| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.71882775781034, | |
| "global_step": 1300, | |
| "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 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.9315319340420465e-05, | |
| "loss": 1.048, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.929838929659015e-05, | |
| "loss": 1.04, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.9281255465244314e-05, | |
| "loss": 1.0686, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.926391799008223e-05, | |
| "loss": 1.0933, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.924637701651111e-05, | |
| "loss": 1.0313, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.9228632691644874e-05, | |
| "loss": 1.0811, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.921068516430293e-05, | |
| "loss": 1.0617, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.919253458500892e-05, | |
| "loss": 1.0482, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.9174181105989445e-05, | |
| "loss": 1.0681, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.9155624881172834e-05, | |
| "loss": 1.0476, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.913686606618777e-05, | |
| "loss": 1.0463, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.911790481836208e-05, | |
| "loss": 1.0213, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.909874129672133e-05, | |
| "loss": 0.9855, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.907937566198757e-05, | |
| "loss": 1.045, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.9059808076577914e-05, | |
| "loss": 1.0442, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.904003870460323e-05, | |
| "loss": 1.0354, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.9020067711866735e-05, | |
| "loss": 1.0331, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.899989526586261e-05, | |
| "loss": 1.0585, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.8979521535774636e-05, | |
| "loss": 1.0223, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.895894669247468e-05, | |
| "loss": 1.0118, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.8938170908521356e-05, | |
| "loss": 1.0508, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.8917194358158534e-05, | |
| "loss": 1.0656, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.8896017217313886e-05, | |
| "loss": 1.0655, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.887463966359741e-05, | |
| "loss": 1.0833, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.8853061876299956e-05, | |
| "loss": 1.068, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.8831284036391684e-05, | |
| "loss": 1.0487, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.880930632652058e-05, | |
| "loss": 1.059, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.878712893101092e-05, | |
| "loss": 1.0408, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.876475203586171e-05, | |
| "loss": 0.9976, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.874217582874514e-05, | |
| "loss": 1.0358, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.8719400499005e-05, | |
| "loss": 1.0073, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.869642623765509e-05, | |
| "loss": 1.0742, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.867325323737765e-05, | |
| "loss": 1.0463, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.864988169252168e-05, | |
| "loss": 1.0242, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.8626311799101375e-05, | |
| "loss": 1.0773, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.860254375479446e-05, | |
| "loss": 1.0371, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.8578577758940504e-05, | |
| "loss": 1.0399, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.855441401253928e-05, | |
| "loss": 1.0411, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.8530052718249076e-05, | |
| "loss": 1.0476, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.850549408038498e-05, | |
| "loss": 1.0075, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.848073830491717e-05, | |
| "loss": 1.0045, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.845578559946923e-05, | |
| "loss": 1.0368, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.8430636173316306e-05, | |
| "loss": 1.0156, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.840529023738348e-05, | |
| "loss": 1.0334, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.837974800424389e-05, | |
| "loss": 0.9994, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.8354009688117026e-05, | |
| "loss": 1.0467, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.8328075504866874e-05, | |
| "loss": 0.9968, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.8301945672000164e-05, | |
| "loss": 1.041, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.8275620408664487e-05, | |
| "loss": 0.9811, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.8249099935646494e-05, | |
| "loss": 1.0235, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.822238447537003e-05, | |
| "loss": 1.0152, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.819547425189429e-05, | |
| "loss": 1.038, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.81683694909119e-05, | |
| "loss": 1.0387, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.814107041974707e-05, | |
| "loss": 1.019, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.811357726735366e-05, | |
| "loss": 1.0403, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.808589026431324e-05, | |
| "loss": 1.0828, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.805800964283322e-05, | |
| "loss": 1.0128, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.802993563674483e-05, | |
| "loss": 1.0502, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.80016684815012e-05, | |
| "loss": 1.0169, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.7973208414175406e-05, | |
| "loss": 1.0316, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.794455567345842e-05, | |
| "loss": 1.0263, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.791571049965714e-05, | |
| "loss": 1.0801, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.7886673134692404e-05, | |
| "loss": 1.0575, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.7857443822096905e-05, | |
| "loss": 0.9767, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.782802280701319e-05, | |
| "loss": 0.9666, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.77984103361916e-05, | |
| "loss": 0.9983, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.776860665798816e-05, | |
| "loss": 1.0219, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.773861202236257e-05, | |
| "loss": 0.9963, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.770842668087602e-05, | |
| "loss": 1.0219, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.767805088668916e-05, | |
| "loss": 1.0029, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.7647484894559936e-05, | |
| "loss": 1.0077, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.7616728960841444e-05, | |
| "loss": 0.9908, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.758578334347981e-05, | |
| "loss": 1.0092, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.7554648302012015e-05, | |
| "loss": 0.9874, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.7523324097563706e-05, | |
| "loss": 1.0185, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.749181099284703e-05, | |
| "loss": 1.0533, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.746010925215839e-05, | |
| "loss": 1.0083, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.74282191413763e-05, | |
| "loss": 1.0191, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.7396140927959045e-05, | |
| "loss": 0.9786, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.7363874880942574e-05, | |
| "loss": 1.0197, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.733142127093813e-05, | |
| "loss": 1.0471, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.7298780370130014e-05, | |
| "loss": 1.0134, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.726595245227336e-05, | |
| "loss": 0.9801, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.723293779269173e-05, | |
| "loss": 0.9944, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.7199736668274924e-05, | |
| "loss": 1.0358, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.716634935747655e-05, | |
| "loss": 0.9907, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.713277614031177e-05, | |
| "loss": 1.0698, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.70990172983549e-05, | |
| "loss": 1.0466, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.706507311473707e-05, | |
| "loss": 0.9489, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.703094387414385e-05, | |
| "loss": 0.9936, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.699662986281288e-05, | |
| "loss": 0.9672, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.696213136853141e-05, | |
| "loss": 1.0164, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.6927448680633954e-05, | |
| "loss": 0.9817, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.689258208999983e-05, | |
| "loss": 1.028, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.6857531889050716e-05, | |
| "loss": 1.0239, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.682229837174821e-05, | |
| "loss": 1.0378, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.678688183359135e-05, | |
| "loss": 0.9908, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.675128257161418e-05, | |
| "loss": 0.9384, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.671550088438319e-05, | |
| "loss": 0.9481, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.6679537071994874e-05, | |
| "loss": 0.9541, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.6643391436073165e-05, | |
| "loss": 0.9787, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.660706427976693e-05, | |
| "loss": 0.9897, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.657055590774745e-05, | |
| "loss": 0.9947, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.6533866626205805e-05, | |
| "loss": 0.9614, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.649699674285036e-05, | |
| "loss": 1.0323, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.645994656690417e-05, | |
| "loss": 1.0345, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.642271640910235e-05, | |
| "loss": 1.0432, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.638530658168954e-05, | |
| "loss": 1.0102, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.6347717398417203e-05, | |
| "loss": 1.0361, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.6309949174541096e-05, | |
| "loss": 0.9818, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.627200222681851e-05, | |
| "loss": 1.0293, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.6233876873505694e-05, | |
| "loss": 1.0125, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.619557343435516e-05, | |
| "loss": 1.0039, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.615709223061302e-05, | |
| "loss": 1.0146, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.611843358501624e-05, | |
| "loss": 1.0352, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.6079597821789993e-05, | |
| "loss": 1.0405, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.604058526664491e-05, | |
| "loss": 0.9834, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.600139624677436e-05, | |
| "loss": 1.0252, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.596203109085168e-05, | |
| "loss": 1.0516, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.5922490129027464e-05, | |
| "loss": 0.9762, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.588277369292674e-05, | |
| "loss": 0.9801, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.5842882115646234e-05, | |
| "loss": 1.0127, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.580281573175157e-05, | |
| "loss": 1.0415, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.576257487727442e-05, | |
| "loss": 0.9974, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.572215988970974e-05, | |
| "loss": 1.0097, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.568157110801293e-05, | |
| "loss": 1.0004, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.5640808872596944e-05, | |
| "loss": 0.9932, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.5599873525329505e-05, | |
| "loss": 0.9437, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.555876540953019e-05, | |
| "loss": 1.0004, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.551748486996755e-05, | |
| "loss": 0.9743, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.547603225285626e-05, | |
| "loss": 1.0303, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.543440790585417e-05, | |
| "loss": 1.0225, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.539261217805939e-05, | |
| "loss": 1.0102, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.535064542000743e-05, | |
| "loss": 1.0087, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.5308507983668165e-05, | |
| "loss": 0.952, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.526620022244293e-05, | |
| "loss": 0.9767, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.522372249116158e-05, | |
| "loss": 1.0049, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.5181075146079456e-05, | |
| "loss": 1.0397, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.5138258544874455e-05, | |
| "loss": 1.0167, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.5095273046643985e-05, | |
| "loss": 0.9966, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.5052119011901986e-05, | |
| "loss": 0.983, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.500879680257587e-05, | |
| "loss": 1.0301, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.4965306782003535e-05, | |
| "loss": 1.0371, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.492164931493028e-05, | |
| "loss": 1.0083, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.487782476750575e-05, | |
| "loss": 0.9728, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.4833833507280884e-05, | |
| "loss": 0.9892, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.4789675903204805e-05, | |
| "loss": 0.9847, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.474535232562176e-05, | |
| "loss": 1.0545, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 4.470086314626797e-05, | |
| "loss": 1.0204, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.465620873826856e-05, | |
| "loss": 0.9648, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.46113894761344e-05, | |
| "loss": 0.9545, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.456640573575896e-05, | |
| "loss": 1.0256, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.4521257894415183e-05, | |
| "loss": 0.99, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.44759463307523e-05, | |
| "loss": 0.9906, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.443047142479266e-05, | |
| "loss": 0.9876, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.4384833557928553e-05, | |
| "loss": 1.0495, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.4339033112918966e-05, | |
| "loss": 0.9869, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.4293070473886456e-05, | |
| "loss": 1.0299, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.424694602631385e-05, | |
| "loss": 1.0073, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.420066015704105e-05, | |
| "loss": 1.0023, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.41542132542618e-05, | |
| "loss": 0.9462, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.410760570752037e-05, | |
| "loss": 1.0116, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.4060837907708375e-05, | |
| "loss": 0.9652, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.401391024706142e-05, | |
| "loss": 1.0411, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.396682311915586e-05, | |
| "loss": 0.9691, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.3919576918905495e-05, | |
| "loss": 0.961, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.387217204255819e-05, | |
| "loss": 0.9602, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.3824608887692666e-05, | |
| "loss": 1.0197, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.377688785321507e-05, | |
| "loss": 0.9982, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.372900933935569e-05, | |
| "loss": 0.986, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.368097374766556e-05, | |
| "loss": 0.9744, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.3632781481013105e-05, | |
| "loss": 1.0078, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.358443294358077e-05, | |
| "loss": 0.9865, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.35359285408616e-05, | |
| "loss": 0.976, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.348726867965591e-05, | |
| "loss": 1.0141, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.343845376806777e-05, | |
| "loss": 0.952, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.338948421550169e-05, | |
| "loss": 1.0239, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 4.334036043265909e-05, | |
| "loss": 0.994, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.329108283153492e-05, | |
| "loss": 1.0505, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.3241651825414195e-05, | |
| "loss": 0.9919, | |
| "step": 1300 | |
| } | |
| ], | |
| "max_steps": 5424, | |
| "num_train_epochs": 3, | |
| "total_flos": 4.22461405396992e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |