| { | |
| "best_metric": 41.611670718999655, | |
| "best_model_checkpoint": "./checkpoint-2000", | |
| "epoch": 25.12562814070352, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 2.032573113826056e-06, | |
| "loss": 4.0759, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 2.5560637913393917e-06, | |
| "loss": 2.3838, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.8458349997907386e-06, | |
| "loss": 1.7165, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.047233714281761e-06, | |
| "loss": 1.4279, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.2017484258746856e-06, | |
| "loss": 1.2448, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.327137518731624e-06, | |
| "loss": 1.0629, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.4326589207538503e-06, | |
| "loss": 0.9782, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.52375591492281e-06, | |
| "loss": 0.8892, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.603901956908609e-06, | |
| "loss": 0.7122, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.6754496284817233e-06, | |
| "loss": 0.6931, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.7400664073586386e-06, | |
| "loss": 0.6402, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.7989773015151948e-06, | |
| "loss": 0.6003, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.853108753462177e-06, | |
| "loss": 0.5986, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.903178415954904e-06, | |
| "loss": 0.5907, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.949753530938024e-06, | |
| "loss": 0.5449, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 3.993290230944718e-06, | |
| "loss": 0.5208, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 3.9826086956521735e-06, | |
| "loss": 0.3515, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 3.960869565217391e-06, | |
| "loss": 0.3225, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 3.939130434782609e-06, | |
| "loss": 0.3388, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.917391304347825e-06, | |
| "loss": 0.2936, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 3.895652173913044e-06, | |
| "loss": 0.3145, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 3.8739130434782605e-06, | |
| "loss": 0.3157, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.852173913043478e-06, | |
| "loss": 0.3372, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 3.830434782608696e-06, | |
| "loss": 0.2957, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.808695652173913e-06, | |
| "loss": 0.1675, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.7869565217391303e-06, | |
| "loss": 0.1609, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.765217391304348e-06, | |
| "loss": 0.1738, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 3.743478260869565e-06, | |
| "loss": 0.1605, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 3.721739130434782e-06, | |
| "loss": 0.1595, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.7e-06, | |
| "loss": 0.1484, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.6782608695652173e-06, | |
| "loss": 0.164, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.6565217391304344e-06, | |
| "loss": 0.1289, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.634782608695652e-06, | |
| "loss": 0.0745, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.613043478260869e-06, | |
| "loss": 0.074, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.591304347826087e-06, | |
| "loss": 0.0826, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 3.5695652173913043e-06, | |
| "loss": 0.0699, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.5478260869565214e-06, | |
| "loss": 0.0771, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 3.526086956521739e-06, | |
| "loss": 0.0795, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 3.5043478260869565e-06, | |
| "loss": 0.0662, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.4826086956521737e-06, | |
| "loss": 0.0602, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "eval_loss": 0.732421875, | |
| "eval_runtime": 308.8997, | |
| "eval_samples_per_second": 6.86, | |
| "eval_steps_per_second": 0.431, | |
| "eval_wer": 43.69572768322334, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 3.4608695652173912e-06, | |
| "loss": 0.0329, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 3.4391304347826084e-06, | |
| "loss": 0.038, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 3.417391304347826e-06, | |
| "loss": 0.0346, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 3.3956521739130435e-06, | |
| "loss": 0.0321, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 3.3739130434782606e-06, | |
| "loss": 0.0308, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 3.352173913043478e-06, | |
| "loss": 0.0342, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 3.3304347826086953e-06, | |
| "loss": 0.0359, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 3.3086956521739133e-06, | |
| "loss": 0.0328, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 3.2869565217391305e-06, | |
| "loss": 0.0171, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 6.28, | |
| "learning_rate": 3.2652173913043476e-06, | |
| "loss": 0.0165, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 3.243478260869565e-06, | |
| "loss": 0.0186, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 3.2217391304347823e-06, | |
| "loss": 0.0226, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 3.2e-06, | |
| "loss": 0.0151, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 3.1782608695652174e-06, | |
| "loss": 0.0227, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 3.1565217391304346e-06, | |
| "loss": 0.0175, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.134782608695652e-06, | |
| "loss": 0.0151, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 3.1130434782608693e-06, | |
| "loss": 0.0087, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 7.29, | |
| "learning_rate": 3.091304347826087e-06, | |
| "loss": 0.0083, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 3.0695652173913044e-06, | |
| "loss": 0.0087, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 3.0478260869565215e-06, | |
| "loss": 0.0118, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 3.0260869565217387e-06, | |
| "loss": 0.0075, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 3.0043478260869566e-06, | |
| "loss": 0.0082, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 7.91, | |
| "learning_rate": 2.9826086956521738e-06, | |
| "loss": 0.012, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 2.9608695652173913e-06, | |
| "loss": 0.0102, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 8.17, | |
| "learning_rate": 2.9391304347826085e-06, | |
| "loss": 0.0065, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 2.9173913043478256e-06, | |
| "loss": 0.006, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "learning_rate": 2.8956521739130436e-06, | |
| "loss": 0.0051, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 2.8739130434782608e-06, | |
| "loss": 0.0058, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 2.852173913043478e-06, | |
| "loss": 0.0065, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 2.8304347826086955e-06, | |
| "loss": 0.0085, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 2.808695652173913e-06, | |
| "loss": 0.0077, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 2.7869565217391306e-06, | |
| "loss": 0.006, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 9.17, | |
| "learning_rate": 2.7652173913043477e-06, | |
| "loss": 0.0044, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 2.743478260869565e-06, | |
| "loss": 0.0037, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 2.7217391304347824e-06, | |
| "loss": 0.0065, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 2.7e-06, | |
| "loss": 0.0052, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 2.6782608695652175e-06, | |
| "loss": 0.0036, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 2.6565217391304347e-06, | |
| "loss": 0.0027, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 9.92, | |
| "learning_rate": 2.634782608695652e-06, | |
| "loss": 0.003, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 2.6130434782608694e-06, | |
| "loss": 0.0036, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "eval_loss": 0.8486328125, | |
| "eval_runtime": 307.4259, | |
| "eval_samples_per_second": 6.893, | |
| "eval_steps_per_second": 0.433, | |
| "eval_wer": 41.611670718999655, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 10.18, | |
| "learning_rate": 2.591304347826087e-06, | |
| "loss": 0.0032, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 2.569565217391304e-06, | |
| "loss": 0.0026, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 2.5478260869565217e-06, | |
| "loss": 0.0033, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "learning_rate": 2.526086956521739e-06, | |
| "loss": 0.0037, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 10.68, | |
| "learning_rate": 2.5043478260869568e-06, | |
| "loss": 0.0038, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 10.8, | |
| "learning_rate": 2.482608695652174e-06, | |
| "loss": 0.0031, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 10.93, | |
| "learning_rate": 2.460869565217391e-06, | |
| "loss": 0.0032, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "learning_rate": 2.4391304347826086e-06, | |
| "loss": 0.0021, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 11.18, | |
| "learning_rate": 2.4173913043478258e-06, | |
| "loss": 0.0032, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 11.31, | |
| "learning_rate": 2.3956521739130433e-06, | |
| "loss": 0.0023, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 11.43, | |
| "learning_rate": 2.373913043478261e-06, | |
| "loss": 0.0015, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "learning_rate": 2.352173913043478e-06, | |
| "loss": 0.0016, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 11.68, | |
| "learning_rate": 2.3304347826086956e-06, | |
| "loss": 0.0015, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 2.308695652173913e-06, | |
| "loss": 0.002, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 2.2869565217391303e-06, | |
| "loss": 0.0017, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 2.265217391304348e-06, | |
| "loss": 0.0015, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 2.243478260869565e-06, | |
| "loss": 0.002, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 2.2217391304347825e-06, | |
| "loss": 0.0011, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 2.2e-06, | |
| "loss": 0.0013, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 2.1782608695652173e-06, | |
| "loss": 0.0018, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 12.69, | |
| "learning_rate": 2.156521739130435e-06, | |
| "loss": 0.0011, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "learning_rate": 2.134782608695652e-06, | |
| "loss": 0.0012, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 2.113043478260869e-06, | |
| "loss": 0.0017, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 13.07, | |
| "learning_rate": 2.091304347826087e-06, | |
| "loss": 0.0026, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 2.0695652173913042e-06, | |
| "loss": 0.0014, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "learning_rate": 2.0478260869565218e-06, | |
| "loss": 0.001, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 2.026086956521739e-06, | |
| "loss": 0.0008, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 2.0043478260869565e-06, | |
| "loss": 0.001, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 13.69, | |
| "learning_rate": 1.982608695652174e-06, | |
| "loss": 0.0008, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 13.82, | |
| "learning_rate": 1.960869565217391e-06, | |
| "loss": 0.0008, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "learning_rate": 1.9391304347826083e-06, | |
| "loss": 0.0012, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 14.07, | |
| "learning_rate": 1.917391304347826e-06, | |
| "loss": 0.0015, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 14.2, | |
| "learning_rate": 1.8956521739130432e-06, | |
| "loss": 0.0006, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 1.8739130434782608e-06, | |
| "loss": 0.0006, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 14.45, | |
| "learning_rate": 1.8521739130434781e-06, | |
| "loss": 0.0021, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "learning_rate": 1.8304347826086957e-06, | |
| "loss": 0.0007, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 1.8086956521739129e-06, | |
| "loss": 0.0007, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 14.82, | |
| "learning_rate": 1.7869565217391302e-06, | |
| "loss": 0.0007, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 14.95, | |
| "learning_rate": 1.7652173913043478e-06, | |
| "loss": 0.0012, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 15.08, | |
| "learning_rate": 1.7434782608695651e-06, | |
| "loss": 0.001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 15.08, | |
| "eval_loss": 0.9033203125, | |
| "eval_runtime": 306.7936, | |
| "eval_samples_per_second": 6.907, | |
| "eval_steps_per_second": 0.434, | |
| "eval_wer": 42.04584925321292, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 1.723478260869565e-06, | |
| "loss": 0.0006, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 15.33, | |
| "learning_rate": 1.7017391304347827e-06, | |
| "loss": 0.001, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "learning_rate": 1.6799999999999998e-06, | |
| "loss": 0.0006, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 15.58, | |
| "learning_rate": 1.6582608695652174e-06, | |
| "loss": 0.0006, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 1.6365217391304347e-06, | |
| "loss": 0.0006, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 15.83, | |
| "learning_rate": 1.6147826086956523e-06, | |
| "loss": 0.0005, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 15.95, | |
| "learning_rate": 1.5930434782608694e-06, | |
| "loss": 0.0011, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 16.08, | |
| "learning_rate": 1.5713043478260868e-06, | |
| "loss": 0.0005, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 16.21, | |
| "learning_rate": 1.5495652173913043e-06, | |
| "loss": 0.0005, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 16.33, | |
| "learning_rate": 1.5278260869565217e-06, | |
| "loss": 0.0005, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 16.46, | |
| "learning_rate": 1.506086956521739e-06, | |
| "loss": 0.0005, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 16.58, | |
| "learning_rate": 1.4843478260869564e-06, | |
| "loss": 0.0006, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 16.71, | |
| "learning_rate": 1.462608695652174e-06, | |
| "loss": 0.0006, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 16.83, | |
| "learning_rate": 1.4408695652173913e-06, | |
| "loss": 0.0006, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 16.96, | |
| "learning_rate": 1.4191304347826084e-06, | |
| "loss": 0.0005, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 17.09, | |
| "learning_rate": 1.397391304347826e-06, | |
| "loss": 0.0005, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 17.21, | |
| "learning_rate": 1.3756521739130433e-06, | |
| "loss": 0.0005, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 17.34, | |
| "learning_rate": 1.353913043478261e-06, | |
| "loss": 0.0005, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 17.46, | |
| "learning_rate": 1.332173913043478e-06, | |
| "loss": 0.0005, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 17.59, | |
| "learning_rate": 1.3104347826086956e-06, | |
| "loss": 0.0004, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 17.71, | |
| "learning_rate": 1.288695652173913e-06, | |
| "loss": 0.0005, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 17.84, | |
| "learning_rate": 1.2669565217391305e-06, | |
| "loss": 0.0004, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 17.96, | |
| "learning_rate": 1.2452173913043479e-06, | |
| "loss": 0.0005, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 18.09, | |
| "learning_rate": 1.223478260869565e-06, | |
| "loss": 0.0004, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 1.2017391304347826e-06, | |
| "loss": 0.0004, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 18.34, | |
| "learning_rate": 1.18e-06, | |
| "loss": 0.0004, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 18.47, | |
| "learning_rate": 1.1582608695652175e-06, | |
| "loss": 0.0004, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 18.59, | |
| "learning_rate": 1.1365217391304346e-06, | |
| "loss": 0.0004, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 18.72, | |
| "learning_rate": 1.1147826086956522e-06, | |
| "loss": 0.0004, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 18.84, | |
| "learning_rate": 1.0930434782608695e-06, | |
| "loss": 0.0004, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "learning_rate": 1.0713043478260869e-06, | |
| "loss": 0.0004, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 19.1, | |
| "learning_rate": 1.0495652173913042e-06, | |
| "loss": 0.0004, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 19.22, | |
| "learning_rate": 1.0278260869565216e-06, | |
| "loss": 0.0004, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 19.35, | |
| "learning_rate": 1.0060869565217392e-06, | |
| "loss": 0.0004, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 19.47, | |
| "learning_rate": 9.843478260869565e-07, | |
| "loss": 0.0004, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 19.6, | |
| "learning_rate": 9.626086956521739e-07, | |
| "loss": 0.0004, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 19.72, | |
| "learning_rate": 9.408695652173912e-07, | |
| "loss": 0.0004, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 19.85, | |
| "learning_rate": 9.191304347826087e-07, | |
| "loss": 0.0004, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 19.97, | |
| "learning_rate": 8.97391304347826e-07, | |
| "loss": 0.0004, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 20.1, | |
| "learning_rate": 8.756521739130435e-07, | |
| "loss": 0.0004, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 20.1, | |
| "eval_loss": 0.93505859375, | |
| "eval_runtime": 308.5813, | |
| "eval_samples_per_second": 6.867, | |
| "eval_steps_per_second": 0.431, | |
| "eval_wer": 41.68113928447378, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 20.23, | |
| "learning_rate": 8.556521739130435e-07, | |
| "loss": 0.0004, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 20.35, | |
| "learning_rate": 8.339130434782609e-07, | |
| "loss": 0.0004, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 20.48, | |
| "learning_rate": 8.121739130434783e-07, | |
| "loss": 0.0004, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 20.6, | |
| "learning_rate": 7.904347826086957e-07, | |
| "loss": 0.0004, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 20.73, | |
| "learning_rate": 7.68695652173913e-07, | |
| "loss": 0.0003, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 20.85, | |
| "learning_rate": 7.469565217391304e-07, | |
| "loss": 0.0004, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 20.98, | |
| "learning_rate": 7.252173913043478e-07, | |
| "loss": 0.0004, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 21.11, | |
| "learning_rate": 7.034782608695652e-07, | |
| "loss": 0.0003, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 21.23, | |
| "learning_rate": 6.817391304347826e-07, | |
| "loss": 0.0003, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 21.36, | |
| "learning_rate": 6.6e-07, | |
| "loss": 0.0003, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 21.48, | |
| "learning_rate": 6.382608695652174e-07, | |
| "loss": 0.0003, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 21.61, | |
| "learning_rate": 6.165217391304347e-07, | |
| "loss": 0.0003, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 21.73, | |
| "learning_rate": 5.947826086956521e-07, | |
| "loss": 0.0003, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 21.86, | |
| "learning_rate": 5.730434782608695e-07, | |
| "loss": 0.0003, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 21.98, | |
| "learning_rate": 5.513043478260869e-07, | |
| "loss": 0.0003, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 22.11, | |
| "learning_rate": 5.295652173913044e-07, | |
| "loss": 0.0003, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 22.24, | |
| "learning_rate": 5.078260869565217e-07, | |
| "loss": 0.0003, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 22.36, | |
| "learning_rate": 4.86086956521739e-07, | |
| "loss": 0.0003, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 22.49, | |
| "learning_rate": 4.643478260869565e-07, | |
| "loss": 0.0003, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 22.61, | |
| "learning_rate": 4.426086956521739e-07, | |
| "loss": 0.0003, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 22.74, | |
| "learning_rate": 4.208695652173913e-07, | |
| "loss": 0.0003, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 22.86, | |
| "learning_rate": 3.9913043478260867e-07, | |
| "loss": 0.0003, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 22.99, | |
| "learning_rate": 3.7739130434782607e-07, | |
| "loss": 0.0003, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 23.12, | |
| "learning_rate": 3.556521739130435e-07, | |
| "loss": 0.0003, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 23.24, | |
| "learning_rate": 3.339130434782609e-07, | |
| "loss": 0.0003, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 23.37, | |
| "learning_rate": 3.1217391304347823e-07, | |
| "loss": 0.0003, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 23.49, | |
| "learning_rate": 2.9043478260869563e-07, | |
| "loss": 0.0003, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 23.62, | |
| "learning_rate": 2.6869565217391304e-07, | |
| "loss": 0.0003, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 23.74, | |
| "learning_rate": 2.4695652173913044e-07, | |
| "loss": 0.0003, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 23.87, | |
| "learning_rate": 2.252173913043478e-07, | |
| "loss": 0.0003, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 23.99, | |
| "learning_rate": 2.0347826086956522e-07, | |
| "loss": 0.0003, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 24.12, | |
| "learning_rate": 1.817391304347826e-07, | |
| "loss": 0.0003, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 24.25, | |
| "learning_rate": 1.6e-07, | |
| "loss": 0.0003, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 24.37, | |
| "learning_rate": 1.3826086956521738e-07, | |
| "loss": 0.0003, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 24.5, | |
| "learning_rate": 1.1652173913043478e-07, | |
| "loss": 0.0003, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 24.62, | |
| "learning_rate": 9.478260869565218e-08, | |
| "loss": 0.0003, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 24.75, | |
| "learning_rate": 7.304347826086957e-08, | |
| "loss": 0.0003, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 24.87, | |
| "learning_rate": 5.1304347826086956e-08, | |
| "loss": 0.0003, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 2.9565217391304347e-08, | |
| "loss": 0.0003, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 25.13, | |
| "learning_rate": 7.826086956521739e-09, | |
| "loss": 0.0003, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 25.13, | |
| "eval_loss": 0.94677734375, | |
| "eval_runtime": 308.4233, | |
| "eval_samples_per_second": 6.87, | |
| "eval_steps_per_second": 0.431, | |
| "eval_wer": 41.78534213268496, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 25.13, | |
| "step": 5000, | |
| "total_flos": 1.6324575581427466e+20, | |
| "train_loss": 0.11887830903530121, | |
| "train_runtime": 19763.0817, | |
| "train_samples_per_second": 8.096, | |
| "train_steps_per_second": 0.253 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 26, | |
| "total_flos": 1.6324575581427466e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |