| { | |
| "best_metric": 15.356302345444602, | |
| "best_model_checkpoint": "./checkpoint-4000", | |
| "epoch": 7.0732, | |
| "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": 0.9703, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 0.9008, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 0.7319, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.94e-06, | |
| "loss": 0.661, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 0.8801, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 0.6611, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.44e-06, | |
| "loss": 0.4738, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.94e-06, | |
| "loss": 0.4535, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 0.6728, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.94e-06, | |
| "loss": 0.5901, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 0.7471, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.7742, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 0.6693, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 0.6028, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.5499, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.5324, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.527, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.52, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.6611, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.5553, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.951111111111111e-06, | |
| "loss": 0.4167, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.895555555555557e-06, | |
| "loss": 0.4138, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.84e-06, | |
| "loss": 0.6233, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.784444444444445e-06, | |
| "loss": 0.5831, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.72888888888889e-06, | |
| "loss": 0.6376, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.673333333333334e-06, | |
| "loss": 0.4743, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.617777777777778e-06, | |
| "loss": 0.495, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.562222222222223e-06, | |
| "loss": 0.5174, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.506666666666667e-06, | |
| "loss": 0.5594, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.451111111111112e-06, | |
| "loss": 0.4933, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.395555555555556e-06, | |
| "loss": 0.558, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.340000000000002e-06, | |
| "loss": 0.5609, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.284444444444444e-06, | |
| "loss": 0.3933, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.22888888888889e-06, | |
| "loss": 0.3492, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.173333333333334e-06, | |
| "loss": 0.3862, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.117777777777778e-06, | |
| "loss": 0.5433, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 9.062222222222224e-06, | |
| "loss": 0.4832, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 9.006666666666666e-06, | |
| "loss": 0.5699, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.951111111111112e-06, | |
| "loss": 0.6008, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.895555555555556e-06, | |
| "loss": 0.481, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "eval_cer": 16.17355184004031, | |
| "eval_loss": 0.34429243206977844, | |
| "eval_runtime": 1169.5321, | |
| "eval_samples_per_second": 9.047, | |
| "eval_steps_per_second": 0.566, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.84e-06, | |
| "loss": 0.3525, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.784444444444446e-06, | |
| "loss": 0.3565, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.72888888888889e-06, | |
| "loss": 0.3716, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.673333333333334e-06, | |
| "loss": 0.2919, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.617777777777778e-06, | |
| "loss": 0.3715, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.562222222222224e-06, | |
| "loss": 0.4703, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.506666666666668e-06, | |
| "loss": 0.2849, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.451111111111112e-06, | |
| "loss": 0.2747, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.395555555555557e-06, | |
| "loss": 0.2983, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.34e-06, | |
| "loss": 0.4124, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.284444444444446e-06, | |
| "loss": 0.4607, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.22888888888889e-06, | |
| "loss": 0.4049, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 8.173333333333334e-06, | |
| "loss": 0.2788, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.11777777777778e-06, | |
| "loss": 0.3577, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 8.062222222222222e-06, | |
| "loss": 0.4031, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 8.006666666666667e-06, | |
| "loss": 0.3453, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.951111111111111e-06, | |
| "loss": 0.3141, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.895555555555557e-06, | |
| "loss": 0.5154, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.840000000000001e-06, | |
| "loss": 0.3213, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.784444444444445e-06, | |
| "loss": 0.2187, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.72888888888889e-06, | |
| "loss": 0.227, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.673333333333333e-06, | |
| "loss": 0.3558, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.617777777777778e-06, | |
| "loss": 0.3023, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.562222222222223e-06, | |
| "loss": 0.4035, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 7.506666666666668e-06, | |
| "loss": 0.4757, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.451111111111111e-06, | |
| "loss": 0.3773, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.395555555555556e-06, | |
| "loss": 0.276, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.340000000000001e-06, | |
| "loss": 0.2463, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.284444444444445e-06, | |
| "loss": 0.2148, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.22888888888889e-06, | |
| "loss": 0.2195, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 7.173333333333335e-06, | |
| "loss": 0.2056, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.117777777777778e-06, | |
| "loss": 0.314, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.062222222222223e-06, | |
| "loss": 0.2783, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 7.006666666666667e-06, | |
| "loss": 0.1516, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.951111111111112e-06, | |
| "loss": 0.1622, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.8955555555555565e-06, | |
| "loss": 0.2934, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.8400000000000014e-06, | |
| "loss": 0.2843, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.784444444444445e-06, | |
| "loss": 0.286, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.7288888888888895e-06, | |
| "loss": 0.2432, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 6.6733333333333335e-06, | |
| "loss": 0.2912, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_cer": 16.52134373022164, | |
| "eval_loss": 0.34871363639831543, | |
| "eval_runtime": 1174.6783, | |
| "eval_samples_per_second": 9.008, | |
| "eval_steps_per_second": 0.564, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.617777777777778e-06, | |
| "loss": 0.2628, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 6.562222222222223e-06, | |
| "loss": 0.2413, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.5066666666666665e-06, | |
| "loss": 0.2547, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 6.451111111111111e-06, | |
| "loss": 0.3324, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.395555555555556e-06, | |
| "loss": 0.2777, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 6.34e-06, | |
| "loss": 0.1575, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.284444444444445e-06, | |
| "loss": 0.1492, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 6.22888888888889e-06, | |
| "loss": 0.1817, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.173333333333333e-06, | |
| "loss": 0.2754, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 6.117777777777778e-06, | |
| "loss": 0.2833, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.062222222222223e-06, | |
| "loss": 0.3104, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 6.006666666666667e-06, | |
| "loss": 0.3034, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.951111111111112e-06, | |
| "loss": 0.2417, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.895555555555557e-06, | |
| "loss": 0.1262, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.84e-06, | |
| "loss": 0.1571, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.784444444444445e-06, | |
| "loss": 0.1592, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.72888888888889e-06, | |
| "loss": 0.1047, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.673333333333334e-06, | |
| "loss": 0.1954, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.617777777777779e-06, | |
| "loss": 0.2349, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 5.562222222222222e-06, | |
| "loss": 0.1399, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.506666666666667e-06, | |
| "loss": 0.0758, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.451111111111112e-06, | |
| "loss": 0.1353, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.3955555555555565e-06, | |
| "loss": 0.2259, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 5.3400000000000005e-06, | |
| "loss": 0.2181, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.2844444444444454e-06, | |
| "loss": 0.2209, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 5.228888888888889e-06, | |
| "loss": 0.1323, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.1733333333333335e-06, | |
| "loss": 0.1845, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 5.117777777777778e-06, | |
| "loss": 0.1883, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.062222222222222e-06, | |
| "loss": 0.1729, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.006666666666667e-06, | |
| "loss": 0.1551, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.951111111111111e-06, | |
| "loss": 0.3004, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 4.895555555555556e-06, | |
| "loss": 0.1558, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.84e-06, | |
| "loss": 0.0912, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.784444444444445e-06, | |
| "loss": 0.0739, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.728888888888889e-06, | |
| "loss": 0.213, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.673333333333333e-06, | |
| "loss": 0.1468, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.617777777777778e-06, | |
| "loss": 0.2467, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 4.562222222222222e-06, | |
| "loss": 0.2318, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.506666666666667e-06, | |
| "loss": 0.212, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.451111111111112e-06, | |
| "loss": 0.1303, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "eval_cer": 15.395014163609662, | |
| "eval_loss": 0.3707258105278015, | |
| "eval_runtime": 1200.8636, | |
| "eval_samples_per_second": 8.811, | |
| "eval_steps_per_second": 0.551, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.395555555555556e-06, | |
| "loss": 0.0857, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 4.34e-06, | |
| "loss": 0.0936, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.284444444444445e-06, | |
| "loss": 0.0952, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.228888888888889e-06, | |
| "loss": 0.0832, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.173333333333334e-06, | |
| "loss": 0.1604, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.117777777777779e-06, | |
| "loss": 0.1396, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.062222222222223e-06, | |
| "loss": 0.0547, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 4.006666666666667e-06, | |
| "loss": 0.0545, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.951111111111112e-06, | |
| "loss": 0.1489, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.895555555555556e-06, | |
| "loss": 0.1656, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.8400000000000005e-06, | |
| "loss": 0.1681, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 3.784444444444445e-06, | |
| "loss": 0.0909, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.728888888888889e-06, | |
| "loss": 0.136, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.673333333333334e-06, | |
| "loss": 0.1511, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 3.617777777777778e-06, | |
| "loss": 0.1075, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.5622222222222224e-06, | |
| "loss": 0.1202, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 3.5066666666666673e-06, | |
| "loss": 0.1459, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.4511111111111113e-06, | |
| "loss": 0.1529, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 3.3955555555555558e-06, | |
| "loss": 0.0954, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.3400000000000006e-06, | |
| "loss": 0.0596, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 3.2844444444444447e-06, | |
| "loss": 0.1037, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.228888888888889e-06, | |
| "loss": 0.1482, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 3.173333333333334e-06, | |
| "loss": 0.1302, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.117777777777778e-06, | |
| "loss": 0.1976, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 3.0622222222222225e-06, | |
| "loss": 0.1663, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 3.0066666666666674e-06, | |
| "loss": 0.119, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 2.9511111111111114e-06, | |
| "loss": 0.0488, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.895555555555556e-06, | |
| "loss": 0.0709, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.84e-06, | |
| "loss": 0.0586, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.784444444444445e-06, | |
| "loss": 0.0422, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 2.7288888888888893e-06, | |
| "loss": 0.0881, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.6733333333333333e-06, | |
| "loss": 0.121, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.617777777777778e-06, | |
| "loss": 0.0606, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.5622222222222226e-06, | |
| "loss": 0.0368, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.5066666666666667e-06, | |
| "loss": 0.0803, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.451111111111111e-06, | |
| "loss": 0.1152, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 2.3955555555555556e-06, | |
| "loss": 0.116, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.3400000000000005e-06, | |
| "loss": 0.0957, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 2.2844444444444445e-06, | |
| "loss": 0.0694, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.228888888888889e-06, | |
| "loss": 0.0997, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "eval_cer": 15.356302345444602, | |
| "eval_loss": 0.38130488991737366, | |
| "eval_runtime": 1220.4825, | |
| "eval_samples_per_second": 8.67, | |
| "eval_steps_per_second": 0.542, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 2.1733333333333334e-06, | |
| "loss": 0.1109, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.0767, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.0792, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.1708, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.0803, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.0376, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.0493, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.1158, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.0724, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.142, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.1388, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.1073, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.0561, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.4511111111111112e-06, | |
| "loss": 0.0457, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.3955555555555556e-06, | |
| "loss": 0.0369, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.34e-06, | |
| "loss": 0.0456, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 1.2844444444444445e-06, | |
| "loss": 0.0329, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.228888888888889e-06, | |
| "loss": 0.1039, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.1733333333333335e-06, | |
| "loss": 0.0642, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.117777777777778e-06, | |
| "loss": 0.0254, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.0622222222222222e-06, | |
| "loss": 0.0349, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 1.0066666666666668e-06, | |
| "loss": 0.0757, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 0.0954, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 8.955555555555557e-07, | |
| "loss": 0.0767, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 0.0631, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 7.844444444444445e-07, | |
| "loss": 0.0733, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 7.28888888888889e-07, | |
| "loss": 0.0846, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 6.733333333333334e-07, | |
| "loss": 0.0636, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 6.177777777777778e-07, | |
| "loss": 0.0703, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 5.622222222222223e-07, | |
| "loss": 0.1078, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 5.066666666666667e-07, | |
| "loss": 0.0973, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 4.511111111111111e-07, | |
| "loss": 0.0382, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.9555555555555557e-07, | |
| "loss": 0.0349, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 0.0639, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.844444444444445e-07, | |
| "loss": 0.0761, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 2.2888888888888892e-07, | |
| "loss": 0.0851, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.7333333333333335e-07, | |
| "loss": 0.1122, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.1777777777777778e-07, | |
| "loss": 0.119, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 6.222222222222223e-08, | |
| "loss": 0.0684, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 6.666666666666667e-09, | |
| "loss": 0.0388, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "eval_cer": 15.509920671496428, | |
| "eval_loss": 0.38818055391311646, | |
| "eval_runtime": 1170.511, | |
| "eval_samples_per_second": 9.04, | |
| "eval_steps_per_second": 0.566, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "step": 5000, | |
| "total_flos": 4.612922189955072e+19, | |
| "train_loss": 0.2561741242170334, | |
| "train_runtime": 24221.1572, | |
| "train_samples_per_second": 6.606, | |
| "train_steps_per_second": 0.206 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 4.612922189955072e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |