| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.0532, | |
| "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": 4.0000000000000003e-07, | |
| "loss": 2.3624, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 8.8e-07, | |
| "loss": 2.1188, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.3800000000000001e-06, | |
| "loss": 1.6209, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.8800000000000002e-06, | |
| "loss": 1.3116, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.38e-06, | |
| "loss": 1.2712, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.88e-06, | |
| "loss": 1.3543, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.3800000000000007e-06, | |
| "loss": 1.2611, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 3.86e-06, | |
| "loss": 1.0402, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.360000000000001e-06, | |
| "loss": 1.1532, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.86e-06, | |
| "loss": 1.3084, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.36e-06, | |
| "loss": 1.2273, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 5.86e-06, | |
| "loss": 1.0857, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.360000000000001e-06, | |
| "loss": 1.2258, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 6.860000000000001e-06, | |
| "loss": 1.2328, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.360000000000001e-06, | |
| "loss": 1.213, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 7.860000000000001e-06, | |
| "loss": 1.3364, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.36e-06, | |
| "loss": 1.1307, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 8.860000000000002e-06, | |
| "loss": 1.2142, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.360000000000002e-06, | |
| "loss": 1.2554, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.86e-06, | |
| "loss": 1.453, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 1.0811, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.904444444444445e-06, | |
| "loss": 1.4234, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.84888888888889e-06, | |
| "loss": 1.1177, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.793333333333333e-06, | |
| "loss": 1.2537, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.737777777777779e-06, | |
| "loss": 1.4309, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.682222222222223e-06, | |
| "loss": 1.1939, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.626666666666667e-06, | |
| "loss": 1.4469, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.571111111111113e-06, | |
| "loss": 1.1278, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.515555555555557e-06, | |
| "loss": 1.2459, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.9669, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.404444444444445e-06, | |
| "loss": 1.0934, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 9.348888888888889e-06, | |
| "loss": 0.8223, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.293333333333335e-06, | |
| "loss": 0.8306, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 9.237777777777779e-06, | |
| "loss": 0.8573, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.182222222222223e-06, | |
| "loss": 0.818, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 9.126666666666667e-06, | |
| "loss": 0.7734, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.07111111111111e-06, | |
| "loss": 0.8211, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 9.015555555555557e-06, | |
| "loss": 0.7027, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.7397, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.904444444444446e-06, | |
| "loss": 0.6011, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "eval_cer": 55.20676124358588, | |
| "eval_loss": 1.1100292205810547, | |
| "eval_runtime": 311.6884, | |
| "eval_samples_per_second": 3.163, | |
| "eval_steps_per_second": 1.582, | |
| "eval_wer": 55.022900763358784, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.848888888888889e-06, | |
| "loss": 0.7035, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 8.793333333333334e-06, | |
| "loss": 0.5823, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.737777777777778e-06, | |
| "loss": 0.6668, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.682222222222222e-06, | |
| "loss": 0.7825, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.626666666666668e-06, | |
| "loss": 0.5401, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 8.571111111111112e-06, | |
| "loss": 0.687, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.515555555555556e-06, | |
| "loss": 0.557, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.5952, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.404444444444444e-06, | |
| "loss": 0.6319, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.34888888888889e-06, | |
| "loss": 0.6004, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.293333333333334e-06, | |
| "loss": 0.6284, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 8.237777777777778e-06, | |
| "loss": 0.6487, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.182222222222222e-06, | |
| "loss": 0.397, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.126666666666668e-06, | |
| "loss": 0.6252, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.071111111111112e-06, | |
| "loss": 0.4981, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 8.015555555555556e-06, | |
| "loss": 0.4106, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.385, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.904444444444444e-06, | |
| "loss": 0.5597, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.84888888888889e-06, | |
| "loss": 0.4176, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.793333333333334e-06, | |
| "loss": 0.5235, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.737777777777778e-06, | |
| "loss": 0.6513, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.682222222222224e-06, | |
| "loss": 0.514, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.626666666666668e-06, | |
| "loss": 0.549, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.571111111111112e-06, | |
| "loss": 0.2948, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.515555555555556e-06, | |
| "loss": 0.1904, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.2632, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.4044444444444455e-06, | |
| "loss": 0.2815, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.3488888888888895e-06, | |
| "loss": 0.1711, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.2933333333333335e-06, | |
| "loss": 0.2852, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 7.237777777777778e-06, | |
| "loss": 0.2344, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.1822222222222224e-06, | |
| "loss": 0.216, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.126666666666667e-06, | |
| "loss": 0.1479, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.071111111111112e-06, | |
| "loss": 0.2791, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.015555555555556e-06, | |
| "loss": 0.1207, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.1538, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.904444444444444e-06, | |
| "loss": 0.1898, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.848888888888889e-06, | |
| "loss": 0.1789, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.793333333333334e-06, | |
| "loss": 0.1941, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.737777777777779e-06, | |
| "loss": 0.2072, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.682222222222223e-06, | |
| "loss": 0.1773, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "eval_cer": 58.768487775430124, | |
| "eval_loss": 1.2054609060287476, | |
| "eval_runtime": 316.5335, | |
| "eval_samples_per_second": 3.115, | |
| "eval_steps_per_second": 1.557, | |
| "eval_wer": 58.656488549618324, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.626666666666667e-06, | |
| "loss": 0.2132, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.571111111111111e-06, | |
| "loss": 0.241, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.515555555555556e-06, | |
| "loss": 0.1587, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.1256, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.404444444444446e-06, | |
| "loss": 0.1353, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 6.348888888888889e-06, | |
| "loss": 0.105, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.293333333333334e-06, | |
| "loss": 0.1043, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.237777777777778e-06, | |
| "loss": 0.2282, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.182222222222223e-06, | |
| "loss": 0.1383, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 6.126666666666668e-06, | |
| "loss": 0.2288, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.0711111111111125e-06, | |
| "loss": 0.188, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 6.015555555555556e-06, | |
| "loss": 0.1187, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.1893, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.9044444444444446e-06, | |
| "loss": 0.1996, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 5.8488888888888895e-06, | |
| "loss": 0.1588, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 5.793333333333334e-06, | |
| "loss": 0.05, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 5.737777777777778e-06, | |
| "loss": 0.0634, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 5.682222222222222e-06, | |
| "loss": 0.0964, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 5.626666666666667e-06, | |
| "loss": 0.0753, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.571111111111111e-06, | |
| "loss": 0.0634, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.515555555555556e-06, | |
| "loss": 0.0248, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.0698, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.404444444444444e-06, | |
| "loss": 0.0496, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.348888888888889e-06, | |
| "loss": 0.0641, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 5.293333333333334e-06, | |
| "loss": 0.0718, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.237777777777778e-06, | |
| "loss": 0.0129, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.182222222222223e-06, | |
| "loss": 0.0223, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.126666666666668e-06, | |
| "loss": 0.0181, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 5.071111111111111e-06, | |
| "loss": 0.0608, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.015555555555556e-06, | |
| "loss": 0.0298, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.0335, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 4.904444444444445e-06, | |
| "loss": 0.0556, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 4.848888888888889e-06, | |
| "loss": 0.0338, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.793333333333334e-06, | |
| "loss": 0.0627, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 4.737777777777779e-06, | |
| "loss": 0.0272, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.682222222222223e-06, | |
| "loss": 0.0332, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 4.626666666666667e-06, | |
| "loss": 0.0464, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.571111111111112e-06, | |
| "loss": 0.0433, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 4.515555555555556e-06, | |
| "loss": 0.0413, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.015, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "eval_cer": 48.80772713552671, | |
| "eval_loss": 1.093173861503601, | |
| "eval_runtime": 300.4137, | |
| "eval_samples_per_second": 3.282, | |
| "eval_steps_per_second": 1.641, | |
| "eval_wer": 48.6412213740458, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 4.404444444444445e-06, | |
| "loss": 0.0324, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.348888888888889e-06, | |
| "loss": 0.037, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 4.2933333333333334e-06, | |
| "loss": 0.0418, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.2377777777777775e-06, | |
| "loss": 0.0705, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 4.182222222222222e-06, | |
| "loss": 0.0239, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 4.126666666666667e-06, | |
| "loss": 0.0386, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.071111111111111e-06, | |
| "loss": 0.0222, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 4.015555555555556e-06, | |
| "loss": 0.0083, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.012, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.904444444444444e-06, | |
| "loss": 0.0263, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.848888888888889e-06, | |
| "loss": 0.0062, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.793333333333334e-06, | |
| "loss": 0.0526, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 3.737777777777778e-06, | |
| "loss": 0.0074, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.6822222222222225e-06, | |
| "loss": 0.0063, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 3.6266666666666674e-06, | |
| "loss": 0.0062, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.5711111111111114e-06, | |
| "loss": 0.0288, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 3.515555555555556e-06, | |
| "loss": 0.0152, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.0073, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 3.404444444444445e-06, | |
| "loss": 0.0167, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.3488888888888892e-06, | |
| "loss": 0.0044, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 3.2933333333333333e-06, | |
| "loss": 0.0225, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.237777777777778e-06, | |
| "loss": 0.0068, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 3.1822222222222226e-06, | |
| "loss": 0.0101, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.1266666666666667e-06, | |
| "loss": 0.0028, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 3.0711111111111115e-06, | |
| "loss": 0.0041, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 3.015555555555556e-06, | |
| "loss": 0.0048, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.003, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 2.904444444444445e-06, | |
| "loss": 0.0085, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 2.8488888888888894e-06, | |
| "loss": 0.0176, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 2.7933333333333334e-06, | |
| "loss": 0.006, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 2.7377777777777783e-06, | |
| "loss": 0.0028, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 2.6822222222222223e-06, | |
| "loss": 0.0098, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 2.6266666666666668e-06, | |
| "loss": 0.015, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 2.5711111111111112e-06, | |
| "loss": 0.0118, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 2.5155555555555557e-06, | |
| "loss": 0.0043, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.0429, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 2.4044444444444446e-06, | |
| "loss": 0.0011, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 2.348888888888889e-06, | |
| "loss": 0.0031, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.2933333333333335e-06, | |
| "loss": 0.0011, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.237777777777778e-06, | |
| "loss": 0.0131, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "eval_cer": 45.849683066706916, | |
| "eval_loss": 1.0531306266784668, | |
| "eval_runtime": 303.983, | |
| "eval_samples_per_second": 3.244, | |
| "eval_steps_per_second": 1.622, | |
| "eval_wer": 45.70992366412214, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 2.1822222222222225e-06, | |
| "loss": 0.0027, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 2.126666666666667e-06, | |
| "loss": 0.0003, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.0711111111111114e-06, | |
| "loss": 0.0008, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.0155555555555554e-06, | |
| "loss": 0.0004, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.0003, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 1.9044444444444445e-06, | |
| "loss": 0.0094, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 1.848888888888889e-06, | |
| "loss": 0.0003, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 1.7933333333333337e-06, | |
| "loss": 0.0029, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 1.737777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 1.6822222222222224e-06, | |
| "loss": 0.012, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 1.6266666666666666e-06, | |
| "loss": 0.0002, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 1.5711111111111113e-06, | |
| "loss": 0.0002, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 1.5155555555555558e-06, | |
| "loss": 0.0007, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.0003, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 1.4044444444444447e-06, | |
| "loss": 0.0002, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 1.3488888888888891e-06, | |
| "loss": 0.0003, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 1.2933333333333334e-06, | |
| "loss": 0.0002, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 1.2377777777777778e-06, | |
| "loss": 0.0009, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.1822222222222223e-06, | |
| "loss": 0.0002, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 1.1266666666666667e-06, | |
| "loss": 0.0059, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 1.0711111111111112e-06, | |
| "loss": 0.0003, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 1.0155555555555557e-06, | |
| "loss": 0.0022, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 0.0003, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 9.044444444444445e-07, | |
| "loss": 0.0005, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 8.488888888888889e-07, | |
| "loss": 0.0002, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 7.933333333333335e-07, | |
| "loss": 0.0034, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 7.377777777777779e-07, | |
| "loss": 0.0075, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 6.822222222222223e-07, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 6.266666666666667e-07, | |
| "loss": 0.0004, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 5.711111111111111e-07, | |
| "loss": 0.0007, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 5.155555555555556e-07, | |
| "loss": 0.0002, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 0.0001, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 4.0444444444444445e-07, | |
| "loss": 0.0003, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 3.488888888888889e-07, | |
| "loss": 0.0001, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.9333333333333337e-07, | |
| "loss": 0.0002, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 2.3777777777777777e-07, | |
| "loss": 0.0001, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.8222222222222226e-07, | |
| "loss": 0.0002, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 1.2666666666666666e-07, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 7.111111111111112e-08, | |
| "loss": 0.0002, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 1.5555555555555557e-08, | |
| "loss": 0.0001, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "eval_cer": 45.45728946574102, | |
| "eval_loss": 1.0350826978683472, | |
| "eval_runtime": 305.8919, | |
| "eval_samples_per_second": 3.223, | |
| "eval_steps_per_second": 1.612, | |
| "eval_wer": 45.37404580152672, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "step": 5000, | |
| "total_flos": 2.12189711818752e+19, | |
| "train_loss": 0.3433635727233719, | |
| "train_runtime": 5460.4494, | |
| "train_samples_per_second": 1.831, | |
| "train_steps_per_second": 0.916 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 2.12189711818752e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |