| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.7025862068965517, | |
| "eval_steps": 500, | |
| "global_step": 1141, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9999981288975627e-05, | |
| "loss": 4.3158, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.9999532227891695e-05, | |
| "loss": 3.8969, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9998128955328933e-05, | |
| "loss": 3.4446, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9995790313594056e-05, | |
| "loss": 3.2177, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.999251652147735e-05, | |
| "loss": 3.0663, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.998830788525654e-05, | |
| "loss": 2.9812, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9983164798668152e-05, | |
| "loss": 2.9121, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9977087742870683e-05, | |
| "loss": 2.8353, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.997007728639956e-05, | |
| "loss": 2.7443, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9962134085114e-05, | |
| "loss": 2.7275, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9953258882135594e-05, | |
| "loss": 2.6533, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9943452507778832e-05, | |
| "loss": 2.6084, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9932715879473385e-05, | |
| "loss": 2.5925, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9921050001678316e-05, | |
| "loss": 2.5301, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.990845596578807e-05, | |
| "loss": 2.5413, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.989493495003039e-05, | |
| "loss": 2.5204, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9880488219356086e-05, | |
| "loss": 2.5032, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9865117125320693e-05, | |
| "loss": 2.457, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9848823105958023e-05, | |
| "loss": 2.4431, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9831607685645636e-05, | |
| "loss": 2.4166, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.981347247496222e-05, | |
| "loss": 2.4134, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9794419170536918e-05, | |
| "loss": 2.402, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9774449554890615e-05, | |
| "loss": 2.391, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.975356549626915e-05, | |
| "loss": 2.35, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.973176894846855e-05, | |
| "loss": 2.3311, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9709061950652237e-05, | |
| "loss": 2.3511, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.968544662716026e-05, | |
| "loss": 2.3209, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.966092518731055e-05, | |
| "loss": 2.3085, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.963549992519223e-05, | |
| "loss": 2.3357, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9609173219450998e-05, | |
| "loss": 2.3342, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9581947533066577e-05, | |
| "loss": 2.2873, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9553825413122313e-05, | |
| "loss": 2.2924, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9524809490566878e-05, | |
| "loss": 2.2737, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.949490247996812e-05, | |
| "loss": 2.2558, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9464107179259126e-05, | |
| "loss": 2.2746, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9432426469476432e-05, | |
| "loss": 2.2444, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.939986331449053e-05, | |
| "loss": 2.2759, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9366420760728546e-05, | |
| "loss": 2.2103, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9332101936889262e-05, | |
| "loss": 2.2509, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9296910053650388e-05, | |
| "loss": 2.2346, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.926084840336821e-05, | |
| "loss": 2.244, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.922392035976957e-05, | |
| "loss": 2.2383, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.918612937763622e-05, | |
| "loss": 2.2148, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9147478992481642e-05, | |
| "loss": 2.2032, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.910797282022027e-05, | |
| "loss": 2.2073, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9067614556829195e-05, | |
| "loss": 2.2182, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.9026407978002415e-05, | |
| "loss": 2.202, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.8984356938797573e-05, | |
| "loss": 2.1799, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.894146537327533e-05, | |
| "loss": 2.1912, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8897737294131288e-05, | |
| "loss": 2.1655, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.8853176792320598e-05, | |
| "loss": 2.1951, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8807788036675243e-05, | |
| "loss": 2.1734, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8761575273514005e-05, | |
| "loss": 2.1795, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8714542826245214e-05, | |
| "loss": 2.1636, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.866669509496227e-05, | |
| "loss": 2.1574, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.861803655603201e-05, | |
| "loss": 2.1472, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8568571761675893e-05, | |
| "loss": 2.1517, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8518305339544157e-05, | |
| "loss": 2.1589, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8467241992282842e-05, | |
| "loss": 2.1256, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.841538649709388e-05, | |
| "loss": 2.143, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.8362743705288127e-05, | |
| "loss": 2.1449, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.830931854183153e-05, | |
| "loss": 2.1219, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.825511600488435e-05, | |
| "loss": 2.1384, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8200141165333592e-05, | |
| "loss": 2.1217, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.814439916631857e-05, | |
| "loss": 2.1361, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8087895222749773e-05, | |
| "loss": 2.1502, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.803063462082095e-05, | |
| "loss": 2.1199, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7972622717514608e-05, | |
| "loss": 2.1221, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.791386494010081e-05, | |
| "loss": 2.1148, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.7854366785629435e-05, | |
| "loss": 2.1336, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7794133820415916e-05, | |
| "loss": 2.1189, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7733171679520485e-05, | |
| "loss": 2.1288, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7671486066220965e-05, | |
| "loss": 2.1152, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.7609082751479246e-05, | |
| "loss": 2.0836, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7545967573401336e-05, | |
| "loss": 2.0997, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7482146436691227e-05, | |
| "loss": 2.0885, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.7417625312098453e-05, | |
| "loss": 2.1141, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.7352410235859503e-05, | |
| "loss": 2.097, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.728650730913312e-05, | |
| "loss": 2.0704, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.721992269742951e-05, | |
| "loss": 2.0772, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7152662630033506e-05, | |
| "loss": 2.0714, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7084733399421808e-05, | |
| "loss": 2.0872, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.7016141360674307e-05, | |
| "loss": 2.1176, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6946892930879516e-05, | |
| "loss": 2.0739, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6876994588534234e-05, | |
| "loss": 2.0703, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.680645287293746e-05, | |
| "loss": 2.1082, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.6735274383578593e-05, | |
| "loss": 2.1068, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.6663465779520042e-05, | |
| "loss": 2.0818, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.659103377877423e-05, | |
| "loss": 2.0882, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.6517985157675097e-05, | |
| "loss": 2.0793, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.6444326750244145e-05, | |
| "loss": 2.0798, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.637006544755108e-05, | |
| "loss": 2.0706, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.629520819706912e-05, | |
| "loss": 2.0798, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6219762002025055e-05, | |
| "loss": 2.038, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.614373392074402e-05, | |
| "loss": 2.049, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.6067131065989188e-05, | |
| "loss": 2.0522, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.598996060429634e-05, | |
| "loss": 2.054, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.591222975530338e-05, | |
| "loss": 2.071, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5833945791074943e-05, | |
| "loss": 2.0592, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.5755116035422015e-05, | |
| "loss": 2.0569, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.56757478632168e-05, | |
| "loss": 2.0351, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5595848699702752e-05, | |
| "loss": 2.0399, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.5515426019799898e-05, | |
| "loss": 2.0554, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5434487347405553e-05, | |
| "loss": 2.0537, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5353040254690396e-05, | |
| "loss": 2.0512, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.5271092361390078e-05, | |
| "loss": 2.0567, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5188651334092373e-05, | |
| "loss": 2.0431, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5105724885519908e-05, | |
| "loss": 2.0529, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.5022320773808612e-05, | |
| "loss": 2.0356, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4938446801781932e-05, | |
| "loss": 2.0464, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4854110816220805e-05, | |
| "loss": 2.0784, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4769320707129588e-05, | |
| "loss": 2.0447, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.4684084406997903e-05, | |
| "loss": 2.0425, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.459840989005851e-05, | |
| "loss": 2.0235, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4512305171541287e-05, | |
| "loss": 2.0436, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.4425778306923388e-05, | |
| "loss": 2.0223, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4338837391175582e-05, | |
| "loss": 2.01, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4251490558004965e-05, | |
| "loss": 2.0334, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.4163745979093997e-05, | |
| "loss": 2.035, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.4075611863336012e-05, | |
| "loss": 2.0281, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.3987096456067236e-05, | |
| "loss": 2.0242, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.3898208038295407e-05, | |
| "loss": 2.0241, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3808954925925042e-05, | |
| "loss": 1.9977, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3719345468979448e-05, | |
| "loss": 2.0308, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.3629388050819547e-05, | |
| "loss": 2.0139, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.353909108735957e-05, | |
| "loss": 2.0126, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3448463026279706e-05, | |
| "loss": 2.0249, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.33575123462358e-05, | |
| "loss": 2.0044, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.3266247556066122e-05, | |
| "loss": 2.023, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.317467719399534e-05, | |
| "loss": 2.0183, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.3082809826835707e-05, | |
| "loss": 1.9861, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.2990654049185636e-05, | |
| "loss": 1.9841, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2898218482625606e-05, | |
| "loss": 2.0091, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2805511774911585e-05, | |
| "loss": 1.9951, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.2712542599165998e-05, | |
| "loss": 1.9862, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.261931965306632e-05, | |
| "loss": 2.0169, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.252585165803135e-05, | |
| "loss": 2.0239, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.2432147358405313e-05, | |
| "loss": 2.0428, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.233821552063977e-05, | |
| "loss": 1.98, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2244064932473467e-05, | |
| "loss": 1.9918, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2149704402110243e-05, | |
| "loss": 1.9996, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.2055142757394936e-05, | |
| "loss": 1.9917, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1960388844987529e-05, | |
| "loss": 2.0338, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1865451529535496e-05, | |
| "loss": 2.003, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.1770339692844484e-05, | |
| "loss": 1.978, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1675062233047365e-05, | |
| "loss": 1.992, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.157962806377179e-05, | |
| "loss": 1.998, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.1484046113306262e-05, | |
| "loss": 2.003, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1388325323764889e-05, | |
| "loss": 1.9771, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1292474650250776e-05, | |
| "loss": 1.9918, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1196503060018243e-05, | |
| "loss": 1.9963, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.1100419531633917e-05, | |
| "loss": 1.9926, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.1004233054136726e-05, | |
| "loss": 2.0016, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.090795262619694e-05, | |
| "loss": 1.9745, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.0811587255274313e-05, | |
| "loss": 1.9952, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0715145956775395e-05, | |
| "loss": 2.009, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0618637753210086e-05, | |
| "loss": 2.0072, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.0522071673347558e-05, | |
| "loss": 1.9818, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0425456751371564e-05, | |
| "loss": 1.9788, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0328802026035254e-05, | |
| "loss": 1.994, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.0232116539815558e-05, | |
| "loss": 1.9926, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.013540933806722e-05, | |
| "loss": 1.9933, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.0038689468176575e-05, | |
| "loss": 1.9848, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.941965978715105e-06, | |
| "loss": 2.0002, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.845247918592937e-06, | |
| "loss": 1.9681, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.748544336212246e-06, | |
| "loss": 1.9588, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.651864278620752e-06, | |
| "loss": 1.9817, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.555216790665334e-06, | |
| "loss": 1.9991, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.458610914145826e-06, | |
| "loss": 1.9723, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.362055686969141e-06, | |
| "loss": 1.9725, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 9.26556014230371e-06, | |
| "loss": 1.9663, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.169133307734415e-06, | |
| "loss": 1.9989, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.072784204417995e-06, | |
| "loss": 1.9618, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 8.976521846239076e-06, | |
| "loss": 1.988, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.880355238966923e-06, | |
| "loss": 1.9748, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.784293379412851e-06, | |
| "loss": 1.9679, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.688345254588579e-06, | |
| "loss": 1.977, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 8.592519840865432e-06, | |
| "loss": 1.9859, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.496826103134572e-06, | |
| "loss": 1.9748, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.401272993968288e-06, | |
| "loss": 1.9848, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 8.305869452782446e-06, | |
| "loss": 1.9853, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.210624405000169e-06, | |
| "loss": 1.9828, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.115546761216822e-06, | |
| "loss": 1.9695, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 8.020645416366367e-06, | |
| "loss": 1.9649, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.92592924888925e-06, | |
| "loss": 1.9531, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.831407119901737e-06, | |
| "loss": 1.9787, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 7.73708787236694e-06, | |
| "loss": 1.978, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.642980330267516e-06, | |
| "loss": 1.9794, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.549093297780133e-06, | |
| "loss": 1.9848, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.455435558451824e-06, | |
| "loss": 1.9669, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.362015874378219e-06, | |
| "loss": 1.9532, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.268842985383841e-06, | |
| "loss": 1.9685, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.175925608204428e-06, | |
| "loss": 1.974, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.0832724356714715e-06, | |
| "loss": 1.9563, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.990892135898937e-06, | |
| "loss": 1.9648, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.898793351472341e-06, | |
| "loss": 1.9784, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.806984698640202e-06, | |
| "loss": 1.9783, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.715474766507932e-06, | |
| "loss": 1.9725, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.624272116234319e-06, | |
| "loss": 1.9531, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.5333852802305556e-06, | |
| "loss": 1.9508, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.442822761362015e-06, | |
| "loss": 1.9733, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.352593032152787e-06, | |
| "loss": 1.9748, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.262704533992995e-06, | |
| "loss": 1.9624, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 6.173165676349103e-06, | |
| "loss": 1.9534, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.083984835977154e-06, | |
| "loss": 1.9787, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.995170356139088e-06, | |
| "loss": 1.9767, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 5.9067305458222054e-06, | |
| "loss": 1.9736, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.8186736789618105e-06, | |
| "loss": 1.9583, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.731007993667155e-06, | |
| "loss": 1.9596, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 5.643741691450733e-06, | |
| "loss": 1.9644, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.556882936460966e-06, | |
| "loss": 1.9409, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.4704398547184535e-06, | |
| "loss": 1.9409, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 5.38442053335571e-06, | |
| "loss": 1.9568, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.298833019860599e-06, | |
| "loss": 1.9416, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.213685321323451e-06, | |
| "loss": 1.9614, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 5.128985403687949e-06, | |
| "loss": 1.9659, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 5.044741191005908e-06, | |
| "loss": 1.9459, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.960960564695922e-06, | |
| "loss": 1.9613, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.877651362806038e-06, | |
| "loss": 1.9665, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 4.7948213792804475e-06, | |
| "loss": 1.9656, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.712478363230361e-06, | |
| "loss": 1.9454, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.630630018209027e-06, | |
| "loss": 1.9412, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.5492840014910145e-06, | |
| "loss": 1.9516, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.468447923355888e-06, | |
| "loss": 1.9871, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.388129346376177e-06, | |
| "loss": 1.9627, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.308335784709902e-06, | |
| "loss": 1.9765, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.22907470339758e-06, | |
| "loss": 1.9573, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.1503535176638264e-06, | |
| "loss": 1.9348, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.07217959222365e-06, | |
| "loss": 1.9642, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 1.9577815532684326, | |
| "eval_runtime": 81.9871, | |
| "eval_samples_per_second": 281.874, | |
| "eval_steps_per_second": 8.818, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "step": 1141, | |
| "total_flos": 17914719633408.0, | |
| "train_loss": 2.1211067035885676, | |
| "train_runtime": 2085.5853, | |
| "train_samples_per_second": 99.667, | |
| "train_steps_per_second": 0.779 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1624, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "total_flos": 17914719633408.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |