| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1350, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4814814814814817e-06, | |
| "loss": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "loss": 0.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4814814814814815e-05, | |
| "loss": 0.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2222222222222223e-05, | |
| "loss": 0.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.962962962962963e-05, | |
| "loss": 0.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.7037037037037037e-05, | |
| "loss": 0.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.4444444444444447e-05, | |
| "loss": 0.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.185185185185185e-05, | |
| "loss": 0.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.925925925925926e-05, | |
| "loss": 0.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 0.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.407407407407407e-05, | |
| "loss": 0.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.148148148148148e-05, | |
| "loss": 0.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.888888888888889e-05, | |
| "loss": 0.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.62962962962963e-05, | |
| "loss": 0.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001037037037037037, | |
| "loss": 0.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011111111111111112, | |
| "loss": 0.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011851851851851852, | |
| "loss": 0.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00012592592592592592, | |
| "loss": 0.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013333333333333334, | |
| "loss": 0.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00014074074074074076, | |
| "loss": 0.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00014814814814814815, | |
| "loss": 0.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015555555555555556, | |
| "loss": 0.0, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00016296296296296295, | |
| "loss": 0.0, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017037037037037037, | |
| "loss": 0.0, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017777777777777779, | |
| "loss": 0.0, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001851851851851852, | |
| "loss": 0.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001925925925925926, | |
| "loss": 0.0, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0002, | |
| "loss": 0.0, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019999164298554375, | |
| "loss": 0.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019996657333896877, | |
| "loss": 0.0, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019992479525042303, | |
| "loss": 0.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019986631570270832, | |
| "loss": 0.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019979114447011323, | |
| "loss": 0.0, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001996992941167792, | |
| "loss": 0.0, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019959077999460095, | |
| "loss": 0.0, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019946562024066014, | |
| "loss": 0.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019932383577419432, | |
| "loss": 0.0, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019916545029310012, | |
| "loss": 0.0, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019899049026997272, | |
| "loss": 0.0, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019879898494768093, | |
| "loss": 0.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019859096633447965, | |
| "loss": 0.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001983664691986601, | |
| "loss": 0.0, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019812553106273847, | |
| "loss": 0.0, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019786819219718443, | |
| "loss": 0.0, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019759449561369038, | |
| "loss": 0.0, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019730448705798239, | |
| "loss": 0.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019699821500217434, | |
| "loss": 0.0, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001966757306366662, | |
| "loss": 0.0, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019633708786158806, | |
| "loss": 0.0, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019598234327779118, | |
| "loss": 0.0, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019561155617738797, | |
| "loss": 0.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019522478853384155, | |
| "loss": 0.0, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019482210499160765, | |
| "loss": 0.0, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019440357285533, | |
| "loss": 0.0, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019396926207859084, | |
| "loss": 0.0, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.000193519245252219, | |
| "loss": 0.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019305359759215685, | |
| "loss": 0.0, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019257239692688907, | |
| "loss": 0.0, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019207572368443385, | |
| "loss": 0.0, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001915636608789006, | |
| "loss": 0.0, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001910362940966147, | |
| "loss": 0.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00019049371148181253, | |
| "loss": 0.0, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018993600372190932, | |
| "loss": 0.0, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018936326403234125, | |
| "loss": 0.0, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018877558814098561, | |
| "loss": 0.0, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001881730742721608, | |
| "loss": 0.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001875558231302091, | |
| "loss": 0.0, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018692393788266479, | |
| "loss": 0.0, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018627752414301086, | |
| "loss": 0.0, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018561668995302667, | |
| "loss": 0.0, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018494154576472976, | |
| "loss": 0.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018425220442191495, | |
| "loss": 0.0, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018354878114129367, | |
| "loss": 0.0, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018283139349323634, | |
| "loss": 0.0, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018210016138212187, | |
| "loss": 0.0, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018135520702629675, | |
| "loss": 0.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00018059665493764743, | |
| "loss": 0.0, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001798246319007893, | |
| "loss": 0.0, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017903926695187595, | |
| "loss": 0.0, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017824069135703198, | |
| "loss": 0.0, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017742903859041325, | |
| "loss": 0.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001766044443118978, | |
| "loss": 0.0, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001757670463444118, | |
| "loss": 0.0, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017491698465089362, | |
| "loss": 0.0, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017405440131090048, | |
| "loss": 0.0, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017317944049686124, | |
| "loss": 0.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017229224844997928, | |
| "loss": 0.0, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017139297345578994, | |
| "loss": 0.0, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00017048176581937563, | |
| "loss": 0.0, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001695587778402442, | |
| "loss": 0.0, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001686241637868734, | |
| "loss": 0.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00016767807987092621, | |
| "loss": 0.0, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00016672068422114196, | |
| "loss": 0.0, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001657521368569064, | |
| "loss": 0.0, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00016477259966150588, | |
| "loss": 0.0, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001637822363550706, | |
| "loss": 0.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00016278121246720987, | |
| "loss": 0.0, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00016176969530934572, | |
| "loss": 0.0, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00016074785394674837, | |
| "loss": 0.0, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015971585917027862, | |
| "loss": 0.0, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001586738834678418, | |
| "loss": 0.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015762210099555803, | |
| "loss": 0.0, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015656068754865387, | |
| "loss": 0.0, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001554898205320797, | |
| "loss": 0.0, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015440967893085828, | |
| "loss": 0.0, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015332044328016914, | |
| "loss": 0.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015222229563517385, | |
| "loss": 0.0, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015111541954058734, | |
| "loss": 0.0, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 0.0, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00014887622344495643, | |
| "loss": 0.0, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001477442777037949, | |
| "loss": 0.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001466043519702539, | |
| "loss": 0.0, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00014545663677185006, | |
| "loss": 0.0, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00014430132393803352, | |
| "loss": 0.0, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00014313860656812536, | |
| "loss": 0.0, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001419686789990429, | |
| "loss": 0.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00014079173677281837, | |
| "loss": 0.0, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001396079766039157, | |
| "loss": 0.0, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013841759634635178, | |
| "loss": 0.0, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013722079496062702, | |
| "loss": 0.0, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013601777248047105, | |
| "loss": 0.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013480872997940905, | |
| "loss": 0.0, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013359386953715421, | |
| "loss": 0.0, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013237339420583212, | |
| "loss": 0.0, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00013114750797604247, | |
| "loss": 0.0, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00012991641574276418, | |
| "loss": 0.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00012868032327110904, | |
| "loss": 0.0, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00012743943716193016, | |
| "loss": 0.0, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001261939648172906, | |
| "loss": 0.0, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00012494411440579814, | |
| "loss": 0.0, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00012369009482781192, | |
| "loss": 0.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00012243211568052677, | |
| "loss": 0.0, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001211703872229411, | |
| "loss": 0.0, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011990512034071406, | |
| "loss": 0.0, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011863652651091823, | |
| "loss": 0.0, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011736481776669306, | |
| "loss": 0.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011609020666180575, | |
| "loss": 0.0, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0001148129062351249, | |
| "loss": 0.0, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011353312997501313, | |
| "loss": 0.0, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011225109178364455, | |
| "loss": 0.0, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00011096700594125318, | |
| "loss": 0.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010968108707031792, | |
| "loss": 0.0, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010839355009969068, | |
| "loss": 0.0, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010710461022867302, | |
| "loss": 0.0, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010581448289104758, | |
| "loss": 0.0, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010452338371907064, | |
| "loss": 0.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010323152850743107, | |
| "loss": 0.0, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010193913317718244, | |
| "loss": 0.0, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.00010064641373965393, | |
| "loss": 0.0, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.935358626034606e-05, | |
| "loss": 0.0, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.806086682281758e-05, | |
| "loss": 0.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.676847149256895e-05, | |
| "loss": 0.0, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.547661628092937e-05, | |
| "loss": 0.0, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.418551710895243e-05, | |
| "loss": 0.0, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.289538977132703e-05, | |
| "loss": 0.0, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.160644990030931e-05, | |
| "loss": 0.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.03189129296821e-05, | |
| "loss": 0.0, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.903299405874684e-05, | |
| "loss": 0.0, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.774890821635548e-05, | |
| "loss": 0.0, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.646687002498692e-05, | |
| "loss": 0.0, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.518709376487515e-05, | |
| "loss": 0.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.390979333819426e-05, | |
| "loss": 0.0, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.263518223330697e-05, | |
| "loss": 0.0, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.13634734890818e-05, | |
| "loss": 0.0, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.009487965928596e-05, | |
| "loss": 0.0, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.882961277705895e-05, | |
| "loss": 0.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.756788431947326e-05, | |
| "loss": 0.0, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.630990517218808e-05, | |
| "loss": 0.0, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.505588559420189e-05, | |
| "loss": 0.0, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.380603518270941e-05, | |
| "loss": 0.0, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.256056283806986e-05, | |
| "loss": 0.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.131967672889101e-05, | |
| "loss": 0.0, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.008358425723585e-05, | |
| "loss": 0.0, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.885249202395754e-05, | |
| "loss": 0.0, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.762660579416791e-05, | |
| "loss": 0.0, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.640613046284581e-05, | |
| "loss": 0.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.519127002059095e-05, | |
| "loss": 0.0, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.398222751952899e-05, | |
| "loss": 0.0, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.277920503937303e-05, | |
| "loss": 0.0, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.158240365364823e-05, | |
| "loss": 0.0, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.039202339608432e-05, | |
| "loss": 0.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.920826322718165e-05, | |
| "loss": 0.0, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.80313210009571e-05, | |
| "loss": 0.0, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.6861393431874675e-05, | |
| "loss": 0.0, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.5698676061966515e-05, | |
| "loss": 0.0, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.4543363228149946e-05, | |
| "loss": 0.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.339564802974615e-05, | |
| "loss": 0.0, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.22557222962051e-05, | |
| "loss": 0.0, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.112377655504359e-05, | |
| "loss": 0.0, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.000000000000002e-05, | |
| "loss": 0.0, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.888458045941269e-05, | |
| "loss": 0.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.777770436482617e-05, | |
| "loss": 0.0, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.66795567198309e-05, | |
| "loss": 0.0, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.559032106914173e-05, | |
| "loss": 0.0, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.451017946792032e-05, | |
| "loss": 0.0, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.343931245134616e-05, | |
| "loss": 0.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.2377899004441966e-05, | |
| "loss": 0.0, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.132611653215822e-05, | |
| "loss": 0.0, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.028414082972141e-05, | |
| "loss": 0.0, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.9252146053251636e-05, | |
| "loss": 0.0, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.8230304690654304e-05, | |
| "loss": 0.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.721878753279017e-05, | |
| "loss": 0.0, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.621776364492939e-05, | |
| "loss": 0.0, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.522740033849411e-05, | |
| "loss": 0.0, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.424786314309365e-05, | |
| "loss": 0.0, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.3279315778858036e-05, | |
| "loss": 0.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.2321920129073816e-05, | |
| "loss": 0.0, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.137583621312665e-05, | |
| "loss": 0.0, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.04412221597558e-05, | |
| "loss": 0.0, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.9518234180624393e-05, | |
| "loss": 0.0, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.8607026544210114e-05, | |
| "loss": 0.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.770775155002071e-05, | |
| "loss": 0.0, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6820559503138797e-05, | |
| "loss": 0.0, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.594559868909956e-05, | |
| "loss": 0.0, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.50830153491064e-05, | |
| "loss": 0.0, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.423295365558821e-05, | |
| "loss": 0.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.339555568810221e-05, | |
| "loss": 0.0, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.2570961409586754e-05, | |
| "loss": 0.0, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1759308642968025e-05, | |
| "loss": 0.0, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.0960733048124083e-05, | |
| "loss": 0.0, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.01753680992107e-05, | |
| "loss": 0.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.9403345062352573e-05, | |
| "loss": 0.0, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.864479297370325e-05, | |
| "loss": 0.0, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7899838617878163e-05, | |
| "loss": 0.0, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.7168606506763695e-05, | |
| "loss": 0.0, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6451218858706374e-05, | |
| "loss": 0.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.5747795578085046e-05, | |
| "loss": 0.0, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.505845423527027e-05, | |
| "loss": 0.0, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.4383310046973365e-05, | |
| "loss": 0.0, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3722475856989158e-05, | |
| "loss": 0.0, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.307606211733522e-05, | |
| "loss": 0.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.2444176869790925e-05, | |
| "loss": 0.0, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.18269257278392e-05, | |
| "loss": 0.0, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.1224411859014417e-05, | |
| "loss": 0.0, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0636735967658784e-05, | |
| "loss": 0.0, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0063996278090704e-05, | |
| "loss": 0.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.0, | |
| "learning_rate": 9.506288518187467e-06, | |
| "loss": 0.0, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.963705903385345e-06, | |
| "loss": 0.0, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.436339121099412e-06, | |
| "loss": 0.0, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.92427631556617e-06, | |
| "loss": 0.0, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.427603073110967e-06, | |
| "loss": 0.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.946402407843155e-06, | |
| "loss": 0.0, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.480754747781037e-06, | |
| "loss": 0.0, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.030737921409169e-06, | |
| "loss": 0.0, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.596427144670002e-06, | |
| "loss": 0.0, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.177895008392353e-06, | |
| "loss": 0.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.775211466158469e-06, | |
| "loss": 0.0, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.3884438226120424e-06, | |
| "loss": 0.0, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.017656722208807e-06, | |
| "loss": 0.0, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.662912138411967e-06, | |
| "loss": 0.0, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.3242693633337983e-06, | |
| "loss": 0.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.0017849978256516e-06, | |
| "loss": 0.0, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.6955129420176196e-06, | |
| "loss": 0.0, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.4055043863096428e-06, | |
| "loss": 0.0, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.1318078028155888e-06, | |
| "loss": 0.0, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.874468937261531e-06, | |
| "loss": 0.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.6335308013398886e-06, | |
| "loss": 0.0, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.409033665520354e-06, | |
| "loss": 0.0, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.201015052319099e-06, | |
| "loss": 0.0, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.0095097300273026e-06, | |
| "loss": 0.0, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.345497068998897e-07, | |
| "loss": 0.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 0.0, | |
| "learning_rate": 6.761642258056978e-07, | |
| "loss": 0.0, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 0.0, | |
| "learning_rate": 5.343797593398536e-07, | |
| "loss": 0.0, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 0.0, | |
| "learning_rate": 4.092200053990691e-07, | |
| "loss": 0.0, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.007058832207976e-07, | |
| "loss": 0.0, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.0, | |
| "learning_rate": 2.088555298867978e-07, | |
| "loss": 0.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.0, | |
| "learning_rate": 1.3368429729168076e-07, | |
| "loss": 0.0, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.0, | |
| "learning_rate": 7.520474957699586e-08, | |
| "loss": 0.0, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 0.0, | |
| "learning_rate": 3.3426661031255026e-08, | |
| "loss": 0.0, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 8.357014456272794e-09, | |
| "loss": 0.0, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": NaN, | |
| "eval_runtime": 865.6078, | |
| "eval_samples_per_second": 2.669, | |
| "eval_steps_per_second": 0.334, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1350, | |
| "total_flos": 2763707490304000.0, | |
| "train_loss": 0.0, | |
| "train_runtime": 23143.2556, | |
| "train_samples_per_second": 0.933, | |
| "train_steps_per_second": 0.058 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1350, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 2763707490304000.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |