| { | |
| "best_metric": 0.1498871147632599, | |
| "best_model_checkpoint": "seq2seq_results/combined_nl_prompt_base_features_contd_codet5small/checkpoint-144856", | |
| "epoch": 2.0, | |
| "global_step": 144856, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.001999861931849561, | |
| "loss": 0.1634, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.001999723863699122, | |
| "loss": 0.166, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.001999585795548683, | |
| "loss": 0.1659, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0019994477273982438, | |
| "loss": 0.1835, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0019993096592478047, | |
| "loss": 0.1772, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0019991715910973656, | |
| "loss": 0.182, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.001999033522946927, | |
| "loss": 0.1739, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0019988954547964875, | |
| "loss": 0.1776, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0019987573866460484, | |
| "loss": 0.1836, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0019986193184956093, | |
| "loss": 0.1808, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0019984812503451702, | |
| "loss": 0.181, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.001998343182194731, | |
| "loss": 0.1794, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.001998205114044292, | |
| "loss": 0.1722, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.001998067045893853, | |
| "loss": 0.1682, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0019979289777434144, | |
| "loss": 0.1707, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0019977909095929753, | |
| "loss": 0.1843, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0019976528414425362, | |
| "loss": 0.1775, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.001997514773292097, | |
| "loss": 0.1716, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.001997376705141658, | |
| "loss": 0.1756, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.001997238636991219, | |
| "loss": 0.1726, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00199710056884078, | |
| "loss": 0.17, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.001996962500690341, | |
| "loss": 0.1726, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.001996824432539902, | |
| "loss": 0.1782, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0019966863643894627, | |
| "loss": 0.1786, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0019965482962390237, | |
| "loss": 0.1706, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0019964102280885846, | |
| "loss": 0.1736, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0019962721599381455, | |
| "loss": 0.1806, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0019961340917877064, | |
| "loss": 0.1767, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0019959960236372674, | |
| "loss": 0.1743, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0019958579554868283, | |
| "loss": 0.1713, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0019957198873363892, | |
| "loss": 0.1722, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00199558181918595, | |
| "loss": 0.18, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001995443751035511, | |
| "loss": 0.1731, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.001995305682885072, | |
| "loss": 0.1893, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.001995167614734633, | |
| "loss": 0.1738, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0019950295465841943, | |
| "loss": 0.17, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.001994891478433755, | |
| "loss": 0.1699, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0019947534102833157, | |
| "loss": 0.178, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0019946153421328766, | |
| "loss": 0.1836, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0019944772739824376, | |
| "loss": 0.1805, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0019943392058319985, | |
| "loss": 0.1817, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0019942011376815594, | |
| "loss": 0.1776, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.001994063069531121, | |
| "loss": 0.1567, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0019939250013806817, | |
| "loss": 0.188, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0019937869332302426, | |
| "loss": 0.1683, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0019936488650798036, | |
| "loss": 0.1854, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0019935107969293645, | |
| "loss": 0.1762, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0019933727287789254, | |
| "loss": 0.1778, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0019932346606284864, | |
| "loss": 0.1819, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0019930965924780473, | |
| "loss": 0.1816, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.001992958524327608, | |
| "loss": 0.1701, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.001992820456177169, | |
| "loss": 0.1731, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00199268238802673, | |
| "loss": 0.1728, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.001992544319876291, | |
| "loss": 0.1754, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.001992406251725852, | |
| "loss": 0.1779, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.001992268183575413, | |
| "loss": 0.1757, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0019921301154249738, | |
| "loss": 0.1817, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.0019919920472745347, | |
| "loss": 0.1712, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0019918539791240956, | |
| "loss": 0.1643, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0019917159109736566, | |
| "loss": 0.1758, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.0019915778428232175, | |
| "loss": 0.1892, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0019914397746727784, | |
| "loss": 0.1809, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0019913017065223393, | |
| "loss": 0.1706, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.0019911636383719007, | |
| "loss": 0.1801, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.001991025570221461, | |
| "loss": 0.1791, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.001990887502071022, | |
| "loss": 0.1706, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.001990749433920583, | |
| "loss": 0.1665, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.001990611365770144, | |
| "loss": 0.1693, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.001990473297619705, | |
| "loss": 0.161, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.001990335229469266, | |
| "loss": 0.1749, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.001990197161318827, | |
| "loss": 0.1668, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.001990059093168388, | |
| "loss": 0.1624, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.001989921025017949, | |
| "loss": 0.1812, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00198978295686751, | |
| "loss": 0.1757, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.001989644888717071, | |
| "loss": 0.1766, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.001989506820566632, | |
| "loss": 0.1812, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0019893687524161928, | |
| "loss": 0.1695, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0019892306842657537, | |
| "loss": 0.1873, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0019890926161153146, | |
| "loss": 0.1799, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.0019889545479648755, | |
| "loss": 0.1707, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.0019888164798144365, | |
| "loss": 0.1724, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0019886784116639974, | |
| "loss": 0.1806, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0019885403435135583, | |
| "loss": 0.1688, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0019884022753631193, | |
| "loss": 0.1697, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00198826420721268, | |
| "loss": 0.1684, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.001988126139062241, | |
| "loss": 0.1685, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.001987988070911802, | |
| "loss": 0.1743, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.001987850002761363, | |
| "loss": 0.1663, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.001987711934610924, | |
| "loss": 0.1685, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.001987573866460485, | |
| "loss": 0.1866, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0019874357983100457, | |
| "loss": 0.1772, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.001987297730159607, | |
| "loss": 0.1829, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.001987159662009168, | |
| "loss": 0.1657, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0019870215938587285, | |
| "loss": 0.1681, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0019868835257082895, | |
| "loss": 0.174, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0019867454575578504, | |
| "loss": 0.1759, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0019866073894074113, | |
| "loss": 0.178, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0019864693212569722, | |
| "loss": 0.1679, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.0019863312531065336, | |
| "loss": 0.168, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0019861931849560945, | |
| "loss": 0.1697, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0019860551168056555, | |
| "loss": 0.1748, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.0019859170486552164, | |
| "loss": 0.1691, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0019857789805047773, | |
| "loss": 0.1885, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.0019856409123543382, | |
| "loss": 0.1734, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.001985502844203899, | |
| "loss": 0.1787, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00198536477605346, | |
| "loss": 0.1741, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.001985226707903021, | |
| "loss": 0.1672, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.001985088639752582, | |
| "loss": 0.1779, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.001984950571602143, | |
| "loss": 0.1697, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.001984812503451704, | |
| "loss": 0.1699, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0019846744353012647, | |
| "loss": 0.1693, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.0019845363671508257, | |
| "loss": 0.1718, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0019843982990003866, | |
| "loss": 0.161, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0019842602308499475, | |
| "loss": 0.1768, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0019841221626995084, | |
| "loss": 0.1793, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.0019839840945490694, | |
| "loss": 0.1802, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0019838460263986303, | |
| "loss": 0.1634, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0019837079582481912, | |
| "loss": 0.1551, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.001983569890097752, | |
| "loss": 0.1767, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.0019834318219473135, | |
| "loss": 0.1739, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0019832937537968744, | |
| "loss": 0.1747, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.0019831556856464354, | |
| "loss": 0.1696, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.001983017617495996, | |
| "loss": 0.1768, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0019828795493455568, | |
| "loss": 0.1829, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0019827414811951177, | |
| "loss": 0.1619, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0019826034130446786, | |
| "loss": 0.1698, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.00198246534489424, | |
| "loss": 0.1762, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.001982327276743801, | |
| "loss": 0.1795, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.001982189208593362, | |
| "loss": 0.1576, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0019820511404429228, | |
| "loss": 0.163, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0019819130722924837, | |
| "loss": 0.173, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.0019817750041420446, | |
| "loss": 0.1733, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0019816369359916056, | |
| "loss": 0.1681, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0019814988678411665, | |
| "loss": 0.1713, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.0019813607996907274, | |
| "loss": 0.1675, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0019812227315402883, | |
| "loss": 0.1612, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.0019810846633898493, | |
| "loss": 0.1699, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00198094659523941, | |
| "loss": 0.1705, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.001980808527088971, | |
| "loss": 0.1639, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.001980670458938532, | |
| "loss": 0.1716, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.001980532390788093, | |
| "loss": 0.1616, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.001980394322637654, | |
| "loss": 0.1685, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.001980256254487215, | |
| "loss": 0.1669, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0019801181863367758, | |
| "loss": 0.1651, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.15423069894313812, | |
| "eval_runtime": 111.7927, | |
| "eval_samples_per_second": 176.675, | |
| "eval_steps_per_second": 11.047, | |
| "step": 72428 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0019799801181863367, | |
| "loss": 0.1643, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0019798420500358976, | |
| "loss": 0.1803, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0019797039818854585, | |
| "loss": 0.164, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00197956591373502, | |
| "loss": 0.1629, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.001979427845584581, | |
| "loss": 0.1606, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0019792897774341418, | |
| "loss": 0.1683, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0019791517092837027, | |
| "loss": 0.1658, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.001979013641133263, | |
| "loss": 0.1736, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.001978875572982824, | |
| "loss": 0.1592, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.001978737504832385, | |
| "loss": 0.1653, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0019785994366819464, | |
| "loss": 0.1665, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0019784613685315073, | |
| "loss": 0.1649, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.0019783233003810683, | |
| "loss": 0.1724, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.001978185232230629, | |
| "loss": 0.1775, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00197804716408019, | |
| "loss": 0.1603, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.001977909095929751, | |
| "loss": 0.1765, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.001977771027779312, | |
| "loss": 0.1705, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.001977632959628873, | |
| "loss": 0.1513, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.001977494891478434, | |
| "loss": 0.173, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0019773568233279947, | |
| "loss": 0.1769, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.0019772187551775557, | |
| "loss": 0.1754, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0019770806870271166, | |
| "loss": 0.1647, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.0019769426188766775, | |
| "loss": 0.162, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.0019768045507262385, | |
| "loss": 0.1597, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0019766664825757994, | |
| "loss": 0.1738, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.0019765284144253603, | |
| "loss": 0.1822, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0019763903462749212, | |
| "loss": 0.1694, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.001976252278124482, | |
| "loss": 0.175, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.001976114209974043, | |
| "loss": 0.1691, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.001975976141823604, | |
| "loss": 0.1676, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.001975838073673165, | |
| "loss": 0.1582, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.0019757000055227263, | |
| "loss": 0.1743, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0019755619373722872, | |
| "loss": 0.1707, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.001975423869221848, | |
| "loss": 0.163, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.001975285801071409, | |
| "loss": 0.1704, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00197514773292097, | |
| "loss": 0.1646, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0019750096647705305, | |
| "loss": 0.1639, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0019748715966200914, | |
| "loss": 0.1767, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.001974733528469653, | |
| "loss": 0.1677, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0019745954603192137, | |
| "loss": 0.1703, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0019744573921687747, | |
| "loss": 0.1632, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.0019743193240183356, | |
| "loss": 0.1712, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0019741812558678965, | |
| "loss": 0.1863, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0019740431877174574, | |
| "loss": 0.1608, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0019739051195670184, | |
| "loss": 0.1566, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.0019737670514165793, | |
| "loss": 0.1641, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.0019736289832661402, | |
| "loss": 0.1684, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.001973490915115701, | |
| "loss": 0.1678, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.001973352846965262, | |
| "loss": 0.1801, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.001973214778814823, | |
| "loss": 0.1646, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.001973076710664384, | |
| "loss": 0.1612, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.001972938642513945, | |
| "loss": 0.169, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.001972800574363506, | |
| "loss": 0.1672, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0019726625062130667, | |
| "loss": 0.1688, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0019725244380626276, | |
| "loss": 0.1644, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.0019723863699121886, | |
| "loss": 0.1646, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0019722483017617495, | |
| "loss": 0.1592, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0019721102336113104, | |
| "loss": 0.1674, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0019719721654608714, | |
| "loss": 0.1654, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0019718340973104327, | |
| "loss": 0.1692, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0019716960291599936, | |
| "loss": 0.1719, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0019715579610095546, | |
| "loss": 0.1726, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.0019714198928591155, | |
| "loss": 0.1634, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0019712818247086764, | |
| "loss": 0.1683, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.001971143756558237, | |
| "loss": 0.1692, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.001971005688407798, | |
| "loss": 0.1781, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.001970867620257359, | |
| "loss": 0.1606, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00197072955210692, | |
| "loss": 0.1649, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.001970591483956481, | |
| "loss": 0.178, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.001970453415806042, | |
| "loss": 0.1699, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.001970315347655603, | |
| "loss": 0.1646, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.001970177279505164, | |
| "loss": 0.1666, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0019700392113547248, | |
| "loss": 0.1674, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0019699011432042857, | |
| "loss": 0.1777, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0019697630750538466, | |
| "loss": 0.161, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.0019696250069034076, | |
| "loss": 0.1633, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0019694869387529685, | |
| "loss": 0.1693, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.0019693488706025294, | |
| "loss": 0.172, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0019692108024520903, | |
| "loss": 0.1702, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.0019690727343016513, | |
| "loss": 0.163, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.001968934666151212, | |
| "loss": 0.1609, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.001968796598000773, | |
| "loss": 0.1706, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.001968658529850334, | |
| "loss": 0.1709, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.001968520461699895, | |
| "loss": 0.174, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.001968382393549456, | |
| "loss": 0.1648, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.001968244325399017, | |
| "loss": 0.1641, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.0019681062572485778, | |
| "loss": 0.1593, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.001967968189098139, | |
| "loss": 0.1733, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0019678301209477, | |
| "loss": 0.1598, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.001967692052797261, | |
| "loss": 0.1736, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.001967553984646822, | |
| "loss": 0.1672, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.001967415916496383, | |
| "loss": 0.1767, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0019672778483459438, | |
| "loss": 0.1765, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.0019671397801955042, | |
| "loss": 0.1591, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0019670017120450656, | |
| "loss": 0.1679, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.0019668636438946265, | |
| "loss": 0.1705, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.0019667255757441875, | |
| "loss": 0.1675, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.0019665875075937484, | |
| "loss": 0.1841, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0019664494394433093, | |
| "loss": 0.1748, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0019663113712928702, | |
| "loss": 0.1769, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.001966173303142431, | |
| "loss": 0.1677, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.001966035234991992, | |
| "loss": 0.1725, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.001965897166841553, | |
| "loss": 0.1695, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.001965759098691114, | |
| "loss": 0.1682, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.001965621030540675, | |
| "loss": 0.1682, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.001965482962390236, | |
| "loss": 0.1818, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.0019653448942397967, | |
| "loss": 0.1551, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0019652068260893577, | |
| "loss": 0.1666, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.0019650687579389186, | |
| "loss": 0.1636, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.0019649306897884795, | |
| "loss": 0.1565, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.0019647926216380404, | |
| "loss": 0.1732, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.0019646545534876014, | |
| "loss": 0.1628, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.0019645164853371623, | |
| "loss": 0.1686, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0019643784171867232, | |
| "loss": 0.1607, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.001964240349036284, | |
| "loss": 0.1744, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.0019641022808858455, | |
| "loss": 0.1642, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.0019639642127354064, | |
| "loss": 0.1794, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0019638261445849674, | |
| "loss": 0.1605, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0019636880764345283, | |
| "loss": 0.1702, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0019635500082840892, | |
| "loss": 0.1708, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.00196341194013365, | |
| "loss": 0.1629, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.001963273871983211, | |
| "loss": 0.1683, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.0019631358038327716, | |
| "loss": 0.1634, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.001962997735682333, | |
| "loss": 0.1713, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.001962859667531894, | |
| "loss": 0.1577, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.001962721599381455, | |
| "loss": 0.1633, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.0019625835312310157, | |
| "loss": 0.1564, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0019624454630805767, | |
| "loss": 0.1676, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0019623073949301376, | |
| "loss": 0.1706, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.0019621693267796985, | |
| "loss": 0.1764, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0019620312586292594, | |
| "loss": 0.1725, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0019618931904788204, | |
| "loss": 0.1721, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0019617551223283813, | |
| "loss": 0.1639, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.001961617054177942, | |
| "loss": 0.1651, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.001961478986027503, | |
| "loss": 0.1678, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.001961340917877064, | |
| "loss": 0.1633, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.001961202849726625, | |
| "loss": 0.1711, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.001961064781576186, | |
| "loss": 0.1872, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.001960926713425747, | |
| "loss": 0.1716, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0019607886452753078, | |
| "loss": 0.1517, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0019606505771248687, | |
| "loss": 0.1743, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0019605125089744296, | |
| "loss": 0.1698, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.0019603744408239906, | |
| "loss": 0.1692, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.0019602363726735515, | |
| "loss": 0.1721, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.001960098304523113, | |
| "loss": 0.1627, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.1498871147632599, | |
| "eval_runtime": 111.769, | |
| "eval_samples_per_second": 176.713, | |
| "eval_steps_per_second": 11.05, | |
| "step": 144856 | |
| } | |
| ], | |
| "max_steps": 7242800, | |
| "num_train_epochs": 100, | |
| "total_flos": 3.136735727635661e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |