| { | |
| "best_metric": 100.0, | |
| "best_model_checkpoint": "./Malasar_ASR_DICtandLuke/checkpoint-1000", | |
| "epoch": 5.692599620493358, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 4.682, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.8e-07, | |
| "loss": 3.5297, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 2.0438, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8800000000000002e-06, | |
| "loss": 1.0543, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 2.38e-06, | |
| "loss": 0.2948, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.88e-06, | |
| "loss": 0.1761, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.3800000000000007e-06, | |
| "loss": 0.1481, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.88e-06, | |
| "loss": 0.1391, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.38e-06, | |
| "loss": 0.1175, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.880000000000001e-06, | |
| "loss": 0.1302, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.380000000000001e-06, | |
| "loss": 0.1227, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 0.1428, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.380000000000001e-06, | |
| "loss": 0.1069, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.88e-06, | |
| "loss": 0.125, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.3800000000000005e-06, | |
| "loss": 0.1326, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.88e-06, | |
| "loss": 0.1077, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.380000000000001e-06, | |
| "loss": 0.1142, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 8.880000000000001e-06, | |
| "loss": 0.107, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.38e-06, | |
| "loss": 0.0887, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.88e-06, | |
| "loss": 0.0898, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.945714285714286e-06, | |
| "loss": 0.1365, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.874285714285715e-06, | |
| "loss": 0.1158, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 9.802857142857144e-06, | |
| "loss": 0.0671, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 9.731428571428573e-06, | |
| "loss": 0.0619, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.0854, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 9.58857142857143e-06, | |
| "loss": 0.086, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 9.517142857142859e-06, | |
| "loss": 0.0782, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 9.445714285714288e-06, | |
| "loss": 0.0862, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.374285714285715e-06, | |
| "loss": 0.09, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 9.302857142857144e-06, | |
| "loss": 0.1011, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.231428571428573e-06, | |
| "loss": 0.0949, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.16e-06, | |
| "loss": 0.1001, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.08857142857143e-06, | |
| "loss": 0.0733, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.017142857142858e-06, | |
| "loss": 0.1032, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.945714285714286e-06, | |
| "loss": 0.1131, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.874285714285715e-06, | |
| "loss": 0.0747, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.802857142857144e-06, | |
| "loss": 0.0618, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.731428571428571e-06, | |
| "loss": 0.0719, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 8.66e-06, | |
| "loss": 0.0675, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 8.588571428571429e-06, | |
| "loss": 0.0795, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "eval_loss": 0.07146687060594559, | |
| "eval_runtime": 569.2334, | |
| "eval_samples_per_second": 3.29, | |
| "eval_steps_per_second": 0.207, | |
| "eval_wer": 100.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.517142857142858e-06, | |
| "loss": 0.0718, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 8.445714285714285e-06, | |
| "loss": 0.0684, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 8.374285714285714e-06, | |
| "loss": 0.0814, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 8.302857142857143e-06, | |
| "loss": 0.0464, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 8.231428571428572e-06, | |
| "loss": 0.056, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 8.16e-06, | |
| "loss": 0.0396, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 8.088571428571429e-06, | |
| "loss": 0.0507, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 8.017142857142858e-06, | |
| "loss": 0.0394, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 7.945714285714287e-06, | |
| "loss": 0.0625, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 7.874285714285716e-06, | |
| "loss": 0.0517, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 7.802857142857143e-06, | |
| "loss": 0.0404, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 7.731428571428572e-06, | |
| "loss": 0.0836, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 0.0602, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.588571428571429e-06, | |
| "loss": 0.0707, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 7.5171428571428575e-06, | |
| "loss": 0.0475, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 7.445714285714286e-06, | |
| "loss": 0.037, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 7.374285714285715e-06, | |
| "loss": 0.0594, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 7.302857142857144e-06, | |
| "loss": 0.0462, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 7.231428571428573e-06, | |
| "loss": 0.0462, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 7.16e-06, | |
| "loss": 0.0335, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.088571428571429e-06, | |
| "loss": 0.0632, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 7.017142857142858e-06, | |
| "loss": 0.0447, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 6.945714285714287e-06, | |
| "loss": 0.0444, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.874285714285714e-06, | |
| "loss": 0.0373, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 6.8028571428571434e-06, | |
| "loss": 0.0293, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 6.7314285714285724e-06, | |
| "loss": 0.0386, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 0.0259, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 6.588571428571429e-06, | |
| "loss": 0.0233, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 6.517142857142858e-06, | |
| "loss": 0.0404, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 6.445714285714286e-06, | |
| "loss": 0.0323, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 6.374285714285715e-06, | |
| "loss": 0.0301, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 6.302857142857144e-06, | |
| "loss": 0.0212, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 6.231428571428571e-06, | |
| "loss": 0.0425, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 6.16e-06, | |
| "loss": 0.0217, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 6.088571428571429e-06, | |
| "loss": 0.0359, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 6.017142857142858e-06, | |
| "loss": 0.0574, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 5.945714285714286e-06, | |
| "loss": 0.0306, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 5.874285714285715e-06, | |
| "loss": 0.0196, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 5.802857142857144e-06, | |
| "loss": 0.0306, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 5.731428571428572e-06, | |
| "loss": 0.0305, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "eval_loss": 0.07201045006513596, | |
| "eval_runtime": 635.3294, | |
| "eval_samples_per_second": 2.948, | |
| "eval_steps_per_second": 0.186, | |
| "eval_wer": 100.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 5.66e-06, | |
| "loss": 0.0379, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 5.588571428571429e-06, | |
| "loss": 0.0299, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 5.517142857142857e-06, | |
| "loss": 0.0282, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 5.445714285714286e-06, | |
| "loss": 0.0356, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 5.374285714285715e-06, | |
| "loss": 0.0183, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 5.3028571428571425e-06, | |
| "loss": 0.018, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 5.2314285714285716e-06, | |
| "loss": 0.0204, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 5.1600000000000006e-06, | |
| "loss": 0.0226, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 5.08857142857143e-06, | |
| "loss": 0.0196, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 5.017142857142857e-06, | |
| "loss": 0.0137, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 4.945714285714286e-06, | |
| "loss": 0.017, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 4.874285714285715e-06, | |
| "loss": 0.0233, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 4.802857142857143e-06, | |
| "loss": 0.0139, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 4.731428571428572e-06, | |
| "loss": 0.0142, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 4.66e-06, | |
| "loss": 0.0201, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 4.588571428571429e-06, | |
| "loss": 0.0109, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 4.5171428571428575e-06, | |
| "loss": 0.0162, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 4.445714285714286e-06, | |
| "loss": 0.0169, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 4.374285714285715e-06, | |
| "loss": 0.022, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 4.302857142857143e-06, | |
| "loss": 0.0185, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 4.231428571428572e-06, | |
| "loss": 0.0156, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 4.16e-06, | |
| "loss": 0.0188, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 4.088571428571429e-06, | |
| "loss": 0.0245, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 4.017142857142857e-06, | |
| "loss": 0.0192, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 3.945714285714286e-06, | |
| "loss": 0.015, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.874285714285715e-06, | |
| "loss": 0.0365, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 3.802857142857143e-06, | |
| "loss": 0.0065, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 3.731428571428572e-06, | |
| "loss": 0.0102, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 3.66e-06, | |
| "loss": 0.0118, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 3.588571428571429e-06, | |
| "loss": 0.0065, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 3.5171428571428573e-06, | |
| "loss": 0.0143, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 3.4457142857142863e-06, | |
| "loss": 0.0059, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 3.3742857142857145e-06, | |
| "loss": 0.0139, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 3.302857142857143e-06, | |
| "loss": 0.0068, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 3.2314285714285716e-06, | |
| "loss": 0.0153, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 3.1600000000000002e-06, | |
| "loss": 0.0059, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 3.0885714285714284e-06, | |
| "loss": 0.0043, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 3.0171428571428574e-06, | |
| "loss": 0.0083, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 2.945714285714286e-06, | |
| "loss": 0.0092, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 2.8742857142857146e-06, | |
| "loss": 0.0114, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "eval_loss": 0.08549218624830246, | |
| "eval_runtime": 577.6517, | |
| "eval_samples_per_second": 3.242, | |
| "eval_steps_per_second": 0.204, | |
| "eval_wer": 100.0, | |
| "step": 3000 | |
| } | |
| ], | |
| "max_steps": 4000, | |
| "num_train_epochs": 8, | |
| "total_flos": 9.79066287783936e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |