| { | |
| "best_metric": 3.9567816257476807, | |
| "best_model_checkpoint": "chinese-roberta-wwm-ext-finetuned-MC-hyper/checkpoint-1375", | |
| "epoch": 5.0, | |
| "global_step": 6875, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.970909090909091e-05, | |
| "loss": 1.3244, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.941818181818182e-05, | |
| "loss": 0.9671, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.912727272727273e-05, | |
| "loss": 0.8194, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.883636363636364e-05, | |
| "loss": 0.7545, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.854545454545455e-05, | |
| "loss": 0.7631, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.825454545454546e-05, | |
| "loss": 0.6651, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.796363636363637e-05, | |
| "loss": 0.5622, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.767272727272728e-05, | |
| "loss": 0.6096, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.738181818181819e-05, | |
| "loss": 0.4859, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.70909090909091e-05, | |
| "loss": 0.4879, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.680000000000001e-05, | |
| "loss": 0.4388, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.650909090909092e-05, | |
| "loss": 0.478, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.621818181818181e-05, | |
| "loss": 0.3661, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.592727272727274e-05, | |
| "loss": 0.4131, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.563636363636365e-05, | |
| "loss": 0.3415, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.534545454545456e-05, | |
| "loss": 0.3695, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 9.505454545454546e-05, | |
| "loss": 0.3059, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 9.476363636363636e-05, | |
| "loss": 0.268, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 9.447272727272728e-05, | |
| "loss": 0.3019, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.418181818181818e-05, | |
| "loss": 0.2147, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.38909090909091e-05, | |
| "loss": 0.3197, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 9.360000000000001e-05, | |
| "loss": 0.2764, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.330909090909091e-05, | |
| "loss": 0.2022, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.301818181818183e-05, | |
| "loss": 0.2326, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.272727272727273e-05, | |
| "loss": 0.3168, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.243636363636365e-05, | |
| "loss": 0.2157, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 9.214545454545455e-05, | |
| "loss": 0.2783, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.185454545454545e-05, | |
| "loss": 0.1759, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 9.156363636363636e-05, | |
| "loss": 0.169, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 9.127272727272727e-05, | |
| "loss": 0.1731, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.09818181818182e-05, | |
| "loss": 0.2469, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 9.069090909090909e-05, | |
| "loss": 0.1617, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.04e-05, | |
| "loss": 0.1841, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 9.010909090909091e-05, | |
| "loss": 0.1549, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 8.981818181818182e-05, | |
| "loss": 0.1393, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 8.952727272727273e-05, | |
| "loss": 0.1853, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.923636363636364e-05, | |
| "loss": 0.1609, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.894545454545455e-05, | |
| "loss": 0.1493, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 8.865454545454546e-05, | |
| "loss": 0.1436, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 8.836363636363637e-05, | |
| "loss": 0.1626, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 8.807272727272728e-05, | |
| "loss": 0.2139, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 8.778181818181819e-05, | |
| "loss": 0.1661, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 8.74909090909091e-05, | |
| "loss": 0.1579, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 8.72e-05, | |
| "loss": 0.1424, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 8.690909090909091e-05, | |
| "loss": 0.073, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 8.661818181818182e-05, | |
| "loss": 0.1764, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 8.632727272727273e-05, | |
| "loss": 0.1296, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 8.603636363636364e-05, | |
| "loss": 0.144, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 8.574545454545455e-05, | |
| "loss": 0.0955, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 8.545454545454545e-05, | |
| "loss": 0.1262, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 8.516363636363637e-05, | |
| "loss": 0.1599, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.487272727272728e-05, | |
| "loss": 0.0835, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 8.458181818181819e-05, | |
| "loss": 0.1509, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 8.42909090909091e-05, | |
| "loss": 0.0916, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 8.4e-05, | |
| "loss": 0.0863, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.370909090909092e-05, | |
| "loss": 0.1302, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.341818181818181e-05, | |
| "loss": 0.1324, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.312727272727274e-05, | |
| "loss": 0.0838, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.283636363636365e-05, | |
| "loss": 0.1127, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 8.254545454545454e-05, | |
| "loss": 0.1193, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 8.225454545454547e-05, | |
| "loss": 0.1404, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 8.196363636363636e-05, | |
| "loss": 0.125, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.167272727272728e-05, | |
| "loss": 0.0519, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.138181818181818e-05, | |
| "loss": 0.1256, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 8.109090909090909e-05, | |
| "loss": 0.1016, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 0.0944, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 8.050909090909091e-05, | |
| "loss": 0.0666, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.021818181818183e-05, | |
| "loss": 0.0903, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.1940000057220459, | |
| "eval_loss": 3.9567816257476807, | |
| "eval_runtime": 3.3624, | |
| "eval_samples_per_second": 148.705, | |
| "eval_steps_per_second": 4.759, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.992727272727273e-05, | |
| "loss": 0.1101, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 7.963636363636364e-05, | |
| "loss": 0.0603, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 7.934545454545455e-05, | |
| "loss": 0.0622, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 7.905454545454546e-05, | |
| "loss": 0.0463, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 7.876363636363638e-05, | |
| "loss": 0.0404, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 7.847272727272727e-05, | |
| "loss": 0.0476, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 7.818181818181818e-05, | |
| "loss": 0.0763, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 7.789090909090909e-05, | |
| "loss": 0.0363, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 7.76e-05, | |
| "loss": 0.0697, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 7.730909090909091e-05, | |
| "loss": 0.0838, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 7.701818181818182e-05, | |
| "loss": 0.0587, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.672727272727273e-05, | |
| "loss": 0.0246, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 7.643636363636364e-05, | |
| "loss": 0.0461, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 7.614545454545455e-05, | |
| "loss": 0.0695, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 7.585454545454546e-05, | |
| "loss": 0.0736, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 7.556363636363637e-05, | |
| "loss": 0.0261, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 7.527272727272728e-05, | |
| "loss": 0.0438, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 7.498181818181819e-05, | |
| "loss": 0.0858, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 7.469090909090908e-05, | |
| "loss": 0.0248, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 7.44e-05, | |
| "loss": 0.0615, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 7.410909090909092e-05, | |
| "loss": 0.0123, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 7.381818181818182e-05, | |
| "loss": 0.0451, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 7.352727272727273e-05, | |
| "loss": 0.05, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 7.323636363636363e-05, | |
| "loss": 0.0415, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 7.294545454545455e-05, | |
| "loss": 0.0734, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 7.265454545454545e-05, | |
| "loss": 0.037, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 7.236363636363637e-05, | |
| "loss": 0.045, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 7.207272727272728e-05, | |
| "loss": 0.0914, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 7.178181818181818e-05, | |
| "loss": 0.0288, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 7.14909090909091e-05, | |
| "loss": 0.0289, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 7.12e-05, | |
| "loss": 0.0355, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 7.090909090909092e-05, | |
| "loss": 0.0376, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.061818181818181e-05, | |
| "loss": 0.0375, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 7.032727272727272e-05, | |
| "loss": 0.0695, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 7.003636363636365e-05, | |
| "loss": 0.0414, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 6.974545454545454e-05, | |
| "loss": 0.0308, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 6.945454545454547e-05, | |
| "loss": 0.0209, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 6.916363636363636e-05, | |
| "loss": 0.0731, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 6.887272727272727e-05, | |
| "loss": 0.0409, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.858181818181818e-05, | |
| "loss": 0.0544, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 6.829090909090909e-05, | |
| "loss": 0.0677, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 6.800000000000001e-05, | |
| "loss": 0.0345, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 6.770909090909091e-05, | |
| "loss": 0.0262, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 6.741818181818182e-05, | |
| "loss": 0.066, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 6.712727272727273e-05, | |
| "loss": 0.0578, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 6.683636363636364e-05, | |
| "loss": 0.066, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 6.654545454545455e-05, | |
| "loss": 0.0565, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 6.625454545454546e-05, | |
| "loss": 0.0428, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 6.596363636363637e-05, | |
| "loss": 0.0387, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 6.567272727272727e-05, | |
| "loss": 0.0294, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.538181818181818e-05, | |
| "loss": 0.0332, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 6.50909090909091e-05, | |
| "loss": 0.0401, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 6.48e-05, | |
| "loss": 0.0324, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 6.450909090909091e-05, | |
| "loss": 0.0162, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 6.421818181818182e-05, | |
| "loss": 0.0501, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 6.392727272727273e-05, | |
| "loss": 0.0335, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 6.363636363636364e-05, | |
| "loss": 0.025, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6.334545454545455e-05, | |
| "loss": 0.0402, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 6.305454545454546e-05, | |
| "loss": 0.0624, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 6.276363636363637e-05, | |
| "loss": 0.0337, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.247272727272728e-05, | |
| "loss": 0.0264, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.218181818181819e-05, | |
| "loss": 0.057, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 6.18909090909091e-05, | |
| "loss": 0.0381, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 6.16e-05, | |
| "loss": 0.0316, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 6.130909090909092e-05, | |
| "loss": 0.0492, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.101818181818182e-05, | |
| "loss": 0.0405, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.0727272727272735e-05, | |
| "loss": 0.0594, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.043636363636364e-05, | |
| "loss": 0.0263, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.014545454545455e-05, | |
| "loss": 0.0432, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.1720000058412552, | |
| "eval_loss": 4.625512599945068, | |
| "eval_runtime": 3.3701, | |
| "eval_samples_per_second": 148.362, | |
| "eval_steps_per_second": 4.748, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 5.985454545454545e-05, | |
| "loss": 0.0235, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 5.9563636363636366e-05, | |
| "loss": 0.0293, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 5.927272727272728e-05, | |
| "loss": 0.0326, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 5.8981818181818184e-05, | |
| "loss": 0.0124, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 5.8690909090909094e-05, | |
| "loss": 0.0157, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 5.8399999999999997e-05, | |
| "loss": 0.0138, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 5.810909090909091e-05, | |
| "loss": 0.0291, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.7818181818181815e-05, | |
| "loss": 0.0047, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.752727272727273e-05, | |
| "loss": 0.0188, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.723636363636364e-05, | |
| "loss": 0.0277, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.6945454545454544e-05, | |
| "loss": 0.0118, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.665454545454546e-05, | |
| "loss": 0.0277, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.636363636363636e-05, | |
| "loss": 0.006, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.607272727272728e-05, | |
| "loss": 0.0214, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.578181818181818e-05, | |
| "loss": 0.0086, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.549090909090909e-05, | |
| "loss": 0.0091, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.520000000000001e-05, | |
| "loss": 0.0055, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5.490909090909091e-05, | |
| "loss": 0.0326, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 5.4618181818181826e-05, | |
| "loss": 0.0145, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5.432727272727273e-05, | |
| "loss": 0.0087, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 5.403636363636364e-05, | |
| "loss": 0.0084, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 5.374545454545454e-05, | |
| "loss": 0.0116, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 5.3454545454545457e-05, | |
| "loss": 0.0254, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 5.316363636363637e-05, | |
| "loss": 0.0033, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 5.2872727272727275e-05, | |
| "loss": 0.0246, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 5.2581818181818185e-05, | |
| "loss": 0.0222, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 5.229090909090909e-05, | |
| "loss": 0.0093, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 0.0112, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 5.1709090909090906e-05, | |
| "loss": 0.0335, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 5.141818181818182e-05, | |
| "loss": 0.0382, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 5.112727272727273e-05, | |
| "loss": 0.0264, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 5.0836363636363634e-05, | |
| "loss": 0.0131, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 5.054545454545455e-05, | |
| "loss": 0.0198, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 5.025454545454545e-05, | |
| "loss": 0.0027, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 4.996363636363637e-05, | |
| "loss": 0.0116, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 4.967272727272728e-05, | |
| "loss": 0.0066, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 4.938181818181818e-05, | |
| "loss": 0.04, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 4.909090909090909e-05, | |
| "loss": 0.0097, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 4.88e-05, | |
| "loss": 0.0119, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 4.850909090909091e-05, | |
| "loss": 0.0111, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 4.821818181818182e-05, | |
| "loss": 0.0311, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 4.792727272727273e-05, | |
| "loss": 0.0234, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 4.763636363636364e-05, | |
| "loss": 0.0124, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 4.734545454545455e-05, | |
| "loss": 0.0068, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 4.705454545454546e-05, | |
| "loss": 0.0349, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 4.6763636363636366e-05, | |
| "loss": 0.0112, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 4.6472727272727276e-05, | |
| "loss": 0.0122, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 4.618181818181818e-05, | |
| "loss": 0.0071, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.5890909090909094e-05, | |
| "loss": 0.0504, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.5600000000000004e-05, | |
| "loss": 0.0307, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.530909090909091e-05, | |
| "loss": 0.0048, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.501818181818182e-05, | |
| "loss": 0.0136, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.472727272727273e-05, | |
| "loss": 0.038, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 4.4436363636363635e-05, | |
| "loss": 0.0089, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 4.4145454545454544e-05, | |
| "loss": 0.0069, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 4.385454545454546e-05, | |
| "loss": 0.0161, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 4.356363636363637e-05, | |
| "loss": 0.013, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 4.327272727272728e-05, | |
| "loss": 0.0049, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 4.298181818181818e-05, | |
| "loss": 0.0304, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 4.269090909090909e-05, | |
| "loss": 0.03, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 4.24e-05, | |
| "loss": 0.0313, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 4.210909090909091e-05, | |
| "loss": 0.0079, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 4.181818181818182e-05, | |
| "loss": 0.0189, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 4.152727272727273e-05, | |
| "loss": 0.0019, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.123636363636364e-05, | |
| "loss": 0.0325, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 4.094545454545455e-05, | |
| "loss": 0.0276, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.065454545454546e-05, | |
| "loss": 0.0137, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 4.0363636363636367e-05, | |
| "loss": 0.0169, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 4.0072727272727276e-05, | |
| "loss": 0.0087, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.17599999904632568, | |
| "eval_loss": 6.114068508148193, | |
| "eval_runtime": 3.4252, | |
| "eval_samples_per_second": 145.975, | |
| "eval_steps_per_second": 4.671, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 3.978181818181818e-05, | |
| "loss": 0.0005, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 3.9490909090909095e-05, | |
| "loss": 0.0109, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.9200000000000004e-05, | |
| "loss": 0.0128, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.8909090909090914e-05, | |
| "loss": 0.0016, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 3.861818181818182e-05, | |
| "loss": 0.0009, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 3.8327272727272726e-05, | |
| "loss": 0.0036, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 3.8036363636363635e-05, | |
| "loss": 0.0124, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.7745454545454544e-05, | |
| "loss": 0.0128, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.745454545454546e-05, | |
| "loss": 0.0029, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 3.716363636363637e-05, | |
| "loss": 0.0088, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.687272727272727e-05, | |
| "loss": 0.0129, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 3.658181818181818e-05, | |
| "loss": 0.0012, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 3.629090909090909e-05, | |
| "loss": 0.0001, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.0116, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.570909090909091e-05, | |
| "loss": 0.0005, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 3.541818181818182e-05, | |
| "loss": 0.0013, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 3.512727272727273e-05, | |
| "loss": 0.0349, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 3.483636363636364e-05, | |
| "loss": 0.0013, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 3.454545454545455e-05, | |
| "loss": 0.0025, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.425454545454546e-05, | |
| "loss": 0.0106, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 3.396363636363637e-05, | |
| "loss": 0.0107, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 3.367272727272727e-05, | |
| "loss": 0.0055, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 3.338181818181818e-05, | |
| "loss": 0.0021, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 3.3090909090909095e-05, | |
| "loss": 0.0187, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 3.2800000000000004e-05, | |
| "loss": 0.0005, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 3.2509090909090914e-05, | |
| "loss": 0.0126, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 3.2218181818181816e-05, | |
| "loss": 0.0001, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 3.1927272727272726e-05, | |
| "loss": 0.0022, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 3.1636363636363635e-05, | |
| "loss": 0.0018, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 3.1345454545454545e-05, | |
| "loss": 0.001, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 3.105454545454546e-05, | |
| "loss": 0.0048, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 3.0763636363636364e-05, | |
| "loss": 0.0025, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 3.0472727272727276e-05, | |
| "loss": 0.0032, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 3.0181818181818182e-05, | |
| "loss": 0.0003, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 2.9890909090909092e-05, | |
| "loss": 0.0125, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 2.96e-05, | |
| "loss": 0.0009, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 2.9309090909090907e-05, | |
| "loss": 0.0007, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 2.9018181818181823e-05, | |
| "loss": 0.0206, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 2.872727272727273e-05, | |
| "loss": 0.007, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 2.843636363636364e-05, | |
| "loss": 0.0048, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 2.8145454545454548e-05, | |
| "loss": 0.0023, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 2.7854545454545454e-05, | |
| "loss": 0.0015, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 2.7563636363636364e-05, | |
| "loss": 0.0081, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 2.7272727272727273e-05, | |
| "loss": 0.0094, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 2.6981818181818186e-05, | |
| "loss": 0.0001, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 2.6690909090909095e-05, | |
| "loss": 0.01, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 2.64e-05, | |
| "loss": 0.0074, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 2.610909090909091e-05, | |
| "loss": 0.0017, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 2.581818181818182e-05, | |
| "loss": 0.0112, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 2.5527272727272726e-05, | |
| "loss": 0.0022, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 2.5236363636363636e-05, | |
| "loss": 0.0004, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 2.494545454545455e-05, | |
| "loss": 0.0, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 2.4654545454545454e-05, | |
| "loss": 0.0134, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 2.4363636363636364e-05, | |
| "loss": 0.0014, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 2.4072727272727273e-05, | |
| "loss": 0.0001, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 2.3781818181818183e-05, | |
| "loss": 0.0075, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 2.3490909090909092e-05, | |
| "loss": 0.0021, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 2.32e-05, | |
| "loss": 0.0159, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 2.290909090909091e-05, | |
| "loss": 0.0001, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 2.261818181818182e-05, | |
| "loss": 0.0011, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 2.2327272727272726e-05, | |
| "loss": 0.0052, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 2.203636363636364e-05, | |
| "loss": 0.0162, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 2.1745454545454545e-05, | |
| "loss": 0.0066, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 2.1454545454545455e-05, | |
| "loss": 0.0022, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 2.1163636363636367e-05, | |
| "loss": 0.0049, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 2.0872727272727273e-05, | |
| "loss": 0.0147, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 2.0581818181818183e-05, | |
| "loss": 0.0015, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 2.0290909090909092e-05, | |
| "loss": 0.0129, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 2e-05, | |
| "loss": 0.001, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.17399999499320984, | |
| "eval_loss": 4.792283058166504, | |
| "eval_runtime": 3.3921, | |
| "eval_samples_per_second": 147.401, | |
| "eval_steps_per_second": 4.717, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 1.970909090909091e-05, | |
| "loss": 0.0, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 1.9418181818181817e-05, | |
| "loss": 0.0005, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 1.9127272727272726e-05, | |
| "loss": 0.0, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 1.883636363636364e-05, | |
| "loss": 0.0, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 1.8545454545454545e-05, | |
| "loss": 0.0011, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 1.8254545454545455e-05, | |
| "loss": 0.0021, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 1.7963636363636364e-05, | |
| "loss": 0.001, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 1.7672727272727274e-05, | |
| "loss": 0.0001, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 1.7381818181818183e-05, | |
| "loss": 0.0044, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 1.7090909090909092e-05, | |
| "loss": 0.0018, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 1.6800000000000002e-05, | |
| "loss": 0.0012, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 1.650909090909091e-05, | |
| "loss": 0.0001, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 1.6218181818181817e-05, | |
| "loss": 0.0013, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 1.5927272727272727e-05, | |
| "loss": 0.0, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 1.563636363636364e-05, | |
| "loss": 0.0003, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 1.5345454545454545e-05, | |
| "loss": 0.0001, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 1.5054545454545455e-05, | |
| "loss": 0.0003, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 1.4763636363636366e-05, | |
| "loss": 0.019, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 1.4472727272727274e-05, | |
| "loss": 0.001, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 1.4181818181818181e-05, | |
| "loss": 0.0001, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 1.389090909090909e-05, | |
| "loss": 0.0001, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 1.3600000000000002e-05, | |
| "loss": 0.0, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 1.330909090909091e-05, | |
| "loss": 0.0, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 1.3018181818181819e-05, | |
| "loss": 0.0064, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 1.2727272727272727e-05, | |
| "loss": 0.0033, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 1.2436363636363636e-05, | |
| "loss": 0.0002, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 1.2145454545454546e-05, | |
| "loss": 0.0001, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 1.1854545454545455e-05, | |
| "loss": 0.0014, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 1.1563636363636364e-05, | |
| "loss": 0.002, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 1.1272727272727274e-05, | |
| "loss": 0.0003, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 1.0981818181818182e-05, | |
| "loss": 0.0127, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 1.0690909090909091e-05, | |
| "loss": 0.0007, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 1.04e-05, | |
| "loss": 0.0086, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 1.010909090909091e-05, | |
| "loss": 0.0, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 9.818181818181818e-06, | |
| "loss": 0.0072, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 9.527272727272727e-06, | |
| "loss": 0.0002, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 9.236363636363638e-06, | |
| "loss": 0.0, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 8.945454545454546e-06, | |
| "loss": 0.0, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 8.654545454545455e-06, | |
| "loss": 0.001, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 8.363636363636365e-06, | |
| "loss": 0.0001, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 8.072727272727274e-06, | |
| "loss": 0.0, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 7.781818181818182e-06, | |
| "loss": 0.001, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 7.490909090909091e-06, | |
| "loss": 0.0017, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 7.2e-06, | |
| "loss": 0.0, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 6.909090909090909e-06, | |
| "loss": 0.0, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 6.618181818181818e-06, | |
| "loss": 0.0143, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 6.327272727272728e-06, | |
| "loss": 0.0, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 6.0363636363636365e-06, | |
| "loss": 0.0, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 5.745454545454546e-06, | |
| "loss": 0.0029, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 5.4545454545454545e-06, | |
| "loss": 0.0018, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 5.163636363636364e-06, | |
| "loss": 0.0, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 4.872727272727273e-06, | |
| "loss": 0.0001, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 4.581818181818182e-06, | |
| "loss": 0.0001, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 4.290909090909091e-06, | |
| "loss": 0.0, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.0019, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 3.7090909090909092e-06, | |
| "loss": 0.0001, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 3.4181818181818182e-06, | |
| "loss": 0.0003, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 3.127272727272727e-06, | |
| "loss": 0.0004, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 2.8363636363636366e-06, | |
| "loss": 0.0037, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 2.5454545454545456e-06, | |
| "loss": 0.0015, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 2.2545454545454546e-06, | |
| "loss": 0.0002, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 1.9636363636363636e-06, | |
| "loss": 0.0144, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 1.6727272727272728e-06, | |
| "loss": 0.0052, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 1.3818181818181818e-06, | |
| "loss": 0.0005, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 1.090909090909091e-06, | |
| "loss": 0.0003, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 8.000000000000001e-07, | |
| "loss": 0.0082, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 5.090909090909092e-07, | |
| "loss": 0.0, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 2.181818181818182e-07, | |
| "loss": 0.0008, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.15800000727176666, | |
| "eval_loss": 4.613570213317871, | |
| "eval_runtime": 3.3639, | |
| "eval_samples_per_second": 148.638, | |
| "eval_steps_per_second": 4.756, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 6875, | |
| "total_flos": 2.323364869172544e+16, | |
| "train_loss": 0.06812760998129988, | |
| "train_runtime": 4390.0209, | |
| "train_samples_per_second": 50.114, | |
| "train_steps_per_second": 1.566 | |
| } | |
| ], | |
| "max_steps": 6875, | |
| "num_train_epochs": 5, | |
| "total_flos": 2.323364869172544e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |