| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 41090, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9391579459722564e-05, | |
| "loss": 1.7982, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.878315891944512e-05, | |
| "loss": 1.2914, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.817473837916768e-05, | |
| "loss": 1.1456, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.756631783889024e-05, | |
| "loss": 1.0345, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.6957897298612804e-05, | |
| "loss": 0.9665, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.6349476758335365e-05, | |
| "loss": 0.9272, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.574105621805793e-05, | |
| "loss": 0.8571, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 4.513263567778048e-05, | |
| "loss": 0.8139, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.4524215137503043e-05, | |
| "loss": 0.6986, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.3915794597225605e-05, | |
| "loss": 0.6375, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 4.330737405694817e-05, | |
| "loss": 0.6407, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.269895351667073e-05, | |
| "loss": 0.6139, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 4.209053297639329e-05, | |
| "loss": 0.5973, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.1482112436115845e-05, | |
| "loss": 0.6094, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.0873691895838406e-05, | |
| "loss": 0.5824, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 4.026527135556097e-05, | |
| "loss": 0.5459, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 3.965685081528352e-05, | |
| "loss": 0.4764, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 3.9048430275006084e-05, | |
| "loss": 0.427, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 3.8440009734728646e-05, | |
| "loss": 0.4192, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.783158919445121e-05, | |
| "loss": 0.4278, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.722316865417377e-05, | |
| "loss": 0.4157, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 3.6614748113896324e-05, | |
| "loss": 0.4156, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.6006327573618886e-05, | |
| "loss": 0.4137, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 3.539790703334145e-05, | |
| "loss": 0.3731, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 3.4789486493064e-05, | |
| "loss": 0.3516, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 3.418106595278657e-05, | |
| "loss": 0.3098, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 3.357264541250913e-05, | |
| "loss": 0.3051, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 3.296422487223169e-05, | |
| "loss": 0.3029, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 3.235580433195425e-05, | |
| "loss": 0.3107, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 3.174738379167681e-05, | |
| "loss": 0.2969, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 3.1138963251399365e-05, | |
| "loss": 0.2886, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 3.053054271112193e-05, | |
| "loss": 0.2887, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 2.9922122170844492e-05, | |
| "loss": 0.2803, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 2.9313701630567047e-05, | |
| "loss": 0.2256, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 2.8705281090289608e-05, | |
| "loss": 0.2209, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 2.8096860550012173e-05, | |
| "loss": 0.2197, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.7488440009734728e-05, | |
| "loss": 0.2305, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 2.688001946945729e-05, | |
| "loss": 0.2252, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.627159892917985e-05, | |
| "loss": 0.207, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 2.566317838890241e-05, | |
| "loss": 0.2187, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 2.505475784862497e-05, | |
| "loss": 0.207, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.444633730834753e-05, | |
| "loss": 0.1685, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 2.383791676807009e-05, | |
| "loss": 0.1772, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 2.3229496227792653e-05, | |
| "loss": 0.1728, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 2.262107568751521e-05, | |
| "loss": 0.1678, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 2.201265514723777e-05, | |
| "loss": 0.1642, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 2.1404234606960334e-05, | |
| "loss": 0.1664, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 2.0795814066682892e-05, | |
| "loss": 0.1575, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 2.018739352640545e-05, | |
| "loss": 0.1599, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 1.9578972986128012e-05, | |
| "loss": 0.1249, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 6.21, | |
| "learning_rate": 1.8970552445850574e-05, | |
| "loss": 0.1266, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 1.8362131905573132e-05, | |
| "loss": 0.1304, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 1.7753711365295694e-05, | |
| "loss": 0.1161, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 1.7145290825018255e-05, | |
| "loss": 0.1253, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 1.6536870284740814e-05, | |
| "loss": 0.1215, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 1.5928449744463375e-05, | |
| "loss": 0.1158, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 1.5320029204185933e-05, | |
| "loss": 0.1219, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 7.06, | |
| "learning_rate": 1.4711608663908493e-05, | |
| "loss": 0.1039, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 1.4103188123631053e-05, | |
| "loss": 0.081, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 1.3494767583353615e-05, | |
| "loss": 0.0922, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 1.2886347043076175e-05, | |
| "loss": 0.0844, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 1.2277926502798735e-05, | |
| "loss": 0.0759, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 1.1669505962521295e-05, | |
| "loss": 0.0882, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 1.1061085422243855e-05, | |
| "loss": 0.0856, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 7.91, | |
| "learning_rate": 1.0452664881966416e-05, | |
| "loss": 0.0881, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 8.03, | |
| "learning_rate": 9.844244341688976e-06, | |
| "loss": 0.0708, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 9.235823801411536e-06, | |
| "loss": 0.0538, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 8.627403261134098e-06, | |
| "loss": 0.0554, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 8.018982720856656e-06, | |
| "loss": 0.06, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 7.410562180579217e-06, | |
| "loss": 0.0514, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 6.802141640301777e-06, | |
| "loss": 0.0536, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 6.193721100024337e-06, | |
| "loss": 0.0583, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 5.585300559746897e-06, | |
| "loss": 0.052, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 4.976880019469458e-06, | |
| "loss": 0.0481, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 9.13, | |
| "learning_rate": 4.368459479192018e-06, | |
| "loss": 0.0378, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 3.760038938914578e-06, | |
| "loss": 0.0296, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 3.151618398637138e-06, | |
| "loss": 0.037, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 9.49, | |
| "learning_rate": 2.543197858359698e-06, | |
| "loss": 0.0339, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 9.61, | |
| "learning_rate": 1.9347773180822585e-06, | |
| "loss": 0.0369, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 1.3263567778048189e-06, | |
| "loss": 0.0344, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 7.17936237527379e-07, | |
| "loss": 0.0301, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 1.0951569724993917e-07, | |
| "loss": 0.0325, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 41090, | |
| "total_flos": 1.1825796643443088e+17, | |
| "train_loss": 0.30929526851019973, | |
| "train_runtime": 55900.09, | |
| "train_samples_per_second": 11.761, | |
| "train_steps_per_second": 0.735 | |
| } | |
| ], | |
| "max_steps": 41090, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.1825796643443088e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |