| { | |
| "best_metric": 13.030367487894296, | |
| "best_model_checkpoint": "./checkpoint-5000", | |
| "epoch": 2.0911752404851525, | |
| "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": 4.6000000000000004e-07, | |
| "loss": 1.0937, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 0.6164, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.4374, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.3824, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.3665, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.3658, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.3458, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.2992, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.3377, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.3187, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.2772, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.2935, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.2672, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.281, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.2947, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.3013, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.2814, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.2685, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.3129, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.3295, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.94888888888889e-06, | |
| "loss": 0.2646, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.893333333333334e-06, | |
| "loss": 0.2881, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 9.837777777777778e-06, | |
| "loss": 0.2858, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.782222222222222e-06, | |
| "loss": 0.2638, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.726666666666668e-06, | |
| "loss": 0.2462, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 9.671111111111112e-06, | |
| "loss": 0.2591, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.615555555555558e-06, | |
| "loss": 0.2346, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.56e-06, | |
| "loss": 0.2626, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 9.504444444444446e-06, | |
| "loss": 0.2291, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.44888888888889e-06, | |
| "loss": 0.2413, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.393333333333334e-06, | |
| "loss": 0.2875, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.33777777777778e-06, | |
| "loss": 0.242, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.282222222222222e-06, | |
| "loss": 0.25, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.226666666666668e-06, | |
| "loss": 0.2551, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 9.171111111111112e-06, | |
| "loss": 0.2633, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.115555555555556e-06, | |
| "loss": 0.2623, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.060000000000001e-06, | |
| "loss": 0.2225, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.004444444444445e-06, | |
| "loss": 0.2517, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 8.94888888888889e-06, | |
| "loss": 0.2461, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 8.893333333333333e-06, | |
| "loss": 0.2594, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 8.83777777777778e-06, | |
| "loss": 0.2633, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 8.782222222222223e-06, | |
| "loss": 0.2505, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 8.726666666666667e-06, | |
| "loss": 0.2225, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 8.671111111111113e-06, | |
| "loss": 0.2375, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 8.615555555555555e-06, | |
| "loss": 0.2415, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 8.560000000000001e-06, | |
| "loss": 0.2477, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 8.504444444444445e-06, | |
| "loss": 0.2319, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 8.448888888888889e-06, | |
| "loss": 0.2106, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.393333333333335e-06, | |
| "loss": 0.2281, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.337777777777777e-06, | |
| "loss": 0.2141, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.282222222222223e-06, | |
| "loss": 0.2428, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.226666666666667e-06, | |
| "loss": 0.2459, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.171111111111113e-06, | |
| "loss": 0.2341, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.115555555555557e-06, | |
| "loss": 0.2568, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.06e-06, | |
| "loss": 0.2293, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 8.004444444444445e-06, | |
| "loss": 0.2168, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 7.948888888888889e-06, | |
| "loss": 0.2249, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 7.893333333333335e-06, | |
| "loss": 0.1987, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 7.837777777777779e-06, | |
| "loss": 0.2239, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 7.782222222222223e-06, | |
| "loss": 0.2259, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 7.726666666666667e-06, | |
| "loss": 0.2323, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 7.67111111111111e-06, | |
| "loss": 0.1966, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 7.6155555555555564e-06, | |
| "loss": 0.1997, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 7.5600000000000005e-06, | |
| "loss": 0.2391, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 7.504444444444445e-06, | |
| "loss": 0.2003, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 7.44888888888889e-06, | |
| "loss": 0.2257, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 7.393333333333333e-06, | |
| "loss": 0.2334, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 7.337777777777778e-06, | |
| "loss": 0.2213, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.282222222222222e-06, | |
| "loss": 0.2008, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 7.226666666666667e-06, | |
| "loss": 0.2193, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 7.171111111111112e-06, | |
| "loss": 0.2219, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 7.115555555555557e-06, | |
| "loss": 0.245, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 7.06e-06, | |
| "loss": 0.2092, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.004444444444445e-06, | |
| "loss": 0.233, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.948888888888889e-06, | |
| "loss": 0.2324, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.893333333333334e-06, | |
| "loss": 0.1989, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 6.837777777777779e-06, | |
| "loss": 0.2284, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 6.782222222222222e-06, | |
| "loss": 0.1968, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 6.726666666666667e-06, | |
| "loss": 0.1814, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.671111111111112e-06, | |
| "loss": 0.2049, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.615555555555556e-06, | |
| "loss": 0.2183, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.560000000000001e-06, | |
| "loss": 0.198, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.504444444444446e-06, | |
| "loss": 0.1681, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.448888888888889e-06, | |
| "loss": 0.2259, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 6.393333333333334e-06, | |
| "loss": 0.2053, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 6.3377777777777786e-06, | |
| "loss": 0.1947, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 6.282222222222223e-06, | |
| "loss": 0.1854, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 6.2266666666666675e-06, | |
| "loss": 0.2062, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.171111111111112e-06, | |
| "loss": 0.2201, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.1155555555555555e-06, | |
| "loss": 0.2114, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.0600000000000004e-06, | |
| "loss": 0.1697, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 6.004444444444445e-06, | |
| "loss": 0.2189, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 5.948888888888889e-06, | |
| "loss": 0.2028, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.893333333333334e-06, | |
| "loss": 0.1963, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.837777777777777e-06, | |
| "loss": 0.2178, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 5.782222222222222e-06, | |
| "loss": 0.1853, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 5.726666666666667e-06, | |
| "loss": 0.1215, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 5.671111111111112e-06, | |
| "loss": 0.14, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 5.615555555555556e-06, | |
| "loss": 0.1095, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 5.560000000000001e-06, | |
| "loss": 0.1231, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.504444444444444e-06, | |
| "loss": 0.1413, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 5.448888888888889e-06, | |
| "loss": 0.1363, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 5.393333333333334e-06, | |
| "loss": 0.098, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 5.337777777777779e-06, | |
| "loss": 0.1167, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 5.282222222222223e-06, | |
| "loss": 0.1341, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 5.226666666666667e-06, | |
| "loss": 0.1274, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 5.171111111111111e-06, | |
| "loss": 0.1197, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 5.115555555555556e-06, | |
| "loss": 0.1328, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 5.060000000000001e-06, | |
| "loss": 0.1377, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 5.004444444444445e-06, | |
| "loss": 0.1253, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 4.94888888888889e-06, | |
| "loss": 0.116, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.893333333333334e-06, | |
| "loss": 0.1136, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.837777777777778e-06, | |
| "loss": 0.1128, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.7822222222222226e-06, | |
| "loss": 0.1372, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 4.7266666666666674e-06, | |
| "loss": 0.1437, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 4.6711111111111115e-06, | |
| "loss": 0.1215, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.6155555555555555e-06, | |
| "loss": 0.1211, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 4.56e-06, | |
| "loss": 0.0987, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 4.504444444444444e-06, | |
| "loss": 0.1295, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 4.448888888888889e-06, | |
| "loss": 0.1267, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 4.393333333333334e-06, | |
| "loss": 0.1065, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 4.337777777777778e-06, | |
| "loss": 0.1143, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.282222222222222e-06, | |
| "loss": 0.1341, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 4.226666666666667e-06, | |
| "loss": 0.1318, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.171111111111111e-06, | |
| "loss": 0.115, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.115555555555556e-06, | |
| "loss": 0.1168, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 4.060000000000001e-06, | |
| "loss": 0.1261, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.004444444444445e-06, | |
| "loss": 0.1148, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.948888888888889e-06, | |
| "loss": 0.1255, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.893333333333333e-06, | |
| "loss": 0.1236, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.837777777777778e-06, | |
| "loss": 0.1331, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.782222222222223e-06, | |
| "loss": 0.1225, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 3.726666666666667e-06, | |
| "loss": 0.1087, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 3.6711111111111113e-06, | |
| "loss": 0.1319, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 3.615555555555556e-06, | |
| "loss": 0.1165, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 3.5600000000000002e-06, | |
| "loss": 0.1217, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 3.5044444444444447e-06, | |
| "loss": 0.1138, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.4488888888888896e-06, | |
| "loss": 0.1192, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.3933333333333336e-06, | |
| "loss": 0.1056, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.337777777777778e-06, | |
| "loss": 0.1129, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 3.282222222222223e-06, | |
| "loss": 0.1108, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.226666666666667e-06, | |
| "loss": 0.1121, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.1711111111111114e-06, | |
| "loss": 0.1036, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 3.1155555555555555e-06, | |
| "loss": 0.1334, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 3.0600000000000003e-06, | |
| "loss": 0.1208, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 3.004444444444445e-06, | |
| "loss": 0.1061, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.948888888888889e-06, | |
| "loss": 0.1181, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.8933333333333337e-06, | |
| "loss": 0.1098, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.837777777777778e-06, | |
| "loss": 0.1096, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.7822222222222222e-06, | |
| "loss": 0.1231, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.726666666666667e-06, | |
| "loss": 0.1038, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.6711111111111116e-06, | |
| "loss": 0.1171, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.6155555555555556e-06, | |
| "loss": 0.1019, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.56e-06, | |
| "loss": 0.1218, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.504444444444445e-06, | |
| "loss": 0.114, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.448888888888889e-06, | |
| "loss": 0.1159, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.3933333333333334e-06, | |
| "loss": 0.1204, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.337777777777778e-06, | |
| "loss": 0.1026, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.2822222222222223e-06, | |
| "loss": 0.0923, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.226666666666667e-06, | |
| "loss": 0.1151, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.1711111111111113e-06, | |
| "loss": 0.1156, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.117777777777778e-06, | |
| "loss": 0.1015, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.0622222222222223e-06, | |
| "loss": 0.0947, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.006666666666667e-06, | |
| "loss": 0.1138, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.9511111111111113e-06, | |
| "loss": 0.1035, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.8955555555555557e-06, | |
| "loss": 0.1032, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.8400000000000002e-06, | |
| "loss": 0.1112, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 1.7844444444444444e-06, | |
| "loss": 0.1043, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.728888888888889e-06, | |
| "loss": 0.1243, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.6733333333333335e-06, | |
| "loss": 0.1282, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.6177777777777778e-06, | |
| "loss": 0.1043, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 1.5622222222222225e-06, | |
| "loss": 0.1121, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.506666666666667e-06, | |
| "loss": 0.1197, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.4511111111111112e-06, | |
| "loss": 0.109, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.3955555555555556e-06, | |
| "loss": 0.104, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.34e-06, | |
| "loss": 0.1255, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.2844444444444445e-06, | |
| "loss": 0.1004, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.228888888888889e-06, | |
| "loss": 0.1143, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.1733333333333335e-06, | |
| "loss": 0.1105, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.117777777777778e-06, | |
| "loss": 0.1139, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.0622222222222222e-06, | |
| "loss": 0.1048, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.0066666666666668e-06, | |
| "loss": 0.1111, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.511111111111111e-07, | |
| "loss": 0.1105, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 8.955555555555557e-07, | |
| "loss": 0.1084, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 8.400000000000001e-07, | |
| "loss": 0.111, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 7.844444444444445e-07, | |
| "loss": 0.0917, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 7.28888888888889e-07, | |
| "loss": 0.115, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.733333333333334e-07, | |
| "loss": 0.0945, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.177777777777778e-07, | |
| "loss": 0.1023, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 5.622222222222223e-07, | |
| "loss": 0.1107, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.066666666666667e-07, | |
| "loss": 0.1089, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 4.511111111111111e-07, | |
| "loss": 0.0746, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 3.9555555555555557e-07, | |
| "loss": 0.0789, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 3.4000000000000003e-07, | |
| "loss": 0.0705, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 2.844444444444445e-07, | |
| "loss": 0.0572, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.2888888888888892e-07, | |
| "loss": 0.0789, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.7333333333333335e-07, | |
| "loss": 0.0663, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.1777777777777778e-07, | |
| "loss": 0.0686, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.222222222222223e-08, | |
| "loss": 0.076, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.666666666666667e-09, | |
| "loss": 0.0611, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "eval_loss": 0.14747175574302673, | |
| "eval_runtime": 10667.3909, | |
| "eval_samples_per_second": 1.025, | |
| "eval_steps_per_second": 0.032, | |
| "eval_wer": 13.030367487894296, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "step": 5000, | |
| "total_flos": 1.6325800221671424e+20, | |
| "train_loss": 0.18211641297340392, | |
| "train_runtime": 28211.9988, | |
| "train_samples_per_second": 5.671, | |
| "train_steps_per_second": 0.177 | |
| } | |
| ], | |
| "max_steps": 5000, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.6325800221671424e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |