 7.421818181818183e-06, | |
| "loss": 0.0599, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "learning_rate": 7.376363636363637e-06, | |
| "loss": 0.0639, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 16.57, | |
| "learning_rate": 7.330909090909092e-06, | |
| "loss": 0.0578, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 16.78, | |
| "learning_rate": 7.2854545454545455e-06, | |
| "loss": 0.0448, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 16.78, | |
| "eval_loss": 2.7543158531188965, | |
| "eval_runtime": 502.2301, | |
| "eval_samples_per_second": 1.268, | |
| "eval_steps_per_second": 0.159, | |
| "eval_wer": 5006.62100456621, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 16.99, | |
| "learning_rate": 7.24e-06, | |
| "loss": 0.0774, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 17.2, | |
| "learning_rate": 7.194545454545456e-06, | |
| "loss": 0.0488, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 17.41, | |
| "learning_rate": 7.14909090909091e-06, | |
| "loss": 0.0417, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 17.62, | |
| "learning_rate": 7.103636363636364e-06, | |
| "loss": 0.0502, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 17.83, | |
| "learning_rate": 7.0581818181818185e-06, | |
| "loss": 0.0548, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 7.012727272727273e-06, | |
| "loss": 0.055, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 18.25, | |
| "learning_rate": 6.967272727272728e-06, | |
| "loss": 0.0507, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 18.46, | |
| "learning_rate": 6.921818181818183e-06, | |
| "loss": 0.0592, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 18.67, | |
| "learning_rate": 6.876363636363637e-06, | |
| "loss": 0.0313, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 18.88, | |
| "learning_rate": 6.8309090909090915e-06, | |
| "loss": 0.028, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 19.09, | |
| "learning_rate": 6.785454545454546e-06, | |
| "loss": 0.0392, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 19.3, | |
| "learning_rate": 6.740000000000001e-06, | |
| "loss": 0.0346, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 19.51, | |
| "learning_rate": 6.6945454545454544e-06, | |
| "loss": 0.0347, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 19.72, | |
| "learning_rate": 6.64909090909091e-06, | |
| "loss": 0.0279, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 6.6036363636363645e-06, | |
| "loss": 0.0351, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 20.14, | |
| "learning_rate": 6.558181818181819e-06, | |
| "loss": 0.0241, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 20.35, | |
| "learning_rate": 6.512727272727273e-06, | |
| "loss": 0.0337, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 20.56, | |
| "learning_rate": 6.467272727272727e-06, | |
| "loss": 0.0308, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 20.77, | |
| "learning_rate": 6.421818181818182e-06, | |
| "loss": 0.0189, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 20.98, | |
| "learning_rate": 6.3763636363636375e-06, | |
| "loss": 0.0201, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 21.19, | |
| "learning_rate": 6.330909090909091e-06, | |
| "loss": 0.02, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 21.39, | |
| "learning_rate": 6.285454545454546e-06, | |
| "loss": 0.0173, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 21.6, | |
| "learning_rate": 6.24e-06, | |
| "loss": 0.0218, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 21.81, | |
| "learning_rate": 6.194545454545455e-06, | |
| "loss": 0.027, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 22.02, | |
| "learning_rate": 6.1490909090909104e-06, | |
| "loss": 0.0188, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 22.23, | |
| "learning_rate": 6.103636363636364e-06, | |
| "loss": 0.02, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 22.44, | |
| "learning_rate": 6.058181818181819e-06, | |
| "loss": 0.0165, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 22.65, | |
| "learning_rate": 6.012727272727273e-06, | |
| "loss": 0.0115, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 22.86, | |
| "learning_rate": 5.967272727272728e-06, | |
| "loss": 0.0143, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 23.07, | |
| "learning_rate": 5.921818181818182e-06, | |
| "loss": 0.0104, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 23.28, | |
| "learning_rate": 5.876363636363636e-06, | |
| "loss": 0.0209, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 23.49, | |
| "learning_rate": 5.830909090909092e-06, | |
| "loss": 0.0056, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 23.7, | |
| "learning_rate": 5.785454545454546e-06, | |
| "loss": 0.0133, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 23.91, | |
| "learning_rate": 5.74e-06, | |
| "loss": 0.0186, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 24.12, | |
| "learning_rate": 5.694545454545455e-06, | |
| "loss": 0.0094, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 24.33, | |
| "learning_rate": 5.649090909090909e-06, | |
| "loss": 0.0069, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 24.54, | |
| "learning_rate": 5.603636363636365e-06, | |
| "loss": 0.0115, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 24.75, | |
| "learning_rate": 5.5581818181818185e-06, | |
| "loss": 0.0045, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 24.96, | |
| "learning_rate": 5.512727272727273e-06, | |
| "loss": 0.0089, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 25.17, | |
| "learning_rate": 5.467272727272728e-06, | |
| "loss": 0.0089, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 25.17, | |
| "eval_loss": 2.8150265216827393, | |
| "eval_runtime": 495.3178, | |
| "eval_samples_per_second": 1.286, | |
| "eval_steps_per_second": 0.162, | |
| "eval_wer": 5198.934550989346, | |
| "step": 3000 | |
| } | |
| ], | |
| "max_steps": 6000, | |
| "num_train_epochs": 51, | |
| "total_flos": 1.385844809416704e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |