| { | |
| "best_metric": 48.772130211307825, | |
| "best_model_checkpoint": "./Malasar_ASR_Luke/checkpoint-3000", | |
| "epoch": 107.14285714285714, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 1.923, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.9421, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.4848, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.3549, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.2811, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.2073, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.1832, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.1453, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.1225, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.098, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.076, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 10.71, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.0602, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 11.61, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.0477, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.0418, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 13.39, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.0314, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.03, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 15.18, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.0251, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 16.07, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.0264, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 16.96, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.0235, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 17.86, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.0204, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 9.942857142857145e-06, | |
| "loss": 0.0174, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 19.64, | |
| "learning_rate": 9.871428571428572e-06, | |
| "loss": 0.0172, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 20.54, | |
| "learning_rate": 9.800000000000001e-06, | |
| "loss": 0.0167, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 21.43, | |
| "learning_rate": 9.72857142857143e-06, | |
| "loss": 0.0161, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 22.32, | |
| "learning_rate": 9.657142857142859e-06, | |
| "loss": 0.0118, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 23.21, | |
| "learning_rate": 9.585714285714286e-06, | |
| "loss": 0.0112, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 24.11, | |
| "learning_rate": 9.514285714285715e-06, | |
| "loss": 0.01, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 9.442857142857144e-06, | |
| "loss": 0.0096, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 25.89, | |
| "learning_rate": 9.371428571428572e-06, | |
| "loss": 0.0093, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 26.79, | |
| "learning_rate": 9.3e-06, | |
| "loss": 0.0063, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 27.68, | |
| "learning_rate": 9.22857142857143e-06, | |
| "loss": 0.0051, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 28.57, | |
| "learning_rate": 9.157142857142857e-06, | |
| "loss": 0.0043, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 29.46, | |
| "learning_rate": 9.085714285714286e-06, | |
| "loss": 0.0036, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 30.36, | |
| "learning_rate": 9.014285714285715e-06, | |
| "loss": 0.0021, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 8.942857142857142e-06, | |
| "loss": 0.0018, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 32.14, | |
| "learning_rate": 8.871428571428571e-06, | |
| "loss": 0.0013, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 33.04, | |
| "learning_rate": 8.8e-06, | |
| "loss": 0.0013, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 33.93, | |
| "learning_rate": 8.72857142857143e-06, | |
| "loss": 0.0019, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 34.82, | |
| "learning_rate": 8.657142857142858e-06, | |
| "loss": 0.0009, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 35.71, | |
| "learning_rate": 8.585714285714286e-06, | |
| "loss": 0.0005, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 35.71, | |
| "eval_loss": 0.48349568247795105, | |
| "eval_runtime": 86.9581, | |
| "eval_samples_per_second": 1.115, | |
| "eval_steps_per_second": 0.08, | |
| "eval_wer": 51.11364934323244, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 36.61, | |
| "learning_rate": 8.514285714285715e-06, | |
| "loss": 0.0004, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 37.5, | |
| "learning_rate": 8.442857142857144e-06, | |
| "loss": 0.0003, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 38.39, | |
| "learning_rate": 8.371428571428573e-06, | |
| "loss": 0.0006, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 39.29, | |
| "learning_rate": 8.3e-06, | |
| "loss": 0.0012, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 40.18, | |
| "learning_rate": 8.22857142857143e-06, | |
| "loss": 0.0016, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 41.07, | |
| "learning_rate": 8.157142857142858e-06, | |
| "loss": 0.0011, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 41.96, | |
| "learning_rate": 8.085714285714287e-06, | |
| "loss": 0.0011, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 42.86, | |
| "learning_rate": 8.014285714285715e-06, | |
| "loss": 0.0019, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 43.75, | |
| "learning_rate": 7.942857142857144e-06, | |
| "loss": 0.0019, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 44.64, | |
| "learning_rate": 7.871428571428573e-06, | |
| "loss": 0.0015, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 45.54, | |
| "learning_rate": 7.800000000000002e-06, | |
| "loss": 0.0018, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 46.43, | |
| "learning_rate": 7.72857142857143e-06, | |
| "loss": 0.0015, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 47.32, | |
| "learning_rate": 7.657142857142858e-06, | |
| "loss": 0.0028, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 48.21, | |
| "learning_rate": 7.585714285714286e-06, | |
| "loss": 0.0027, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 49.11, | |
| "learning_rate": 7.514285714285715e-06, | |
| "loss": 0.0042, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 7.442857142857144e-06, | |
| "loss": 0.0042, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 50.89, | |
| "learning_rate": 7.371428571428571e-06, | |
| "loss": 0.0031, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 51.79, | |
| "learning_rate": 7.3e-06, | |
| "loss": 0.0026, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 52.68, | |
| "learning_rate": 7.2285714285714294e-06, | |
| "loss": 0.0027, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 53.57, | |
| "learning_rate": 7.1571428571428584e-06, | |
| "loss": 0.0036, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 54.46, | |
| "learning_rate": 7.085714285714286e-06, | |
| "loss": 0.0031, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 55.36, | |
| "learning_rate": 7.014285714285715e-06, | |
| "loss": 0.0021, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 56.25, | |
| "learning_rate": 6.942857142857144e-06, | |
| "loss": 0.0015, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 57.14, | |
| "learning_rate": 6.871428571428572e-06, | |
| "loss": 0.0013, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 58.04, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 0.0012, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 58.93, | |
| "learning_rate": 6.728571428571429e-06, | |
| "loss": 0.0009, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 59.82, | |
| "learning_rate": 6.657142857142857e-06, | |
| "loss": 0.0005, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 60.71, | |
| "learning_rate": 6.585714285714286e-06, | |
| "loss": 0.0003, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 61.61, | |
| "learning_rate": 6.514285714285715e-06, | |
| "loss": 0.001, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 62.5, | |
| "learning_rate": 6.442857142857143e-06, | |
| "loss": 0.0008, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 63.39, | |
| "learning_rate": 6.371428571428572e-06, | |
| "loss": 0.0006, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 64.29, | |
| "learning_rate": 6.300000000000001e-06, | |
| "loss": 0.0004, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 65.18, | |
| "learning_rate": 6.22857142857143e-06, | |
| "loss": 0.0003, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 66.07, | |
| "learning_rate": 6.157142857142858e-06, | |
| "loss": 0.0004, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 66.96, | |
| "learning_rate": 6.085714285714286e-06, | |
| "loss": 0.0003, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 67.86, | |
| "learning_rate": 6.014285714285715e-06, | |
| "loss": 0.0011, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 68.75, | |
| "learning_rate": 5.942857142857143e-06, | |
| "loss": 0.0004, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 69.64, | |
| "learning_rate": 5.871428571428572e-06, | |
| "loss": 0.0002, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 70.54, | |
| "learning_rate": 5.8e-06, | |
| "loss": 0.0002, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 71.43, | |
| "learning_rate": 5.7285714285714285e-06, | |
| "loss": 0.0002, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 71.43, | |
| "eval_loss": 0.5415698289871216, | |
| "eval_runtime": 77.537, | |
| "eval_samples_per_second": 1.251, | |
| "eval_steps_per_second": 0.09, | |
| "eval_wer": 49.0005711022273, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 72.32, | |
| "learning_rate": 5.6571428571428576e-06, | |
| "loss": 0.0002, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 73.21, | |
| "learning_rate": 5.5857142857142866e-06, | |
| "loss": 0.0001, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 74.11, | |
| "learning_rate": 5.514285714285714e-06, | |
| "loss": 0.0001, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "learning_rate": 5.442857142857143e-06, | |
| "loss": 0.0001, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 75.89, | |
| "learning_rate": 5.371428571428572e-06, | |
| "loss": 0.0001, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 76.79, | |
| "learning_rate": 5.300000000000001e-06, | |
| "loss": 0.0001, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 77.68, | |
| "learning_rate": 5.22857142857143e-06, | |
| "loss": 0.0001, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 78.57, | |
| "learning_rate": 5.157142857142857e-06, | |
| "loss": 0.0001, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 79.46, | |
| "learning_rate": 5.085714285714286e-06, | |
| "loss": 0.0001, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 80.36, | |
| "learning_rate": 5.014285714285715e-06, | |
| "loss": 0.0001, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 81.25, | |
| "learning_rate": 4.9428571428571435e-06, | |
| "loss": 0.0001, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 82.14, | |
| "learning_rate": 4.871428571428572e-06, | |
| "loss": 0.0001, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 83.04, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 0.0001, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 83.93, | |
| "learning_rate": 4.728571428571429e-06, | |
| "loss": 0.0001, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 84.82, | |
| "learning_rate": 4.657142857142857e-06, | |
| "loss": 0.0001, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 85.71, | |
| "learning_rate": 4.585714285714286e-06, | |
| "loss": 0.0001, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 86.61, | |
| "learning_rate": 4.514285714285714e-06, | |
| "loss": 0.0001, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 87.5, | |
| "learning_rate": 4.442857142857143e-06, | |
| "loss": 0.0001, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 88.39, | |
| "learning_rate": 4.371428571428572e-06, | |
| "loss": 0.0001, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 89.29, | |
| "learning_rate": 4.3e-06, | |
| "loss": 0.0001, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 90.18, | |
| "learning_rate": 4.228571428571429e-06, | |
| "loss": 0.0001, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 91.07, | |
| "learning_rate": 4.1571428571428575e-06, | |
| "loss": 0.0001, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 91.96, | |
| "learning_rate": 4.0857142857142865e-06, | |
| "loss": 0.0001, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 92.86, | |
| "learning_rate": 4.014285714285715e-06, | |
| "loss": 0.0001, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 93.75, | |
| "learning_rate": 3.942857142857143e-06, | |
| "loss": 0.0001, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 94.64, | |
| "learning_rate": 3.871428571428572e-06, | |
| "loss": 0.0001, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 95.54, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 0.0001, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 96.43, | |
| "learning_rate": 3.7285714285714286e-06, | |
| "loss": 0.0, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 97.32, | |
| "learning_rate": 3.6571428571428576e-06, | |
| "loss": 0.0001, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 98.21, | |
| "learning_rate": 3.5857142857142862e-06, | |
| "loss": 0.0001, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 99.11, | |
| "learning_rate": 3.5142857142857144e-06, | |
| "loss": 0.0001, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "learning_rate": 3.4428571428571434e-06, | |
| "loss": 0.0001, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 100.89, | |
| "learning_rate": 3.3714285714285716e-06, | |
| "loss": 0.0001, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 101.79, | |
| "learning_rate": 3.3000000000000006e-06, | |
| "loss": 0.0001, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 102.68, | |
| "learning_rate": 3.2285714285714288e-06, | |
| "loss": 0.0, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 103.57, | |
| "learning_rate": 3.1571428571428573e-06, | |
| "loss": 0.0001, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 104.46, | |
| "learning_rate": 3.085714285714286e-06, | |
| "loss": 0.0001, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 105.36, | |
| "learning_rate": 3.0142857142857145e-06, | |
| "loss": 0.0001, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 106.25, | |
| "learning_rate": 2.9428571428571427e-06, | |
| "loss": 0.0001, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 107.14, | |
| "learning_rate": 2.8714285714285717e-06, | |
| "loss": 0.0001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 107.14, | |
| "eval_loss": 0.6008931994438171, | |
| "eval_runtime": 76.998, | |
| "eval_samples_per_second": 1.26, | |
| "eval_steps_per_second": 0.091, | |
| "eval_wer": 48.772130211307825, | |
| "step": 3000 | |
| } | |
| ], | |
| "max_steps": 4000, | |
| "num_train_epochs": 143, | |
| "total_flos": 1.974633976184832e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |