| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9893622418779279, | |
| "global_step": 58500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9585581623505137e-05, | |
| "loss": 0.6366, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9162705729122623e-05, | |
| "loss": 0.2794, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.8739829834740104e-05, | |
| "loss": 0.2335, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.8316953940357584e-05, | |
| "loss": 0.2659, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.789407804597507e-05, | |
| "loss": 0.2433, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.747120215159255e-05, | |
| "loss": 0.2383, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.704832625721004e-05, | |
| "loss": 0.2382, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.6625450362827524e-05, | |
| "loss": 0.2224, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.6202574468445005e-05, | |
| "loss": 0.2412, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.5779698574062485e-05, | |
| "loss": 0.2221, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.535682267967997e-05, | |
| "loss": 0.2365, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.493394678529745e-05, | |
| "loss": 0.206, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.451107089091494e-05, | |
| "loss": 0.2189, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.408819499653242e-05, | |
| "loss": 0.2206, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.3665319102149906e-05, | |
| "loss": 0.2238, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.3242443207767386e-05, | |
| "loss": 0.2127, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.2819567313384866e-05, | |
| "loss": 0.2103, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.239669141900235e-05, | |
| "loss": 0.216, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.197381552461983e-05, | |
| "loss": 0.1916, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.155093963023732e-05, | |
| "loss": 0.2208, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.1128063735854807e-05, | |
| "loss": 0.1848, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.070518784147229e-05, | |
| "loss": 0.2126, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.028231194708977e-05, | |
| "loss": 0.1963, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.9859436052707254e-05, | |
| "loss": 0.2159, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.9436560158324734e-05, | |
| "loss": 0.1872, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 3.901368426394222e-05, | |
| "loss": 0.2261, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 3.859080836955971e-05, | |
| "loss": 0.2147, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.816793247517719e-05, | |
| "loss": 0.1938, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.774505658079467e-05, | |
| "loss": 0.1938, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 3.7322180686412155e-05, | |
| "loss": 0.1935, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 3.6899304792029635e-05, | |
| "loss": 0.209, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 3.647642889764712e-05, | |
| "loss": 0.1981, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 3.605355300326461e-05, | |
| "loss": 0.2031, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 3.563067710888209e-05, | |
| "loss": 0.2045, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.520780121449957e-05, | |
| "loss": 0.1931, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 3.478492532011705e-05, | |
| "loss": 0.1936, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 3.4362049425734536e-05, | |
| "loss": 0.2128, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.393917353135202e-05, | |
| "loss": 0.1841, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 3.35162976369695e-05, | |
| "loss": 0.193, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 3.309342174258699e-05, | |
| "loss": 0.1961, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 3.267054584820447e-05, | |
| "loss": 0.1913, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.224766995382195e-05, | |
| "loss": 0.2081, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.182479405943944e-05, | |
| "loss": 0.1866, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.1401918165056924e-05, | |
| "loss": 0.1884, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.0979042270674404e-05, | |
| "loss": 0.1936, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.055616637629189e-05, | |
| "loss": 0.2044, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.013329048190937e-05, | |
| "loss": 0.1916, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.971041458752685e-05, | |
| "loss": 0.1787, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.928753869314434e-05, | |
| "loss": 0.2117, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 2.886466279876182e-05, | |
| "loss": 0.1983, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 2.84417869043793e-05, | |
| "loss": 0.1947, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 2.801891100999679e-05, | |
| "loss": 0.1919, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 2.7596035115614272e-05, | |
| "loss": 0.1799, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.7173159221231752e-05, | |
| "loss": 0.1746, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.675028332684924e-05, | |
| "loss": 0.1768, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 2.6327407432466722e-05, | |
| "loss": 0.1758, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.5904531538084202e-05, | |
| "loss": 0.1882, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 2.5481655643701686e-05, | |
| "loss": 0.1816, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.5058779749319173e-05, | |
| "loss": 0.197, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.4635903854936653e-05, | |
| "loss": 0.1672, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.421302796055414e-05, | |
| "loss": 0.1709, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.379015206617162e-05, | |
| "loss": 0.1688, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.3367276171789103e-05, | |
| "loss": 0.1945, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.2944400277406587e-05, | |
| "loss": 0.1888, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.252152438302407e-05, | |
| "loss": 0.1634, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.2098648488641554e-05, | |
| "loss": 0.2004, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.1675772594259037e-05, | |
| "loss": 0.1886, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.125289669987652e-05, | |
| "loss": 0.1814, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.0830020805494004e-05, | |
| "loss": 0.1939, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.0407144911111488e-05, | |
| "loss": 0.1932, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.998426901672897e-05, | |
| "loss": 0.1694, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9561393122346455e-05, | |
| "loss": 0.1883, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.913851722796394e-05, | |
| "loss": 0.1623, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8715641333581422e-05, | |
| "loss": 0.1703, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8292765439198905e-05, | |
| "loss": 0.1795, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7869889544816386e-05, | |
| "loss": 0.1873, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.7447013650433872e-05, | |
| "loss": 0.1657, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.7024137756051356e-05, | |
| "loss": 0.1806, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.6601261861668836e-05, | |
| "loss": 0.1701, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.6178385967286323e-05, | |
| "loss": 0.1646, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5755510072903806e-05, | |
| "loss": 0.1686, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5332634178521287e-05, | |
| "loss": 0.1752, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.4909758284138772e-05, | |
| "loss": 0.1631, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.4486882389756254e-05, | |
| "loss": 0.1645, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.4064006495373739e-05, | |
| "loss": 0.1815, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.3641130600991222e-05, | |
| "loss": 0.1928, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.3218254706608704e-05, | |
| "loss": 0.1637, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.279537881222619e-05, | |
| "loss": 0.1658, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.2372502917843671e-05, | |
| "loss": 0.1754, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.1949627023461155e-05, | |
| "loss": 0.1787, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.152675112907864e-05, | |
| "loss": 0.1499, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1103875234696122e-05, | |
| "loss": 0.1659, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0680999340313605e-05, | |
| "loss": 0.175, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0258123445931089e-05, | |
| "loss": 0.1659, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.835247551548572e-06, | |
| "loss": 0.1904, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.412371657166056e-06, | |
| "loss": 0.1736, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.989495762783539e-06, | |
| "loss": 0.1873, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 8.566619868401023e-06, | |
| "loss": 0.1809, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 8.143743974018506e-06, | |
| "loss": 0.1714, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.720868079635988e-06, | |
| "loss": 0.162, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 7.297992185253471e-06, | |
| "loss": 0.1758, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 6.875116290870956e-06, | |
| "loss": 0.1705, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 6.452240396488439e-06, | |
| "loss": 0.1632, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 6.029364502105922e-06, | |
| "loss": 0.1767, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.606488607723405e-06, | |
| "loss": 0.1705, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.183612713340889e-06, | |
| "loss": 0.1636, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 4.760736818958372e-06, | |
| "loss": 0.1666, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.337860924575856e-06, | |
| "loss": 0.1765, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.9149850301933386e-06, | |
| "loss": 0.1634, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.4921091358108225e-06, | |
| "loss": 0.159, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.0692332414283056e-06, | |
| "loss": 0.1725, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.646357347045789e-06, | |
| "loss": 0.1568, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.223481452663272e-06, | |
| "loss": 0.165, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.800605558280756e-06, | |
| "loss": 0.1825, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3777296638982391e-06, | |
| "loss": 0.1646, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.548537695157224e-07, | |
| "loss": 0.173, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.319778751332059e-07, | |
| "loss": 0.1723, | |
| "step": 58500 | |
| } | |
| ], | |
| "max_steps": 59129, | |
| "num_train_epochs": 1, | |
| "total_flos": 1.27532698435584e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |