| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.3959733513339, | |
| "global_step": 31500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8532519370744304e-05, | |
| "loss": 3.2928, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.706503874148862e-05, | |
| "loss": 3.2304, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.559755811223292e-05, | |
| "loss": 3.2047, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.413007748297723e-05, | |
| "loss": 3.1857, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.266259685372153e-05, | |
| "loss": 3.1708, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.1195116224465844e-05, | |
| "loss": 3.162, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.9727635595210145e-05, | |
| "loss": 3.1554, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.826015496595445e-05, | |
| "loss": 3.1443, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.6792674336698754e-05, | |
| "loss": 3.1349, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.532519370744307e-05, | |
| "loss": 3.1265, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.385771307818737e-05, | |
| "loss": 3.1208, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 3.239023244893167e-05, | |
| "loss": 3.1183, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 3.092275181967598e-05, | |
| "loss": 3.1154, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.9455271190420287e-05, | |
| "loss": 3.109, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.7987790561164595e-05, | |
| "loss": 3.1061, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.65203099319089e-05, | |
| "loss": 3.1059, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.5052829302653208e-05, | |
| "loss": 3.1018, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.3585348673397513e-05, | |
| "loss": 3.0926, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 2.2117868044141817e-05, | |
| "loss": 3.0933, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 2.0650387414886125e-05, | |
| "loss": 3.0893, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 1.918290678563043e-05, | |
| "loss": 3.0922, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 1.7715426156374738e-05, | |
| "loss": 3.0856, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 1.6247945527119042e-05, | |
| "loss": 3.0831, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.478046489786335e-05, | |
| "loss": 3.0846, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.3312984268607657e-05, | |
| "loss": 3.0803, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.1845503639351961e-05, | |
| "loss": 3.0762, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.0378023010096268e-05, | |
| "loss": 3.0781, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 8.910542380840574e-06, | |
| "loss": 3.0788, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 7.443061751584879e-06, | |
| "loss": 3.0742, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 5.975581122329185e-06, | |
| "loss": 3.0775, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.508100493073491e-06, | |
| "loss": 3.0749, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 3.0406198638177975e-06, | |
| "loss": 3.0721, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 1.573139234562104e-06, | |
| "loss": 3.0719, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 1.0565860530640996e-07, | |
| "loss": 3.0744, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 2.431908898802536e-05, | |
| "loss": 3.0706, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 2.3585348673397513e-05, | |
| "loss": 3.0717, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 2.2851608358769667e-05, | |
| "loss": 3.0702, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 2.2117868044141817e-05, | |
| "loss": 3.0674, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 2.138412772951397e-05, | |
| "loss": 3.066, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 2.0650387414886125e-05, | |
| "loss": 3.065, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 1.991664710025828e-05, | |
| "loss": 3.0648, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 1.918290678563043e-05, | |
| "loss": 3.0624, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 1.8449166471002584e-05, | |
| "loss": 3.0609, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 1.7715426156374738e-05, | |
| "loss": 3.0593, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 1.698168584174689e-05, | |
| "loss": 3.0543, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 1.6247945527119042e-05, | |
| "loss": 3.0545, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 1.5514205212491196e-05, | |
| "loss": 3.0564, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 1.478046489786335e-05, | |
| "loss": 3.0551, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 5.75, | |
| "learning_rate": 1.4046724583235501e-05, | |
| "loss": 3.0548, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 1.3312984268607657e-05, | |
| "loss": 3.0502, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 1.2579243953979807e-05, | |
| "loss": 3.0462, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 1.1845503639351961e-05, | |
| "loss": 3.0477, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 1.1111763324724114e-05, | |
| "loss": 3.0476, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 1.0378023010096268e-05, | |
| "loss": 3.0438, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 9.64428269546842e-06, | |
| "loss": 3.0464, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 8.910542380840574e-06, | |
| "loss": 3.0488, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 8.176802066212726e-06, | |
| "loss": 3.0451, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 7.443061751584879e-06, | |
| "loss": 3.0446, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 6.709321436957032e-06, | |
| "loss": 3.0454, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 5.975581122329185e-06, | |
| "loss": 3.044, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 5.241840807701338e-06, | |
| "loss": 3.0445, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 4.508100493073491e-06, | |
| "loss": 3.0398, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 3.7743601784456448e-06, | |
| "loss": 3.0409, | |
| "step": 31500 | |
| } | |
| ], | |
| "max_steps": 34072, | |
| "num_train_epochs": 8, | |
| "total_flos": 1.2840071685382656e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |