| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.22117777163395078, | |
| "global_step": 400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9999895164082156e-05, | |
| "loss": 1.6349, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999958065720787e-05, | |
| "loss": 1.6199, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999905648201487e-05, | |
| "loss": 1.4834, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999832264289934e-05, | |
| "loss": 1.3882, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999737914601591e-05, | |
| "loss": 1.3679, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999622599927756e-05, | |
| "loss": 1.2396, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999486321235559e-05, | |
| "loss": 1.321, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9993290796679516e-05, | |
| "loss": 1.2874, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.999150876543699e-05, | |
| "loss": 1.2607, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9989517133573694e-05, | |
| "loss": 1.2454, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9987315917793174e-05, | |
| "loss": 1.2799, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.998490513655676e-05, | |
| "loss": 1.2575, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.998228481008337e-05, | |
| "loss": 1.2404, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.997945496034934e-05, | |
| "loss": 1.2219, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9976415611088267e-05, | |
| "loss": 1.2241, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.997316678779079e-05, | |
| "loss": 1.1716, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.996970851770438e-05, | |
| "loss": 1.1883, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9966040829833115e-05, | |
| "loss": 1.205, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9962163754937426e-05, | |
| "loss": 1.1246, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.995807732553384e-05, | |
| "loss": 1.1636, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9953781575894723e-05, | |
| "loss": 1.158, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9949276542048e-05, | |
| "loss": 1.1477, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9944562261776805e-05, | |
| "loss": 1.1678, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9939638774619216e-05, | |
| "loss": 1.1501, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.99345061218679e-05, | |
| "loss": 1.1955, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9929164346569756e-05, | |
| "loss": 1.1724, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.9923613493525576e-05, | |
| "loss": 1.177, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.991785360928968e-05, | |
| "loss": 1.1418, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.991188474216947e-05, | |
| "loss": 1.1898, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.9905706942225094e-05, | |
| "loss": 1.1479, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9899320261268966e-05, | |
| "loss": 1.1356, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.989272475286537e-05, | |
| "loss": 1.1397, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9885920472330004e-05, | |
| "loss": 1.1215, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9878907476729516e-05, | |
| "loss": 1.167, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9871685824881e-05, | |
| "loss": 1.1219, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9864255577351534e-05, | |
| "loss": 1.0835, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.985661679645769e-05, | |
| "loss": 1.0721, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9848769546264915e-05, | |
| "loss": 1.0692, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9840713892587146e-05, | |
| "loss": 1.0488, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.983244990298609e-05, | |
| "loss": 1.1285, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.982397764677081e-05, | |
| "loss": 1.0832, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.981529719499704e-05, | |
| "loss": 1.0652, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.980640862046663e-05, | |
| "loss": 1.1043, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.979731199772693e-05, | |
| "loss": 1.112, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9788007403070146e-05, | |
| "loss": 1.1029, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.977849491453277e-05, | |
| "loss": 1.0869, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.976877461189481e-05, | |
| "loss": 1.0843, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.975884657667922e-05, | |
| "loss": 1.0789, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.974871089215118e-05, | |
| "loss": 1.0449, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9738367643317405e-05, | |
| "loss": 1.1053, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9727816916925395e-05, | |
| "loss": 1.0651, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.971705880146276e-05, | |
| "loss": 1.0828, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.970609338715646e-05, | |
| "loss": 1.0932, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.969492076597203e-05, | |
| "loss": 1.0648, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.968354103161283e-05, | |
| "loss": 1.0948, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.967195427951926e-05, | |
| "loss": 1.0721, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9660160606867936e-05, | |
| "loss": 1.124, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9648160112570896e-05, | |
| "loss": 1.0963, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.9635952897274773e-05, | |
| "loss": 1.1078, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9623539063359925e-05, | |
| "loss": 1.1059, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.961091871493962e-05, | |
| "loss": 1.1032, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.959809195785912e-05, | |
| "loss": 1.0595, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.958505889969481e-05, | |
| "loss": 1.1096, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.957181964975329e-05, | |
| "loss": 1.0589, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.955837431907049e-05, | |
| "loss": 1.0608, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.954472302041069e-05, | |
| "loss": 1.0819, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.9530865868265605e-05, | |
| "loss": 1.0759, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.951680297885342e-05, | |
| "loss": 1.0515, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.950253447011779e-05, | |
| "loss": 1.0371, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.948806046172691e-05, | |
| "loss": 1.0619, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.947338107507245e-05, | |
| "loss": 1.0757, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.945849643326857e-05, | |
| "loss": 1.0686, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.9443406661150874e-05, | |
| "loss": 1.0809, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.942811188527537e-05, | |
| "loss": 1.0704, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.941261223391742e-05, | |
| "loss": 1.0655, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.939690783707063e-05, | |
| "loss": 1.1182, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.938099882644578e-05, | |
| "loss": 1.081, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.9364885335469734e-05, | |
| "loss": 1.0792, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.93485674992843e-05, | |
| "loss": 1.0244, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.933204545474511e-05, | |
| "loss": 1.0531, | |
| "step": 400 | |
| } | |
| ], | |
| "max_steps": 5424, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.29988124737536e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |