| { | |
| "best_metric": 0.22282175719738007, | |
| "best_model_checkpoint": "whisper-base-full-data-v4/checkpoint-60000", | |
| "epoch": 20.003133322888925, | |
| "global_step": 63840, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 2.0000000000000002e-07, | |
| "loss": 2.2652, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 1.8464, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 1.2569, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.8772, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.6906, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.6237, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.5784, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.6000000000000001e-06, | |
| "loss": 0.5455, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.8000000000000001e-06, | |
| "loss": 0.5264, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.5018, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 2.2e-06, | |
| "loss": 0.4829, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.4635, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.6e-06, | |
| "loss": 0.4505, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.8000000000000003e-06, | |
| "loss": 0.4415, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3e-06, | |
| "loss": 0.4276, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.2000000000000003e-06, | |
| "loss": 0.4168, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.4074, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.6000000000000003e-06, | |
| "loss": 0.3999, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.8000000000000005e-06, | |
| "loss": 0.3904, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.3819, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.2000000000000004e-06, | |
| "loss": 0.3717, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.3635, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.600000000000001e-06, | |
| "loss": 0.3586, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.800000000000001e-06, | |
| "loss": 0.3596, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 5e-06, | |
| "loss": 0.3483, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.2e-06, | |
| "loss": 0.3465, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.3389, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.600000000000001e-06, | |
| "loss": 0.3361, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.8e-06, | |
| "loss": 0.3271, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6e-06, | |
| "loss": 0.3221, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 6.200000000000001e-06, | |
| "loss": 0.3173, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.3169, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 6.600000000000001e-06, | |
| "loss": 0.3099, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 6.800000000000001e-06, | |
| "loss": 0.3025, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 7e-06, | |
| "loss": 0.2969, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.2000000000000005e-06, | |
| "loss": 0.2949, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.2907, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.600000000000001e-06, | |
| "loss": 0.2914, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.800000000000002e-06, | |
| "loss": 0.2859, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2829, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 8.2e-06, | |
| "loss": 0.2824, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.2762, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 8.6e-06, | |
| "loss": 0.2739, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 8.8e-06, | |
| "loss": 0.2709, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 9e-06, | |
| "loss": 0.2712, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.200000000000002e-06, | |
| "loss": 0.2695, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.2689, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.600000000000001e-06, | |
| "loss": 0.263, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.800000000000001e-06, | |
| "loss": 0.2583, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 1e-05, | |
| "loss": 0.2569, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "eval_loss": 0.4137764871120453, | |
| "eval_runtime": 73.4793, | |
| "eval_samples_per_second": 121.681, | |
| "eval_steps_per_second": 1.905, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.983004758667573e-06, | |
| "loss": 0.2546, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.966009517335146e-06, | |
| "loss": 0.2544, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 9.94901427600272e-06, | |
| "loss": 0.2497, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 9.932019034670293e-06, | |
| "loss": 0.2491, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.915023793337866e-06, | |
| "loss": 0.2468, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 9.89802855200544e-06, | |
| "loss": 0.2431, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 9.881033310673012e-06, | |
| "loss": 0.2407, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 9.864038069340585e-06, | |
| "loss": 0.2398, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.84704282800816e-06, | |
| "loss": 0.2389, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 9.830047586675732e-06, | |
| "loss": 0.2318, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 9.813052345343305e-06, | |
| "loss": 0.2347, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 9.796057104010877e-06, | |
| "loss": 0.2325, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 9.779061862678451e-06, | |
| "loss": 0.2278, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 9.762066621346024e-06, | |
| "loss": 0.2295, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 9.745071380013597e-06, | |
| "loss": 0.2215, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.728076138681171e-06, | |
| "loss": 0.2177, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 9.711080897348744e-06, | |
| "loss": 0.219, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 9.694085656016316e-06, | |
| "loss": 0.2159, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 9.677090414683889e-06, | |
| "loss": 0.2129, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 9.660095173351463e-06, | |
| "loss": 0.215, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 9.643099932019036e-06, | |
| "loss": 0.2113, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 9.626104690686608e-06, | |
| "loss": 0.2134, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 9.60910944935418e-06, | |
| "loss": 0.2096, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 9.592114208021755e-06, | |
| "loss": 0.2087, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 9.575118966689328e-06, | |
| "loss": 0.2056, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 9.5581237253569e-06, | |
| "loss": 0.2069, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.541128484024475e-06, | |
| "loss": 0.2067, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.524133242692047e-06, | |
| "loss": 0.2057, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.50713800135962e-06, | |
| "loss": 0.2019, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 9.490142760027194e-06, | |
| "loss": 0.203, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 9.473147518694767e-06, | |
| "loss": 0.2025, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 9.45615227736234e-06, | |
| "loss": 0.2002, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 9.439157036029912e-06, | |
| "loss": 0.197, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 9.422161794697484e-06, | |
| "loss": 0.2002, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 9.405166553365059e-06, | |
| "loss": 0.1953, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 9.388171312032631e-06, | |
| "loss": 0.1974, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 9.371176070700204e-06, | |
| "loss": 0.1923, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 9.354180829367778e-06, | |
| "loss": 0.1936, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 9.337185588035351e-06, | |
| "loss": 0.1934, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 9.320190346702923e-06, | |
| "loss": 0.1951, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 9.303195105370498e-06, | |
| "loss": 0.1926, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 9.28619986403807e-06, | |
| "loss": 0.1926, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 9.269204622705643e-06, | |
| "loss": 0.1899, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 9.252209381373216e-06, | |
| "loss": 0.1896, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 9.23521414004079e-06, | |
| "loss": 0.187, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 9.218218898708362e-06, | |
| "loss": 0.1861, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 9.201223657375935e-06, | |
| "loss": 0.1801, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.18422841604351e-06, | |
| "loss": 0.1801, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 9.167233174711082e-06, | |
| "loss": 0.1808, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 9.150237933378655e-06, | |
| "loss": 0.1774, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "eval_loss": 0.32951679825782776, | |
| "eval_runtime": 64.7018, | |
| "eval_samples_per_second": 138.188, | |
| "eval_steps_per_second": 2.164, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 9.133242692046229e-06, | |
| "loss": 0.1785, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 9.116247450713801e-06, | |
| "loss": 0.1758, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 9.099252209381374e-06, | |
| "loss": 0.179, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 9.082256968048947e-06, | |
| "loss": 0.1761, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 9.06526172671652e-06, | |
| "loss": 0.1756, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 9.048266485384094e-06, | |
| "loss": 0.1766, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 9.031271244051666e-06, | |
| "loss": 0.1782, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 9.014276002719239e-06, | |
| "loss": 0.1753, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 8.997280761386813e-06, | |
| "loss": 0.175, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 8.980285520054386e-06, | |
| "loss": 0.1743, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 8.963290278721958e-06, | |
| "loss": 0.1705, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 8.946295037389533e-06, | |
| "loss": 0.1722, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 8.929299796057105e-06, | |
| "loss": 0.1685, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 8.912304554724678e-06, | |
| "loss": 0.1689, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 8.89530931339225e-06, | |
| "loss": 0.1675, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 8.878314072059823e-06, | |
| "loss": 0.1706, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 8.861318830727397e-06, | |
| "loss": 0.1683, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 8.84432358939497e-06, | |
| "loss": 0.169, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 8.827328348062542e-06, | |
| "loss": 0.1675, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 8.810333106730117e-06, | |
| "loss": 0.1686, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 8.79333786539769e-06, | |
| "loss": 0.167, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 8.776342624065262e-06, | |
| "loss": 0.1673, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 8.759347382732836e-06, | |
| "loss": 0.168, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 8.742352141400409e-06, | |
| "loss": 0.1642, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 8.725356900067981e-06, | |
| "loss": 0.1661, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 8.708361658735554e-06, | |
| "loss": 0.1645, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 8.691366417403128e-06, | |
| "loss": 0.1645, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 8.674371176070701e-06, | |
| "loss": 0.1607, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 8.657375934738274e-06, | |
| "loss": 0.1595, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 8.640380693405848e-06, | |
| "loss": 0.1582, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 8.62338545207342e-06, | |
| "loss": 0.1592, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 8.606390210740993e-06, | |
| "loss": 0.1564, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 8.589394969408567e-06, | |
| "loss": 0.158, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 8.57239972807614e-06, | |
| "loss": 0.156, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 8.555404486743713e-06, | |
| "loss": 0.1542, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 8.538409245411285e-06, | |
| "loss": 0.1551, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 8.521414004078858e-06, | |
| "loss": 0.1558, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 8.504418762746432e-06, | |
| "loss": 0.1531, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 8.487423521414005e-06, | |
| "loss": 0.1548, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 8.470428280081577e-06, | |
| "loss": 0.1554, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 8.453433038749152e-06, | |
| "loss": 0.1555, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 8.436437797416724e-06, | |
| "loss": 0.1532, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 8.419442556084297e-06, | |
| "loss": 0.1521, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 8.402447314751871e-06, | |
| "loss": 0.1543, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 8.385452073419444e-06, | |
| "loss": 0.1519, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 8.368456832087016e-06, | |
| "loss": 0.1536, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 8.351461590754589e-06, | |
| "loss": 0.1518, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 8.334466349422161e-06, | |
| "loss": 0.1516, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 8.317471108089736e-06, | |
| "loss": 0.1502, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 8.300475866757308e-06, | |
| "loss": 0.1485, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "eval_loss": 0.29357945919036865, | |
| "eval_runtime": 82.3606, | |
| "eval_samples_per_second": 108.559, | |
| "eval_steps_per_second": 1.7, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 8.283480625424881e-06, | |
| "loss": 0.1515, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 8.266485384092455e-06, | |
| "loss": 0.1488, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 8.249490142760028e-06, | |
| "loss": 0.1504, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 8.2324949014276e-06, | |
| "loss": 0.1449, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 8.215499660095175e-06, | |
| "loss": 0.1504, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 8.198504418762747e-06, | |
| "loss": 0.1472, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 8.18150917743032e-06, | |
| "loss": 0.1502, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 8.164513936097893e-06, | |
| "loss": 0.1461, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 8.147518694765467e-06, | |
| "loss": 0.1475, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 8.13052345343304e-06, | |
| "loss": 0.1449, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 8.113528212100612e-06, | |
| "loss": 0.1402, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 8.096532970768186e-06, | |
| "loss": 0.1423, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 8.079537729435759e-06, | |
| "loss": 0.1402, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 8.062542488103332e-06, | |
| "loss": 0.1407, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 8.045547246770906e-06, | |
| "loss": 0.1408, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 8.028552005438478e-06, | |
| "loss": 0.1415, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 8.011556764106051e-06, | |
| "loss": 0.1405, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 7.994561522773624e-06, | |
| "loss": 0.141, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 7.977566281441196e-06, | |
| "loss": 0.1423, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 7.96057104010877e-06, | |
| "loss": 0.144, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 7.943575798776343e-06, | |
| "loss": 0.1394, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 7.926580557443916e-06, | |
| "loss": 0.1396, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 7.90958531611149e-06, | |
| "loss": 0.1393, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 7.892590074779063e-06, | |
| "loss": 0.138, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 7.875594833446635e-06, | |
| "loss": 0.1393, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 7.85859959211421e-06, | |
| "loss": 0.1403, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 7.841604350781782e-06, | |
| "loss": 0.1357, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 7.824609109449355e-06, | |
| "loss": 0.1368, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 7.807613868116927e-06, | |
| "loss": 0.1371, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 7.790618626784502e-06, | |
| "loss": 0.1384, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 7.773623385452074e-06, | |
| "loss": 0.1356, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 7.756628144119647e-06, | |
| "loss": 0.1378, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 5.73, | |
| "learning_rate": 7.73963290278722e-06, | |
| "loss": 0.1375, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 7.722637661454794e-06, | |
| "loss": 0.1377, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 7.705642420122366e-06, | |
| "loss": 0.1365, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 7.688647178789939e-06, | |
| "loss": 0.138, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 7.671651937457513e-06, | |
| "loss": 0.1344, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 7.654656696125086e-06, | |
| "loss": 0.1328, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 7.637661454792658e-06, | |
| "loss": 0.1363, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 7.620666213460232e-06, | |
| "loss": 0.1366, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 5.98, | |
| "learning_rate": 7.6036709721278044e-06, | |
| "loss": 0.1356, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 7.586675730795378e-06, | |
| "loss": 0.1326, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 6.05, | |
| "learning_rate": 7.5696804894629505e-06, | |
| "loss": 0.1288, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 7.552685248130525e-06, | |
| "loss": 0.1304, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 7.535690006798097e-06, | |
| "loss": 0.1287, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 7.518694765465671e-06, | |
| "loss": 0.13, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 7.5016995241332434e-06, | |
| "loss": 0.127, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 7.484704282800817e-06, | |
| "loss": 0.1301, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 7.4677090414683895e-06, | |
| "loss": 0.1295, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 7.450713800135962e-06, | |
| "loss": 0.1316, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "eval_loss": 0.27446937561035156, | |
| "eval_runtime": 83.4787, | |
| "eval_samples_per_second": 107.105, | |
| "eval_steps_per_second": 1.677, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 7.4337185588035355e-06, | |
| "loss": 0.1278, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 7.416723317471108e-06, | |
| "loss": 0.1273, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 7.399728076138682e-06, | |
| "loss": 0.1309, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 7.382732834806254e-06, | |
| "loss": 0.127, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 7.3657375934738285e-06, | |
| "loss": 0.1278, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 7.348742352141401e-06, | |
| "loss": 0.1286, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 7.3317471108089745e-06, | |
| "loss": 0.1284, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 7.314751869476547e-06, | |
| "loss": 0.1289, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 7.297756628144121e-06, | |
| "loss": 0.1278, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 7.280761386811693e-06, | |
| "loss": 0.1271, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 7.263766145479267e-06, | |
| "loss": 0.1263, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 7.246770904146839e-06, | |
| "loss": 0.1261, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 7.229775662814412e-06, | |
| "loss": 0.1304, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 7.212780421481985e-06, | |
| "loss": 0.1264, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 7.195785180149558e-06, | |
| "loss": 0.1264, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 7.178789938817132e-06, | |
| "loss": 0.1264, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 7.161794697484706e-06, | |
| "loss": 0.1255, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 7.144799456152278e-06, | |
| "loss": 0.1263, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 7.127804214819851e-06, | |
| "loss": 0.1286, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 7.110808973487424e-06, | |
| "loss": 0.1248, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 6.92, | |
| "learning_rate": 7.093813732154997e-06, | |
| "loss": 0.1254, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 7.07681849082257e-06, | |
| "loss": 0.1235, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 6.99, | |
| "learning_rate": 7.059823249490143e-06, | |
| "loss": 0.1275, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 7.042828008157716e-06, | |
| "loss": 0.1222, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 7.025832766825289e-06, | |
| "loss": 0.1198, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.08, | |
| "learning_rate": 7.008837525492863e-06, | |
| "loss": 0.1207, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 6.991842284160436e-06, | |
| "loss": 0.1174, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 6.974847042828009e-06, | |
| "loss": 0.1203, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 6.957851801495582e-06, | |
| "loss": 0.1224, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 6.940856560163155e-06, | |
| "loss": 0.1171, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 7.24, | |
| "learning_rate": 6.923861318830728e-06, | |
| "loss": 0.119, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 6.9068660774983006e-06, | |
| "loss": 0.1211, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 6.889870836165874e-06, | |
| "loss": 0.1204, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 6.872875594833447e-06, | |
| "loss": 0.1174, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 7.36, | |
| "learning_rate": 6.85588035350102e-06, | |
| "loss": 0.1179, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 6.838885112168593e-06, | |
| "loss": 0.121, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 6.821889870836167e-06, | |
| "loss": 0.1195, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 6.8048946295037396e-06, | |
| "loss": 0.1188, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 6.787899388171313e-06, | |
| "loss": 0.119, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 6.770904146838886e-06, | |
| "loss": 0.1194, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 6.753908905506459e-06, | |
| "loss": 0.1185, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 6.736913664174032e-06, | |
| "loss": 0.1185, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 6.719918422841605e-06, | |
| "loss": 0.1185, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 6.702923181509178e-06, | |
| "loss": 0.1189, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 6.685927940176751e-06, | |
| "loss": 0.1189, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 6.668932698844324e-06, | |
| "loss": 0.1191, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 6.651937457511896e-06, | |
| "loss": 0.1189, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 7.77, | |
| "learning_rate": 6.634942216179471e-06, | |
| "loss": 0.1201, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 6.617946974847044e-06, | |
| "loss": 0.1192, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 6.600951733514617e-06, | |
| "loss": 0.1187, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "eval_loss": 0.2598571479320526, | |
| "eval_runtime": 81.4111, | |
| "eval_samples_per_second": 109.825, | |
| "eval_steps_per_second": 1.72, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 7.86, | |
| "learning_rate": 6.583956492182189e-06, | |
| "loss": 0.1201, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 6.566961250849763e-06, | |
| "loss": 0.1177, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 6.549966009517335e-06, | |
| "loss": 0.119, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 6.532970768184909e-06, | |
| "loss": 0.1192, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 6.515975526852481e-06, | |
| "loss": 0.1155, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 6.498980285520055e-06, | |
| "loss": 0.1127, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 8.05, | |
| "learning_rate": 6.4819850441876274e-06, | |
| "loss": 0.1126, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 6.464989802855202e-06, | |
| "loss": 0.1129, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 8.12, | |
| "learning_rate": 6.447994561522774e-06, | |
| "loss": 0.1138, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 6.430999320190348e-06, | |
| "loss": 0.1124, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 6.41400407885792e-06, | |
| "loss": 0.1113, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 8.21, | |
| "learning_rate": 6.397008837525494e-06, | |
| "loss": 0.1125, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 6.3800135961930664e-06, | |
| "loss": 0.1126, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 6.36301835486064e-06, | |
| "loss": 0.1136, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 6.3460231135282125e-06, | |
| "loss": 0.1137, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 6.329027872195785e-06, | |
| "loss": 0.1136, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 6.3120326308633585e-06, | |
| "loss": 0.1108, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 6.295037389530931e-06, | |
| "loss": 0.1117, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 6.278042148198505e-06, | |
| "loss": 0.1114, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 8.46, | |
| "learning_rate": 6.261046906866078e-06, | |
| "loss": 0.1135, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 6.2440516655336515e-06, | |
| "loss": 0.1104, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 6.227056424201224e-06, | |
| "loss": 0.1122, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 6.2100611828687975e-06, | |
| "loss": 0.111, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 6.19306594153637e-06, | |
| "loss": 0.1118, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 6.1760707002039436e-06, | |
| "loss": 0.1117, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 6.159075458871516e-06, | |
| "loss": 0.1102, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 6.14208021753909e-06, | |
| "loss": 0.11, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 8.71, | |
| "learning_rate": 6.125084976206662e-06, | |
| "loss": 0.1087, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 6.108089734874235e-06, | |
| "loss": 0.1096, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 6.091094493541809e-06, | |
| "loss": 0.1119, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 6.0740992522093826e-06, | |
| "loss": 0.1102, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 6.057104010876955e-06, | |
| "loss": 0.1091, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 6.040108769544528e-06, | |
| "loss": 0.1116, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 6.023113528212101e-06, | |
| "loss": 0.1096, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 6.006118286879674e-06, | |
| "loss": 0.1116, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 5.989123045547247e-06, | |
| "loss": 0.11, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 8.99, | |
| "learning_rate": 5.97212780421482e-06, | |
| "loss": 0.1112, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 5.955132562882393e-06, | |
| "loss": 0.1081, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 5.938137321549966e-06, | |
| "loss": 0.1053, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 5.92114208021754e-06, | |
| "loss": 0.106, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 5.904146838885113e-06, | |
| "loss": 0.1043, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 5.887151597552686e-06, | |
| "loss": 0.1066, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 5.870156356220259e-06, | |
| "loss": 0.1043, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 5.853161114887832e-06, | |
| "loss": 0.1042, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 9.24, | |
| "learning_rate": 5.836165873555405e-06, | |
| "loss": 0.1086, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 9.27, | |
| "learning_rate": 5.819170632222978e-06, | |
| "loss": 0.1073, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 9.31, | |
| "learning_rate": 5.802175390890551e-06, | |
| "loss": 0.1059, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 5.7851801495581236e-06, | |
| "loss": 0.1069, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 5.768184908225697e-06, | |
| "loss": 0.109, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 5.75118966689327e-06, | |
| "loss": 0.1037, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "eval_loss": 0.24955707788467407, | |
| "eval_runtime": 72.5131, | |
| "eval_samples_per_second": 123.302, | |
| "eval_steps_per_second": 1.931, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 5.734194425560844e-06, | |
| "loss": 0.1047, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 5.7171991842284165e-06, | |
| "loss": 0.1045, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 9.49, | |
| "learning_rate": 5.70020394289599e-06, | |
| "loss": 0.1055, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 5.6832087015635625e-06, | |
| "loss": 0.1038, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 5.666213460231136e-06, | |
| "loss": 0.1058, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 5.649218218898709e-06, | |
| "loss": 0.1044, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 5.632222977566282e-06, | |
| "loss": 0.1042, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 5.615227736233855e-06, | |
| "loss": 0.1048, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 5.598232494901428e-06, | |
| "loss": 0.107, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 9.71, | |
| "learning_rate": 5.581237253569001e-06, | |
| "loss": 0.1061, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 5.564242012236573e-06, | |
| "loss": 0.1032, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 5.547246770904148e-06, | |
| "loss": 0.1042, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 9.81, | |
| "learning_rate": 5.530251529571721e-06, | |
| "loss": 0.1045, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 9.84, | |
| "learning_rate": 5.513256288239294e-06, | |
| "loss": 0.1042, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 9.87, | |
| "learning_rate": 5.496261046906867e-06, | |
| "loss": 0.1058, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 5.47926580557444e-06, | |
| "loss": 0.104, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 5.462270564242012e-06, | |
| "loss": 0.104, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 5.445275322909586e-06, | |
| "loss": 0.1039, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 5.428280081577158e-06, | |
| "loss": 0.1041, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 5.411284840244732e-06, | |
| "loss": 0.0994, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 5.394289598912304e-06, | |
| "loss": 0.1007, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 10.09, | |
| "learning_rate": 5.377294357579879e-06, | |
| "loss": 0.1016, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "learning_rate": 5.360299116247451e-06, | |
| "loss": 0.0977, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 5.343303874915025e-06, | |
| "loss": 0.0992, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 10.18, | |
| "learning_rate": 5.326308633582597e-06, | |
| "loss": 0.1005, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 10.21, | |
| "learning_rate": 5.309313392250171e-06, | |
| "loss": 0.0987, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 10.25, | |
| "learning_rate": 5.292318150917743e-06, | |
| "loss": 0.1007, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 10.28, | |
| "learning_rate": 5.275322909585317e-06, | |
| "loss": 0.0996, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 10.31, | |
| "learning_rate": 5.258327668252889e-06, | |
| "loss": 0.0982, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 10.34, | |
| "learning_rate": 5.241332426920462e-06, | |
| "loss": 0.103, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 5.2243371855880355e-06, | |
| "loss": 0.0993, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 10.4, | |
| "learning_rate": 5.207341944255608e-06, | |
| "loss": 0.0986, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 10.43, | |
| "learning_rate": 5.190346702923182e-06, | |
| "loss": 0.1006, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 10.47, | |
| "learning_rate": 5.173351461590756e-06, | |
| "loss": 0.0995, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 10.5, | |
| "learning_rate": 5.156356220258328e-06, | |
| "loss": 0.1022, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 5.139360978925901e-06, | |
| "loss": 0.1001, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 10.56, | |
| "learning_rate": 5.1223657375934745e-06, | |
| "loss": 0.0998, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 5.105370496261047e-06, | |
| "loss": 0.0978, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 10.62, | |
| "learning_rate": 5.0883752549286205e-06, | |
| "loss": 0.0997, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 10.65, | |
| "learning_rate": 5.071380013596193e-06, | |
| "loss": 0.1017, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 10.68, | |
| "learning_rate": 5.0543847722637666e-06, | |
| "loss": 0.0988, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 5.037389530931339e-06, | |
| "loss": 0.102, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 10.75, | |
| "learning_rate": 5.020394289598912e-06, | |
| "loss": 0.0986, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "learning_rate": 5.003399048266486e-06, | |
| "loss": 0.0986, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 4.986403806934059e-06, | |
| "loss": 0.0983, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 10.84, | |
| "learning_rate": 4.969408565601632e-06, | |
| "loss": 0.0999, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 4.9524133242692056e-06, | |
| "loss": 0.0994, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 10.9, | |
| "learning_rate": 4.935418082936778e-06, | |
| "loss": 0.1001, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 10.94, | |
| "learning_rate": 4.918422841604351e-06, | |
| "loss": 0.0997, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "learning_rate": 4.901427600271924e-06, | |
| "loss": 0.1011, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 10.97, | |
| "eval_loss": 0.2406775802373886, | |
| "eval_runtime": 78.2369, | |
| "eval_samples_per_second": 114.281, | |
| "eval_steps_per_second": 1.789, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 4.884432358939497e-06, | |
| "loss": 0.1, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 11.03, | |
| "learning_rate": 4.86743711760707e-06, | |
| "loss": 0.0959, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 11.06, | |
| "learning_rate": 4.850441876274644e-06, | |
| "loss": 0.094, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 11.09, | |
| "learning_rate": 4.833446634942216e-06, | |
| "loss": 0.0953, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 4.81645139360979e-06, | |
| "loss": 0.0948, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 11.15, | |
| "learning_rate": 4.799456152277362e-06, | |
| "loss": 0.0932, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "learning_rate": 4.782460910944936e-06, | |
| "loss": 0.095, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 11.22, | |
| "learning_rate": 4.765465669612509e-06, | |
| "loss": 0.0953, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "learning_rate": 4.748470428280082e-06, | |
| "loss": 0.0952, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 4.731475186947655e-06, | |
| "loss": 0.0941, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 11.31, | |
| "learning_rate": 4.714479945615228e-06, | |
| "loss": 0.096, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 4.697484704282801e-06, | |
| "loss": 0.0948, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "learning_rate": 4.680489462950375e-06, | |
| "loss": 0.0945, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 11.41, | |
| "learning_rate": 4.663494221617947e-06, | |
| "loss": 0.0962, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 4.64649898028552e-06, | |
| "loss": 0.0949, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 11.47, | |
| "learning_rate": 4.6295037389530934e-06, | |
| "loss": 0.0953, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 11.5, | |
| "learning_rate": 4.612508497620666e-06, | |
| "loss": 0.0944, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 11.53, | |
| "learning_rate": 4.5955132562882395e-06, | |
| "loss": 0.0963, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "learning_rate": 4.578518014955813e-06, | |
| "loss": 0.096, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 4.5615227736233855e-06, | |
| "loss": 0.0936, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 11.62, | |
| "learning_rate": 4.544527532290959e-06, | |
| "loss": 0.0982, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 11.66, | |
| "learning_rate": 4.527532290958532e-06, | |
| "loss": 0.0951, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 11.69, | |
| "learning_rate": 4.510537049626105e-06, | |
| "loss": 0.0943, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 11.72, | |
| "learning_rate": 4.4935418082936785e-06, | |
| "loss": 0.094, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 11.75, | |
| "learning_rate": 4.476546566961251e-06, | |
| "loss": 0.095, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 11.78, | |
| "learning_rate": 4.4595513256288245e-06, | |
| "loss": 0.0936, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 4.442556084296397e-06, | |
| "loss": 0.0953, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 4.425560842963971e-06, | |
| "loss": 0.0934, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 11.88, | |
| "learning_rate": 4.408565601631544e-06, | |
| "loss": 0.0941, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "learning_rate": 4.391570360299117e-06, | |
| "loss": 0.095, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 11.94, | |
| "learning_rate": 4.374575118966689e-06, | |
| "loss": 0.0942, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 11.97, | |
| "learning_rate": 4.357579877634263e-06, | |
| "loss": 0.0967, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 4.340584636301835e-06, | |
| "loss": 0.0935, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 4.323589394969409e-06, | |
| "loss": 0.0911, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 4.306594153636982e-06, | |
| "loss": 0.0905, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 12.09, | |
| "learning_rate": 4.289598912304555e-06, | |
| "loss": 0.0913, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 4.272603670972128e-06, | |
| "loss": 0.0922, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 12.16, | |
| "learning_rate": 4.255608429639701e-06, | |
| "loss": 0.092, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 4.238613188307274e-06, | |
| "loss": 0.0911, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 4.221617946974848e-06, | |
| "loss": 0.092, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 12.25, | |
| "learning_rate": 4.20462270564242e-06, | |
| "loss": 0.0917, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 12.28, | |
| "learning_rate": 4.187627464309994e-06, | |
| "loss": 0.0909, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 12.31, | |
| "learning_rate": 4.170632222977566e-06, | |
| "loss": 0.0913, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "learning_rate": 4.15363698164514e-06, | |
| "loss": 0.0904, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 12.38, | |
| "learning_rate": 4.136641740312713e-06, | |
| "loss": 0.089, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "learning_rate": 4.119646498980286e-06, | |
| "loss": 0.0909, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 4.1026512576478585e-06, | |
| "loss": 0.0907, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 12.47, | |
| "learning_rate": 4.085656016315432e-06, | |
| "loss": 0.09, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 4.0686607749830045e-06, | |
| "loss": 0.0897, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "learning_rate": 4.051665533650578e-06, | |
| "loss": 0.0905, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 12.53, | |
| "eval_loss": 0.23563739657402039, | |
| "eval_runtime": 76.2602, | |
| "eval_samples_per_second": 117.243, | |
| "eval_steps_per_second": 1.836, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 4.034670292318151e-06, | |
| "loss": 0.091, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 12.6, | |
| "learning_rate": 4.017675050985724e-06, | |
| "loss": 0.0913, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 12.63, | |
| "learning_rate": 4.0006798096532975e-06, | |
| "loss": 0.0901, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 12.66, | |
| "learning_rate": 3.98368456832087e-06, | |
| "loss": 0.0907, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 12.69, | |
| "learning_rate": 3.9666893269884435e-06, | |
| "loss": 0.0899, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 12.72, | |
| "learning_rate": 3.949694085656017e-06, | |
| "loss": 0.0891, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 3.9326988443235896e-06, | |
| "loss": 0.0906, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "learning_rate": 3.915703602991163e-06, | |
| "loss": 0.0918, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 12.82, | |
| "learning_rate": 3.898708361658736e-06, | |
| "loss": 0.0908, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 12.85, | |
| "learning_rate": 3.881713120326309e-06, | |
| "loss": 0.0883, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 12.88, | |
| "learning_rate": 3.8647178789938825e-06, | |
| "loss": 0.0903, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 12.91, | |
| "learning_rate": 3.847722637661455e-06, | |
| "loss": 0.0906, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 3.8307273963290285e-06, | |
| "loss": 0.0919, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 3.813732154996601e-06, | |
| "loss": 0.0896, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 3.796736913664174e-06, | |
| "loss": 0.0908, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 13.03, | |
| "learning_rate": 3.7797416723317476e-06, | |
| "loss": 0.0874, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 13.07, | |
| "learning_rate": 3.7627464309993206e-06, | |
| "loss": 0.0872, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 13.1, | |
| "learning_rate": 3.7457511896668937e-06, | |
| "loss": 0.0872, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "learning_rate": 3.7287559483344667e-06, | |
| "loss": 0.0866, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "learning_rate": 3.7117607070020393e-06, | |
| "loss": 0.0865, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 3.694765465669613e-06, | |
| "loss": 0.0856, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 13.22, | |
| "learning_rate": 3.677770224337186e-06, | |
| "loss": 0.088, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 13.25, | |
| "learning_rate": 3.660774983004759e-06, | |
| "loss": 0.0881, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 13.29, | |
| "learning_rate": 3.643779741672332e-06, | |
| "loss": 0.0889, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 13.32, | |
| "learning_rate": 3.626784500339905e-06, | |
| "loss": 0.0861, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "learning_rate": 3.6097892590074783e-06, | |
| "loss": 0.0891, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 13.38, | |
| "learning_rate": 3.5927940176750513e-06, | |
| "loss": 0.0883, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 3.5757987763426243e-06, | |
| "loss": 0.0859, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 3.5588035350101974e-06, | |
| "loss": 0.0865, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 13.47, | |
| "learning_rate": 3.5418082936777704e-06, | |
| "loss": 0.0851, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 13.5, | |
| "learning_rate": 3.5248130523453434e-06, | |
| "loss": 0.0885, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 13.54, | |
| "learning_rate": 3.507817811012917e-06, | |
| "loss": 0.0851, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 3.49082256968049e-06, | |
| "loss": 0.0862, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "learning_rate": 3.473827328348063e-06, | |
| "loss": 0.0884, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 3.456832087015636e-06, | |
| "loss": 0.0874, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 13.66, | |
| "learning_rate": 3.439836845683209e-06, | |
| "loss": 0.0868, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 13.69, | |
| "learning_rate": 3.4228416043507824e-06, | |
| "loss": 0.0856, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 3.4058463630183554e-06, | |
| "loss": 0.0877, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 13.76, | |
| "learning_rate": 3.388851121685928e-06, | |
| "loss": 0.0863, | |
| "step": 43900 | |
| }, | |
| { | |
| "epoch": 13.79, | |
| "learning_rate": 3.371855880353501e-06, | |
| "loss": 0.0868, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 13.82, | |
| "learning_rate": 3.354860639021074e-06, | |
| "loss": 0.0871, | |
| "step": 44100 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 3.3378653976886475e-06, | |
| "loss": 0.0845, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 13.88, | |
| "learning_rate": 3.3208701563562205e-06, | |
| "loss": 0.0874, | |
| "step": 44300 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 3.3038749150237936e-06, | |
| "loss": 0.0906, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 13.94, | |
| "learning_rate": 3.2868796736913666e-06, | |
| "loss": 0.0863, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 13.97, | |
| "learning_rate": 3.2698844323589396e-06, | |
| "loss": 0.0868, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 14.01, | |
| "learning_rate": 3.2528891910265126e-06, | |
| "loss": 0.086, | |
| "step": 44700 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 3.235893949694086e-06, | |
| "loss": 0.0839, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 14.07, | |
| "learning_rate": 3.218898708361659e-06, | |
| "loss": 0.0837, | |
| "step": 44900 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 3.201903467029232e-06, | |
| "loss": 0.0827, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "eval_loss": 0.23157520592212677, | |
| "eval_runtime": 66.6414, | |
| "eval_samples_per_second": 134.166, | |
| "eval_steps_per_second": 2.101, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 14.13, | |
| "learning_rate": 3.184908225696805e-06, | |
| "loss": 0.0829, | |
| "step": 45100 | |
| }, | |
| { | |
| "epoch": 14.16, | |
| "learning_rate": 3.167912984364378e-06, | |
| "loss": 0.0843, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 3.1509177430319516e-06, | |
| "loss": 0.0837, | |
| "step": 45300 | |
| }, | |
| { | |
| "epoch": 14.23, | |
| "learning_rate": 3.1339225016995247e-06, | |
| "loss": 0.0835, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 3.1169272603670973e-06, | |
| "loss": 0.0834, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 14.29, | |
| "learning_rate": 3.0999320190346703e-06, | |
| "loss": 0.0838, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 3.0829367777022433e-06, | |
| "loss": 0.0841, | |
| "step": 45700 | |
| }, | |
| { | |
| "epoch": 14.35, | |
| "learning_rate": 3.0659415363698168e-06, | |
| "loss": 0.0845, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "learning_rate": 3.0489462950373898e-06, | |
| "loss": 0.0825, | |
| "step": 45900 | |
| }, | |
| { | |
| "epoch": 14.41, | |
| "learning_rate": 3.031951053704963e-06, | |
| "loss": 0.0853, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "learning_rate": 3.014955812372536e-06, | |
| "loss": 0.0845, | |
| "step": 46100 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 2.997960571040109e-06, | |
| "loss": 0.0839, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 14.51, | |
| "learning_rate": 2.980965329707682e-06, | |
| "loss": 0.0835, | |
| "step": 46300 | |
| }, | |
| { | |
| "epoch": 14.54, | |
| "learning_rate": 2.9639700883752553e-06, | |
| "loss": 0.084, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 14.57, | |
| "learning_rate": 2.9469748470428284e-06, | |
| "loss": 0.0848, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 14.6, | |
| "learning_rate": 2.9299796057104014e-06, | |
| "loss": 0.0831, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 14.63, | |
| "learning_rate": 2.9129843643779744e-06, | |
| "loss": 0.0811, | |
| "step": 46700 | |
| }, | |
| { | |
| "epoch": 14.66, | |
| "learning_rate": 2.8959891230455474e-06, | |
| "loss": 0.0843, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 2.878993881713121e-06, | |
| "loss": 0.0821, | |
| "step": 46900 | |
| }, | |
| { | |
| "epoch": 14.73, | |
| "learning_rate": 2.861998640380694e-06, | |
| "loss": 0.0827, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 14.76, | |
| "learning_rate": 2.845003399048267e-06, | |
| "loss": 0.0848, | |
| "step": 47100 | |
| }, | |
| { | |
| "epoch": 14.79, | |
| "learning_rate": 2.8280081577158395e-06, | |
| "loss": 0.0833, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 14.82, | |
| "learning_rate": 2.8110129163834125e-06, | |
| "loss": 0.0835, | |
| "step": 47300 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 2.794017675050986e-06, | |
| "loss": 0.0823, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 14.88, | |
| "learning_rate": 2.777022433718559e-06, | |
| "loss": 0.0831, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 14.91, | |
| "learning_rate": 2.760027192386132e-06, | |
| "loss": 0.0848, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 14.95, | |
| "learning_rate": 2.743031951053705e-06, | |
| "loss": 0.0817, | |
| "step": 47700 | |
| }, | |
| { | |
| "epoch": 14.98, | |
| "learning_rate": 2.726036709721278e-06, | |
| "loss": 0.0838, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 15.01, | |
| "learning_rate": 2.709041468388851e-06, | |
| "loss": 0.0828, | |
| "step": 47900 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 2.6920462270564246e-06, | |
| "loss": 0.0803, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 15.07, | |
| "learning_rate": 2.6750509857239976e-06, | |
| "loss": 0.0799, | |
| "step": 48100 | |
| }, | |
| { | |
| "epoch": 15.1, | |
| "learning_rate": 2.6580557443915706e-06, | |
| "loss": 0.0805, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 15.13, | |
| "learning_rate": 2.6410605030591436e-06, | |
| "loss": 0.0806, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 15.17, | |
| "learning_rate": 2.6240652617267167e-06, | |
| "loss": 0.0825, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 2.60707002039429e-06, | |
| "loss": 0.0801, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 15.23, | |
| "learning_rate": 2.590074779061863e-06, | |
| "loss": 0.081, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 2.573079537729436e-06, | |
| "loss": 0.0827, | |
| "step": 48700 | |
| }, | |
| { | |
| "epoch": 15.29, | |
| "learning_rate": 2.5560842963970088e-06, | |
| "loss": 0.0808, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 15.32, | |
| "learning_rate": 2.5390890550645818e-06, | |
| "loss": 0.0817, | |
| "step": 48900 | |
| }, | |
| { | |
| "epoch": 15.35, | |
| "learning_rate": 2.5220938137321557e-06, | |
| "loss": 0.0797, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 15.38, | |
| "learning_rate": 2.5050985723997283e-06, | |
| "loss": 0.079, | |
| "step": 49100 | |
| }, | |
| { | |
| "epoch": 15.42, | |
| "learning_rate": 2.4881033310673013e-06, | |
| "loss": 0.0811, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "learning_rate": 2.4711080897348743e-06, | |
| "loss": 0.0794, | |
| "step": 49300 | |
| }, | |
| { | |
| "epoch": 15.48, | |
| "learning_rate": 2.4541128484024478e-06, | |
| "loss": 0.0812, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 15.51, | |
| "learning_rate": 2.4371176070700208e-06, | |
| "loss": 0.0822, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 15.54, | |
| "learning_rate": 2.4201223657375934e-06, | |
| "loss": 0.0815, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 15.57, | |
| "learning_rate": 2.403127124405167e-06, | |
| "loss": 0.079, | |
| "step": 49700 | |
| }, | |
| { | |
| "epoch": 15.6, | |
| "learning_rate": 2.38613188307274e-06, | |
| "loss": 0.0793, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 2.369136641740313e-06, | |
| "loss": 0.0798, | |
| "step": 49900 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 2.352141400407886e-06, | |
| "loss": 0.0789, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "eval_loss": 0.2277790755033493, | |
| "eval_runtime": 81.7922, | |
| "eval_samples_per_second": 109.314, | |
| "eval_steps_per_second": 1.712, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 2.335146159075459e-06, | |
| "loss": 0.0812, | |
| "step": 50100 | |
| }, | |
| { | |
| "epoch": 15.73, | |
| "learning_rate": 2.3181509177430324e-06, | |
| "loss": 0.0795, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 15.76, | |
| "learning_rate": 2.3011556764106054e-06, | |
| "loss": 0.0784, | |
| "step": 50300 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 2.284160435078178e-06, | |
| "loss": 0.0826, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 15.82, | |
| "learning_rate": 2.2671651937457514e-06, | |
| "loss": 0.0813, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 2.2501699524133245e-06, | |
| "loss": 0.0793, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 2.2331747110808975e-06, | |
| "loss": 0.0803, | |
| "step": 50700 | |
| }, | |
| { | |
| "epoch": 15.92, | |
| "learning_rate": 2.2161794697484705e-06, | |
| "loss": 0.0792, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 15.95, | |
| "learning_rate": 2.1991842284160435e-06, | |
| "loss": 0.0808, | |
| "step": 50900 | |
| }, | |
| { | |
| "epoch": 15.98, | |
| "learning_rate": 2.182188987083617e-06, | |
| "loss": 0.0809, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 16.01, | |
| "learning_rate": 2.16519374575119e-06, | |
| "loss": 0.0786, | |
| "step": 51100 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 2.1481985044187626e-06, | |
| "loss": 0.0793, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 16.07, | |
| "learning_rate": 2.131203263086336e-06, | |
| "loss": 0.0783, | |
| "step": 51300 | |
| }, | |
| { | |
| "epoch": 16.11, | |
| "learning_rate": 2.114208021753909e-06, | |
| "loss": 0.078, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 16.14, | |
| "learning_rate": 2.097212780421482e-06, | |
| "loss": 0.0783, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 16.17, | |
| "learning_rate": 2.080217539089055e-06, | |
| "loss": 0.0789, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 16.2, | |
| "learning_rate": 2.063222297756628e-06, | |
| "loss": 0.077, | |
| "step": 51700 | |
| }, | |
| { | |
| "epoch": 16.23, | |
| "learning_rate": 2.0462270564242016e-06, | |
| "loss": 0.0768, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 16.26, | |
| "learning_rate": 2.0292318150917746e-06, | |
| "loss": 0.0786, | |
| "step": 51900 | |
| }, | |
| { | |
| "epoch": 16.29, | |
| "learning_rate": 2.0122365737593477e-06, | |
| "loss": 0.0774, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 16.32, | |
| "learning_rate": 1.9952413324269207e-06, | |
| "loss": 0.0763, | |
| "step": 52100 | |
| }, | |
| { | |
| "epoch": 16.36, | |
| "learning_rate": 1.9782460910944937e-06, | |
| "loss": 0.0793, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 16.39, | |
| "learning_rate": 1.9612508497620667e-06, | |
| "loss": 0.0785, | |
| "step": 52300 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 1.9442556084296398e-06, | |
| "loss": 0.0773, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 16.45, | |
| "learning_rate": 1.9272603670972128e-06, | |
| "loss": 0.0759, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 16.48, | |
| "learning_rate": 1.9102651257647862e-06, | |
| "loss": 0.0778, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 16.51, | |
| "learning_rate": 1.893269884432359e-06, | |
| "loss": 0.0795, | |
| "step": 52700 | |
| }, | |
| { | |
| "epoch": 16.54, | |
| "learning_rate": 1.876274643099932e-06, | |
| "loss": 0.079, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 16.58, | |
| "learning_rate": 1.8592794017675053e-06, | |
| "loss": 0.0771, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 16.61, | |
| "learning_rate": 1.8422841604350783e-06, | |
| "loss": 0.0789, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 16.64, | |
| "learning_rate": 1.8252889191026513e-06, | |
| "loss": 0.0785, | |
| "step": 53100 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 1.8082936777702246e-06, | |
| "loss": 0.0773, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "learning_rate": 1.7912984364377974e-06, | |
| "loss": 0.0763, | |
| "step": 53300 | |
| }, | |
| { | |
| "epoch": 16.73, | |
| "learning_rate": 1.7743031951053708e-06, | |
| "loss": 0.0793, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 16.76, | |
| "learning_rate": 1.7573079537729437e-06, | |
| "loss": 0.0771, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 16.79, | |
| "learning_rate": 1.7403127124405167e-06, | |
| "loss": 0.0779, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 16.83, | |
| "learning_rate": 1.72331747110809e-06, | |
| "loss": 0.0765, | |
| "step": 53700 | |
| }, | |
| { | |
| "epoch": 16.86, | |
| "learning_rate": 1.706322229775663e-06, | |
| "loss": 0.077, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "learning_rate": 1.689326988443236e-06, | |
| "loss": 0.0774, | |
| "step": 53900 | |
| }, | |
| { | |
| "epoch": 16.92, | |
| "learning_rate": 1.6723317471108092e-06, | |
| "loss": 0.0771, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "learning_rate": 1.6553365057783822e-06, | |
| "loss": 0.0783, | |
| "step": 54100 | |
| }, | |
| { | |
| "epoch": 16.98, | |
| "learning_rate": 1.6383412644459555e-06, | |
| "loss": 0.0785, | |
| "step": 54200 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 1.6213460231135283e-06, | |
| "loss": 0.0771, | |
| "step": 54300 | |
| }, | |
| { | |
| "epoch": 17.05, | |
| "learning_rate": 1.6043507817811013e-06, | |
| "loss": 0.0776, | |
| "step": 54400 | |
| }, | |
| { | |
| "epoch": 17.08, | |
| "learning_rate": 1.5873555404486745e-06, | |
| "loss": 0.0753, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 17.11, | |
| "learning_rate": 1.5703602991162476e-06, | |
| "loss": 0.0758, | |
| "step": 54600 | |
| }, | |
| { | |
| "epoch": 17.14, | |
| "learning_rate": 1.5533650577838208e-06, | |
| "loss": 0.077, | |
| "step": 54700 | |
| }, | |
| { | |
| "epoch": 17.17, | |
| "learning_rate": 1.5363698164513938e-06, | |
| "loss": 0.0749, | |
| "step": 54800 | |
| }, | |
| { | |
| "epoch": 17.2, | |
| "learning_rate": 1.5193745751189668e-06, | |
| "loss": 0.076, | |
| "step": 54900 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "learning_rate": 1.50237933378654e-06, | |
| "loss": 0.0741, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 17.23, | |
| "eval_loss": 0.224798783659935, | |
| "eval_runtime": 80.315, | |
| "eval_samples_per_second": 111.324, | |
| "eval_steps_per_second": 1.743, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 17.26, | |
| "learning_rate": 1.4853840924541129e-06, | |
| "loss": 0.0763, | |
| "step": 55100 | |
| }, | |
| { | |
| "epoch": 17.3, | |
| "learning_rate": 1.468388851121686e-06, | |
| "loss": 0.0771, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 17.33, | |
| "learning_rate": 1.4513936097892592e-06, | |
| "loss": 0.0768, | |
| "step": 55300 | |
| }, | |
| { | |
| "epoch": 17.36, | |
| "learning_rate": 1.4343983684568322e-06, | |
| "loss": 0.0775, | |
| "step": 55400 | |
| }, | |
| { | |
| "epoch": 17.39, | |
| "learning_rate": 1.4174031271244054e-06, | |
| "loss": 0.075, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 17.42, | |
| "learning_rate": 1.4004078857919784e-06, | |
| "loss": 0.0734, | |
| "step": 55600 | |
| }, | |
| { | |
| "epoch": 17.45, | |
| "learning_rate": 1.3834126444595515e-06, | |
| "loss": 0.0779, | |
| "step": 55700 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 1.3664174031271247e-06, | |
| "loss": 0.0746, | |
| "step": 55800 | |
| }, | |
| { | |
| "epoch": 17.52, | |
| "learning_rate": 1.3494221617946975e-06, | |
| "loss": 0.075, | |
| "step": 55900 | |
| }, | |
| { | |
| "epoch": 17.55, | |
| "learning_rate": 1.3324269204622705e-06, | |
| "loss": 0.0751, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 17.58, | |
| "learning_rate": 1.3154316791298438e-06, | |
| "loss": 0.0746, | |
| "step": 56100 | |
| }, | |
| { | |
| "epoch": 17.61, | |
| "learning_rate": 1.2984364377974168e-06, | |
| "loss": 0.0747, | |
| "step": 56200 | |
| }, | |
| { | |
| "epoch": 17.64, | |
| "learning_rate": 1.28144119646499e-06, | |
| "loss": 0.077, | |
| "step": 56300 | |
| }, | |
| { | |
| "epoch": 17.67, | |
| "learning_rate": 1.264445955132563e-06, | |
| "loss": 0.0731, | |
| "step": 56400 | |
| }, | |
| { | |
| "epoch": 17.7, | |
| "learning_rate": 1.247450713800136e-06, | |
| "loss": 0.077, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 17.73, | |
| "learning_rate": 1.230455472467709e-06, | |
| "loss": 0.0762, | |
| "step": 56600 | |
| }, | |
| { | |
| "epoch": 17.77, | |
| "learning_rate": 1.2134602311352821e-06, | |
| "loss": 0.0764, | |
| "step": 56700 | |
| }, | |
| { | |
| "epoch": 17.8, | |
| "learning_rate": 1.1964649898028554e-06, | |
| "loss": 0.0742, | |
| "step": 56800 | |
| }, | |
| { | |
| "epoch": 17.83, | |
| "learning_rate": 1.1794697484704284e-06, | |
| "loss": 0.0742, | |
| "step": 56900 | |
| }, | |
| { | |
| "epoch": 17.86, | |
| "learning_rate": 1.1624745071380014e-06, | |
| "loss": 0.0755, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 17.89, | |
| "learning_rate": 1.1454792658055744e-06, | |
| "loss": 0.0759, | |
| "step": 57100 | |
| }, | |
| { | |
| "epoch": 17.92, | |
| "learning_rate": 1.1284840244731477e-06, | |
| "loss": 0.075, | |
| "step": 57200 | |
| }, | |
| { | |
| "epoch": 17.95, | |
| "learning_rate": 1.1114887831407207e-06, | |
| "loss": 0.0762, | |
| "step": 57300 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "learning_rate": 1.0944935418082937e-06, | |
| "loss": 0.0741, | |
| "step": 57400 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 1.0774983004758667e-06, | |
| "loss": 0.0751, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 18.05, | |
| "learning_rate": 1.06050305914344e-06, | |
| "loss": 0.0735, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 18.08, | |
| "learning_rate": 1.043507817811013e-06, | |
| "loss": 0.074, | |
| "step": 57700 | |
| }, | |
| { | |
| "epoch": 18.11, | |
| "learning_rate": 1.026512576478586e-06, | |
| "loss": 0.0745, | |
| "step": 57800 | |
| }, | |
| { | |
| "epoch": 18.14, | |
| "learning_rate": 1.009517335146159e-06, | |
| "loss": 0.0745, | |
| "step": 57900 | |
| }, | |
| { | |
| "epoch": 18.17, | |
| "learning_rate": 9.925220938137323e-07, | |
| "loss": 0.0752, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 18.2, | |
| "learning_rate": 9.755268524813053e-07, | |
| "loss": 0.0746, | |
| "step": 58100 | |
| }, | |
| { | |
| "epoch": 18.24, | |
| "learning_rate": 9.585316111488783e-07, | |
| "loss": 0.0755, | |
| "step": 58200 | |
| }, | |
| { | |
| "epoch": 18.27, | |
| "learning_rate": 9.415363698164515e-07, | |
| "loss": 0.0757, | |
| "step": 58300 | |
| }, | |
| { | |
| "epoch": 18.3, | |
| "learning_rate": 9.245411284840246e-07, | |
| "loss": 0.0731, | |
| "step": 58400 | |
| }, | |
| { | |
| "epoch": 18.33, | |
| "learning_rate": 9.075458871515977e-07, | |
| "loss": 0.075, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 18.36, | |
| "learning_rate": 8.905506458191706e-07, | |
| "loss": 0.0733, | |
| "step": 58600 | |
| }, | |
| { | |
| "epoch": 18.39, | |
| "learning_rate": 8.735554044867438e-07, | |
| "loss": 0.0751, | |
| "step": 58700 | |
| }, | |
| { | |
| "epoch": 18.42, | |
| "learning_rate": 8.565601631543169e-07, | |
| "loss": 0.0731, | |
| "step": 58800 | |
| }, | |
| { | |
| "epoch": 18.46, | |
| "learning_rate": 8.3956492182189e-07, | |
| "loss": 0.0743, | |
| "step": 58900 | |
| }, | |
| { | |
| "epoch": 18.49, | |
| "learning_rate": 8.22569680489463e-07, | |
| "loss": 0.0751, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 18.52, | |
| "learning_rate": 8.055744391570361e-07, | |
| "loss": 0.0729, | |
| "step": 59100 | |
| }, | |
| { | |
| "epoch": 18.55, | |
| "learning_rate": 7.885791978246092e-07, | |
| "loss": 0.0745, | |
| "step": 59200 | |
| }, | |
| { | |
| "epoch": 18.58, | |
| "learning_rate": 7.715839564921823e-07, | |
| "loss": 0.0753, | |
| "step": 59300 | |
| }, | |
| { | |
| "epoch": 18.61, | |
| "learning_rate": 7.545887151597553e-07, | |
| "loss": 0.074, | |
| "step": 59400 | |
| }, | |
| { | |
| "epoch": 18.64, | |
| "learning_rate": 7.375934738273284e-07, | |
| "loss": 0.0741, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 18.67, | |
| "learning_rate": 7.205982324949015e-07, | |
| "loss": 0.0727, | |
| "step": 59600 | |
| }, | |
| { | |
| "epoch": 18.71, | |
| "learning_rate": 7.036029911624747e-07, | |
| "loss": 0.0727, | |
| "step": 59700 | |
| }, | |
| { | |
| "epoch": 18.74, | |
| "learning_rate": 6.866077498300476e-07, | |
| "loss": 0.0754, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 18.77, | |
| "learning_rate": 6.696125084976207e-07, | |
| "loss": 0.0738, | |
| "step": 59900 | |
| }, | |
| { | |
| "epoch": 18.8, | |
| "learning_rate": 6.526172671651938e-07, | |
| "loss": 0.0714, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 18.8, | |
| "eval_loss": 0.22282175719738007, | |
| "eval_runtime": 115.3459, | |
| "eval_samples_per_second": 77.515, | |
| "eval_steps_per_second": 1.214, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 18.83, | |
| "learning_rate": 6.35622025832767e-07, | |
| "loss": 0.0727, | |
| "step": 60100 | |
| }, | |
| { | |
| "epoch": 18.86, | |
| "learning_rate": 6.1862678450034e-07, | |
| "loss": 0.0739, | |
| "step": 60200 | |
| }, | |
| { | |
| "epoch": 18.89, | |
| "learning_rate": 6.01631543167913e-07, | |
| "loss": 0.0729, | |
| "step": 60300 | |
| }, | |
| { | |
| "epoch": 18.93, | |
| "learning_rate": 5.846363018354861e-07, | |
| "loss": 0.0729, | |
| "step": 60400 | |
| }, | |
| { | |
| "epoch": 18.96, | |
| "learning_rate": 5.676410605030592e-07, | |
| "loss": 0.0731, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 18.99, | |
| "learning_rate": 5.506458191706323e-07, | |
| "loss": 0.0725, | |
| "step": 60600 | |
| }, | |
| { | |
| "epoch": 19.02, | |
| "learning_rate": 5.336505778382053e-07, | |
| "loss": 0.0727, | |
| "step": 60700 | |
| }, | |
| { | |
| "epoch": 19.05, | |
| "learning_rate": 5.166553365057785e-07, | |
| "loss": 0.0733, | |
| "step": 60800 | |
| }, | |
| { | |
| "epoch": 19.08, | |
| "learning_rate": 4.996600951733515e-07, | |
| "loss": 0.073, | |
| "step": 60900 | |
| }, | |
| { | |
| "epoch": 19.11, | |
| "learning_rate": 4.826648538409246e-07, | |
| "loss": 0.0721, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 19.14, | |
| "learning_rate": 4.6566961250849763e-07, | |
| "loss": 0.0731, | |
| "step": 61100 | |
| }, | |
| { | |
| "epoch": 19.18, | |
| "learning_rate": 4.4867437117607076e-07, | |
| "loss": 0.0749, | |
| "step": 61200 | |
| }, | |
| { | |
| "epoch": 19.21, | |
| "learning_rate": 4.316791298436438e-07, | |
| "loss": 0.0719, | |
| "step": 61300 | |
| }, | |
| { | |
| "epoch": 19.24, | |
| "learning_rate": 4.146838885112169e-07, | |
| "loss": 0.0706, | |
| "step": 61400 | |
| }, | |
| { | |
| "epoch": 19.27, | |
| "learning_rate": 3.9768864717879e-07, | |
| "loss": 0.0753, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 19.3, | |
| "learning_rate": 3.8069340584636307e-07, | |
| "loss": 0.0719, | |
| "step": 61600 | |
| }, | |
| { | |
| "epoch": 19.33, | |
| "learning_rate": 3.6369816451393615e-07, | |
| "loss": 0.0719, | |
| "step": 61700 | |
| }, | |
| { | |
| "epoch": 19.36, | |
| "learning_rate": 3.467029231815092e-07, | |
| "loss": 0.073, | |
| "step": 61800 | |
| }, | |
| { | |
| "epoch": 19.4, | |
| "learning_rate": 3.297076818490823e-07, | |
| "loss": 0.074, | |
| "step": 61900 | |
| }, | |
| { | |
| "epoch": 19.43, | |
| "learning_rate": 3.127124405166554e-07, | |
| "loss": 0.0718, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 19.46, | |
| "learning_rate": 2.9571719918422846e-07, | |
| "loss": 0.0728, | |
| "step": 62100 | |
| }, | |
| { | |
| "epoch": 19.49, | |
| "learning_rate": 2.7872195785180153e-07, | |
| "loss": 0.0728, | |
| "step": 62200 | |
| }, | |
| { | |
| "epoch": 19.52, | |
| "learning_rate": 2.617267165193746e-07, | |
| "loss": 0.0712, | |
| "step": 62300 | |
| }, | |
| { | |
| "epoch": 19.55, | |
| "learning_rate": 2.447314751869477e-07, | |
| "loss": 0.0747, | |
| "step": 62400 | |
| }, | |
| { | |
| "epoch": 19.58, | |
| "learning_rate": 2.2773623385452074e-07, | |
| "loss": 0.072, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 19.61, | |
| "learning_rate": 2.1074099252209382e-07, | |
| "loss": 0.0714, | |
| "step": 62600 | |
| }, | |
| { | |
| "epoch": 19.65, | |
| "learning_rate": 1.937457511896669e-07, | |
| "loss": 0.0717, | |
| "step": 62700 | |
| }, | |
| { | |
| "epoch": 19.68, | |
| "learning_rate": 1.7675050985723997e-07, | |
| "loss": 0.072, | |
| "step": 62800 | |
| }, | |
| { | |
| "epoch": 19.71, | |
| "learning_rate": 1.5975526852481305e-07, | |
| "loss": 0.0734, | |
| "step": 62900 | |
| }, | |
| { | |
| "epoch": 19.74, | |
| "learning_rate": 1.4276002719238612e-07, | |
| "loss": 0.0734, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 19.77, | |
| "learning_rate": 1.2576478585995923e-07, | |
| "loss": 0.072, | |
| "step": 63100 | |
| }, | |
| { | |
| "epoch": 19.8, | |
| "learning_rate": 1.0876954452753229e-07, | |
| "loss": 0.0731, | |
| "step": 63200 | |
| }, | |
| { | |
| "epoch": 19.83, | |
| "learning_rate": 9.177430319510537e-08, | |
| "loss": 0.0714, | |
| "step": 63300 | |
| }, | |
| { | |
| "epoch": 19.87, | |
| "learning_rate": 7.477906186267846e-08, | |
| "loss": 0.0725, | |
| "step": 63400 | |
| }, | |
| { | |
| "epoch": 19.9, | |
| "learning_rate": 5.778382053025154e-08, | |
| "loss": 0.0724, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 4.0788579197824615e-08, | |
| "loss": 0.0717, | |
| "step": 63600 | |
| }, | |
| { | |
| "epoch": 19.96, | |
| "learning_rate": 2.3793337865397692e-08, | |
| "loss": 0.0709, | |
| "step": 63700 | |
| }, | |
| { | |
| "epoch": 19.99, | |
| "learning_rate": 6.798096532970768e-09, | |
| "loss": 0.0726, | |
| "step": 63800 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 63840, | |
| "total_flos": 1.325011889553408e+20, | |
| "train_loss": 0.13975257427247245, | |
| "train_runtime": 211299.4316, | |
| "train_samples_per_second": 77.345, | |
| "train_steps_per_second": 0.302 | |
| } | |
| ], | |
| "max_steps": 63840, | |
| "num_train_epochs": 21, | |
| "total_flos": 1.325011889553408e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |