| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.5098673568424457, | |
| "global_step": 3500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.989214840379638e-05, | |
| "loss": 1.653, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9784296807592755e-05, | |
| "loss": 1.0664, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9676445211389134e-05, | |
| "loss": 1.0555, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.956859361518551e-05, | |
| "loss": 1.0407, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.946074201898188e-05, | |
| "loss": 1.0132, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.935289042277826e-05, | |
| "loss": 0.9909, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.924503882657463e-05, | |
| "loss": 0.9312, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.913718723037101e-05, | |
| "loss": 0.9696, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.902933563416739e-05, | |
| "loss": 0.9298, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.8921484037963765e-05, | |
| "loss": 0.9732, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.8813632441760144e-05, | |
| "loss": 0.9815, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.870578084555652e-05, | |
| "loss": 1.0084, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.859792924935289e-05, | |
| "loss": 0.9314, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.849007765314927e-05, | |
| "loss": 0.9135, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.838222605694564e-05, | |
| "loss": 0.9594, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.827437446074202e-05, | |
| "loss": 0.968, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.81665228645384e-05, | |
| "loss": 0.9542, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8058671268334774e-05, | |
| "loss": 0.937, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.795081967213115e-05, | |
| "loss": 0.9218, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.784296807592753e-05, | |
| "loss": 0.928, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.77351164797239e-05, | |
| "loss": 0.9122, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.762726488352028e-05, | |
| "loss": 0.9071, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.751941328731665e-05, | |
| "loss": 0.9467, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.741156169111303e-05, | |
| "loss": 0.9313, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.730371009490941e-05, | |
| "loss": 0.9362, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.719585849870578e-05, | |
| "loss": 0.9306, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.708800690250216e-05, | |
| "loss": 0.9337, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.698015530629854e-05, | |
| "loss": 0.9193, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.687230371009491e-05, | |
| "loss": 0.9286, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.676445211389129e-05, | |
| "loss": 0.917, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.665660051768766e-05, | |
| "loss": 0.8848, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.654874892148404e-05, | |
| "loss": 0.9077, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.644089732528042e-05, | |
| "loss": 0.9341, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.633304572907679e-05, | |
| "loss": 0.9002, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.622519413287317e-05, | |
| "loss": 0.9171, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.6117342536669546e-05, | |
| "loss": 0.9021, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.600949094046592e-05, | |
| "loss": 0.894, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.59016393442623e-05, | |
| "loss": 0.9241, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.579378774805868e-05, | |
| "loss": 0.8885, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.568593615185505e-05, | |
| "loss": 0.8635, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.5578084555651424e-05, | |
| "loss": 0.824, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.5470232959447804e-05, | |
| "loss": 0.8661, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.5362381363244177e-05, | |
| "loss": 0.8665, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.5254529767040556e-05, | |
| "loss": 0.8888, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.514667817083693e-05, | |
| "loss": 0.8807, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.503882657463331e-05, | |
| "loss": 0.8796, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.493097497842969e-05, | |
| "loss": 0.8702, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.4823123382226054e-05, | |
| "loss": 0.8801, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.4715271786022434e-05, | |
| "loss": 0.8825, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.4607420189818814e-05, | |
| "loss": 0.8886, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.4499568593615186e-05, | |
| "loss": 0.8448, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.4391716997411566e-05, | |
| "loss": 0.9007, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.428386540120794e-05, | |
| "loss": 0.7982, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.417601380500432e-05, | |
| "loss": 0.9384, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.40681622088007e-05, | |
| "loss": 0.8831, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.3960310612597064e-05, | |
| "loss": 0.8702, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.3852459016393444e-05, | |
| "loss": 0.8665, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.374460742018982e-05, | |
| "loss": 0.872, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.3636755823986196e-05, | |
| "loss": 0.8459, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.3528904227782576e-05, | |
| "loss": 0.8411, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.342105263157895e-05, | |
| "loss": 0.8853, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.331320103537533e-05, | |
| "loss": 0.8657, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.32053494391717e-05, | |
| "loss": 0.8364, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.3097497842968074e-05, | |
| "loss": 0.8717, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.2989646246764453e-05, | |
| "loss": 0.8352, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.288179465056083e-05, | |
| "loss": 0.8549, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.2773943054357206e-05, | |
| "loss": 0.8709, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.2666091458153586e-05, | |
| "loss": 0.8434, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.255823986194996e-05, | |
| "loss": 0.8736, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.245038826574633e-05, | |
| "loss": 0.8683, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.234253666954271e-05, | |
| "loss": 0.8525, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.2234685073339084e-05, | |
| "loss": 0.8482, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.212683347713546e-05, | |
| "loss": 0.8434, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.201898188093184e-05, | |
| "loss": 0.8192, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.1911130284728216e-05, | |
| "loss": 0.847, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.1803278688524595e-05, | |
| "loss": 0.8417, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.169542709232097e-05, | |
| "loss": 0.831, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.158757549611734e-05, | |
| "loss": 0.8364, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.147972389991372e-05, | |
| "loss": 0.8438, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.1371872303710093e-05, | |
| "loss": 0.8497, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.126402070750647e-05, | |
| "loss": 0.8448, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.115616911130285e-05, | |
| "loss": 0.831, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.1048317515099225e-05, | |
| "loss": 0.8303, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.0940465918895605e-05, | |
| "loss": 0.8329, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.083261432269198e-05, | |
| "loss": 0.8052, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.072476272648835e-05, | |
| "loss": 0.8283, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.061691113028473e-05, | |
| "loss": 0.8428, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.05090595340811e-05, | |
| "loss": 0.8242, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.040120793787748e-05, | |
| "loss": 0.8271, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.029335634167386e-05, | |
| "loss": 0.839, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.0185504745470235e-05, | |
| "loss": 0.8184, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.007765314926661e-05, | |
| "loss": 0.8053, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.996980155306299e-05, | |
| "loss": 0.7676, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.986194995685936e-05, | |
| "loss": 0.8404, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.975409836065574e-05, | |
| "loss": 0.8084, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.964624676445211e-05, | |
| "loss": 0.8285, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.953839516824849e-05, | |
| "loss": 0.7973, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.943054357204487e-05, | |
| "loss": 0.8617, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.9322691975841245e-05, | |
| "loss": 0.79, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.921484037963762e-05, | |
| "loss": 0.8048, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.9106988783434e-05, | |
| "loss": 0.8726, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.899913718723037e-05, | |
| "loss": 0.8206, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.889128559102675e-05, | |
| "loss": 0.8039, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.878343399482313e-05, | |
| "loss": 0.8041, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.86755823986195e-05, | |
| "loss": 0.7987, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.856773080241588e-05, | |
| "loss": 0.8119, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.8459879206212255e-05, | |
| "loss": 0.8252, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.835202761000863e-05, | |
| "loss": 0.8043, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.824417601380501e-05, | |
| "loss": 0.7446, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.813632441760138e-05, | |
| "loss": 0.8219, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.802847282139776e-05, | |
| "loss": 0.8219, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.792062122519414e-05, | |
| "loss": 0.7932, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.781276962899051e-05, | |
| "loss": 0.7698, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.7704918032786885e-05, | |
| "loss": 0.7855, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.7597066436583265e-05, | |
| "loss": 0.7862, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.748921484037964e-05, | |
| "loss": 0.7791, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.738136324417602e-05, | |
| "loss": 0.7778, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 3.727351164797239e-05, | |
| "loss": 0.7831, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 3.716566005176877e-05, | |
| "loss": 0.8099, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.705780845556515e-05, | |
| "loss": 0.7796, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.694995685936152e-05, | |
| "loss": 0.7923, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.6842105263157895e-05, | |
| "loss": 0.8082, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.6734253666954274e-05, | |
| "loss": 0.8252, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.662640207075065e-05, | |
| "loss": 0.7972, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 3.651855047454703e-05, | |
| "loss": 0.771, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 3.64106988783434e-05, | |
| "loss": 0.8091, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.630284728213978e-05, | |
| "loss": 0.7834, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.619499568593616e-05, | |
| "loss": 0.7765, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.6087144089732525e-05, | |
| "loss": 0.7705, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 3.5979292493528904e-05, | |
| "loss": 0.7727, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.5871440897325284e-05, | |
| "loss": 0.7936, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.576358930112166e-05, | |
| "loss": 0.8337, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.5655737704918037e-05, | |
| "loss": 0.7418, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.554788610871441e-05, | |
| "loss": 0.7572, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.544003451251079e-05, | |
| "loss": 0.7778, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.533218291630716e-05, | |
| "loss": 0.7507, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.5224331320103535e-05, | |
| "loss": 0.7422, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.5116479723899914e-05, | |
| "loss": 0.7935, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.5008628127696294e-05, | |
| "loss": 0.7565, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.490077653149267e-05, | |
| "loss": 0.7539, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.4792924935289046e-05, | |
| "loss": 0.7774, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.468507333908542e-05, | |
| "loss": 0.7558, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.45772217428818e-05, | |
| "loss": 0.7786, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.446937014667817e-05, | |
| "loss": 0.7977, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.4361518550474544e-05, | |
| "loss": 0.7465, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.4253666954270924e-05, | |
| "loss": 0.7633, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.4145815358067304e-05, | |
| "loss": 0.7429, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.4037963761863676e-05, | |
| "loss": 0.7493, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.3930112165660056e-05, | |
| "loss": 0.7553, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.382226056945643e-05, | |
| "loss": 0.7607, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.37144089732528e-05, | |
| "loss": 0.7704, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.360655737704918e-05, | |
| "loss": 0.7861, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.3498705780845554e-05, | |
| "loss": 0.7597, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.3390854184641934e-05, | |
| "loss": 0.7209, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.3283002588438313e-05, | |
| "loss": 0.7562, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.3175150992234686e-05, | |
| "loss": 0.7533, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.3067299396031066e-05, | |
| "loss": 0.7745, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.295944779982744e-05, | |
| "loss": 0.7717, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.285159620362381e-05, | |
| "loss": 0.7423, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.274374460742019e-05, | |
| "loss": 0.7462, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2635893011216564e-05, | |
| "loss": 0.7451, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.2528041415012944e-05, | |
| "loss": 0.7586, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.242018981880932e-05, | |
| "loss": 0.7608, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.2312338222605696e-05, | |
| "loss": 0.7752, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.2204486626402076e-05, | |
| "loss": 0.761, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.209663503019845e-05, | |
| "loss": 0.7504, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.198878343399482e-05, | |
| "loss": 0.7559, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.18809318377912e-05, | |
| "loss": 0.7796, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.177308024158758e-05, | |
| "loss": 0.7372, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.166522864538395e-05, | |
| "loss": 0.7566, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.155737704918033e-05, | |
| "loss": 0.7937, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.1449525452976706e-05, | |
| "loss": 0.751, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.134167385677308e-05, | |
| "loss": 0.7787, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.123382226056946e-05, | |
| "loss": 0.751, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.112597066436583e-05, | |
| "loss": 0.7211, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.101811906816221e-05, | |
| "loss": 0.7832, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.091026747195859e-05, | |
| "loss": 0.7904, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.080241587575496e-05, | |
| "loss": 0.7576, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.069456427955134e-05, | |
| "loss": 0.7603, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.0586712683347716e-05, | |
| "loss": 0.7267, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.047886108714409e-05, | |
| "loss": 0.6739, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.0371009490940465e-05, | |
| "loss": 0.7507, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.0263157894736844e-05, | |
| "loss": 0.742, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.015530629853322e-05, | |
| "loss": 0.7272, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.0047454702329597e-05, | |
| "loss": 0.7401, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.9939603106125973e-05, | |
| "loss": 0.7451, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.9831751509922353e-05, | |
| "loss": 0.7417, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.9723899913718722e-05, | |
| "loss": 0.7484, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.9616048317515098e-05, | |
| "loss": 0.754, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.9508196721311478e-05, | |
| "loss": 0.7558, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.9400345125107854e-05, | |
| "loss": 0.7239, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.929249352890423e-05, | |
| "loss": 0.6976, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.9184641932700606e-05, | |
| "loss": 0.6897, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.9076790336496983e-05, | |
| "loss": 0.7171, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 2.8968938740293356e-05, | |
| "loss": 0.7228, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.8861087144089732e-05, | |
| "loss": 0.7759, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.8753235547886108e-05, | |
| "loss": 0.7583, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.8645383951682488e-05, | |
| "loss": 0.7303, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.8537532355478864e-05, | |
| "loss": 0.6929, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.842968075927524e-05, | |
| "loss": 0.7226, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.8321829163071616e-05, | |
| "loss": 0.744, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.821397756686799e-05, | |
| "loss": 0.6937, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.8106125970664365e-05, | |
| "loss": 0.7244, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.799827437446074e-05, | |
| "loss": 0.7052, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.7890422778257118e-05, | |
| "loss": 0.7069, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.7782571182053497e-05, | |
| "loss": 0.7292, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.7674719585849874e-05, | |
| "loss": 0.7065, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.756686798964625e-05, | |
| "loss": 0.7137, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7459016393442626e-05, | |
| "loss": 0.7395, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.7351164797239e-05, | |
| "loss": 0.6953, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.7243313201035375e-05, | |
| "loss": 0.7287, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.713546160483175e-05, | |
| "loss": 0.7369, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.7027610008628128e-05, | |
| "loss": 0.747, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.6919758412424507e-05, | |
| "loss": 0.7342, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.6811906816220883e-05, | |
| "loss": 0.7194, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.670405522001726e-05, | |
| "loss": 0.7184, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.6596203623813632e-05, | |
| "loss": 0.7232, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.648835202761001e-05, | |
| "loss": 0.7048, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 2.6380500431406385e-05, | |
| "loss": 0.7227, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.627264883520276e-05, | |
| "loss": 0.7026, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.616479723899914e-05, | |
| "loss": 0.7497, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.6056945642795517e-05, | |
| "loss": 0.6849, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.5949094046591893e-05, | |
| "loss": 0.7417, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.5841242450388266e-05, | |
| "loss": 0.7114, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.5733390854184642e-05, | |
| "loss": 0.7019, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.562553925798102e-05, | |
| "loss": 0.735, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.5517687661777395e-05, | |
| "loss": 0.6782, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.540983606557377e-05, | |
| "loss": 0.7492, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.530198446937015e-05, | |
| "loss": 0.72, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.5194132873166527e-05, | |
| "loss": 0.6884, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.5086281276962903e-05, | |
| "loss": 0.7112, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.4978429680759276e-05, | |
| "loss": 0.6953, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.4870578084555652e-05, | |
| "loss": 0.5278, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.4762726488352028e-05, | |
| "loss": 0.5105, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 2.4654874892148404e-05, | |
| "loss": 0.5109, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.454702329594478e-05, | |
| "loss": 0.5153, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.4439171699741157e-05, | |
| "loss": 0.5137, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 2.4331320103537533e-05, | |
| "loss": 0.5157, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 2.422346850733391e-05, | |
| "loss": 0.4999, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.4115616911130286e-05, | |
| "loss": 0.5095, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.4007765314926662e-05, | |
| "loss": 0.5243, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 2.3899913718723038e-05, | |
| "loss": 0.5056, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 2.3792062122519414e-05, | |
| "loss": 0.5118, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 2.368421052631579e-05, | |
| "loss": 0.4913, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2.3576358930112167e-05, | |
| "loss": 0.5117, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 2.3468507333908543e-05, | |
| "loss": 0.5135, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 2.336065573770492e-05, | |
| "loss": 0.5068, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 2.3252804141501295e-05, | |
| "loss": 0.532, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 2.314495254529767e-05, | |
| "loss": 0.5272, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.3037100949094048e-05, | |
| "loss": 0.5164, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.2929249352890424e-05, | |
| "loss": 0.5235, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.28213977566868e-05, | |
| "loss": 0.4908, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 2.2713546160483176e-05, | |
| "loss": 0.5175, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 2.2605694564279553e-05, | |
| "loss": 0.5044, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 2.249784296807593e-05, | |
| "loss": 0.4979, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 2.2389991371872305e-05, | |
| "loss": 0.5075, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 2.228213977566868e-05, | |
| "loss": 0.5201, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 2.2174288179465058e-05, | |
| "loss": 0.51, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 2.2066436583261434e-05, | |
| "loss": 0.5234, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 2.195858498705781e-05, | |
| "loss": 0.5003, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.1850733390854186e-05, | |
| "loss": 0.5112, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.1742881794650562e-05, | |
| "loss": 0.5093, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.163503019844694e-05, | |
| "loss": 0.5165, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 2.1527178602243315e-05, | |
| "loss": 0.535, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 2.141932700603969e-05, | |
| "loss": 0.4887, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 2.1311475409836064e-05, | |
| "loss": 0.4998, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 2.1203623813632444e-05, | |
| "loss": 0.4902, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 2.109577221742882e-05, | |
| "loss": 0.5138, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 2.0987920621225196e-05, | |
| "loss": 0.5424, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 2.088006902502157e-05, | |
| "loss": 0.5082, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 2.077221742881795e-05, | |
| "loss": 0.5198, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 2.0664365832614325e-05, | |
| "loss": 0.503, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 2.05565142364107e-05, | |
| "loss": 0.5263, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 2.0448662640207074e-05, | |
| "loss": 0.4992, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 2.0340811044003453e-05, | |
| "loss": 0.5104, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 2.023295944779983e-05, | |
| "loss": 0.4862, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 2.0125107851596202e-05, | |
| "loss": 0.5245, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.001725625539258e-05, | |
| "loss": 0.5111, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.9909404659188958e-05, | |
| "loss": 0.4914, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.9801553062985334e-05, | |
| "loss": 0.5003, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.9693701466781707e-05, | |
| "loss": 0.5255, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.9585849870578087e-05, | |
| "loss": 0.4957, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.9477998274374463e-05, | |
| "loss": 0.5002, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.937014667817084e-05, | |
| "loss": 0.5131, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.9262295081967212e-05, | |
| "loss": 0.4771, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.9154443485763592e-05, | |
| "loss": 0.4961, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.9046591889559968e-05, | |
| "loss": 0.5043, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.893874029335634e-05, | |
| "loss": 0.5105, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.8830888697152717e-05, | |
| "loss": 0.513, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.8723037100949097e-05, | |
| "loss": 0.5248, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.8615185504745473e-05, | |
| "loss": 0.5239, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.8507333908541846e-05, | |
| "loss": 0.4915, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.8399482312338222e-05, | |
| "loss": 0.5036, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.82916307161346e-05, | |
| "loss": 0.5197, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.8183779119930978e-05, | |
| "loss": 0.4982, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.807592752372735e-05, | |
| "loss": 0.4958, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.7968075927523727e-05, | |
| "loss": 0.4914, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.7860224331320106e-05, | |
| "loss": 0.4931, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.775237273511648e-05, | |
| "loss": 0.4971, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.7644521138912855e-05, | |
| "loss": 0.4797, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.7536669542709232e-05, | |
| "loss": 0.478, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.742881794650561e-05, | |
| "loss": 0.5026, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.7320966350301984e-05, | |
| "loss": 0.4962, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.721311475409836e-05, | |
| "loss": 0.4947, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.7105263157894737e-05, | |
| "loss": 0.4971, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.6997411561691116e-05, | |
| "loss": 0.5163, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.688955996548749e-05, | |
| "loss": 0.51, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.6781708369283865e-05, | |
| "loss": 0.4748, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.667385677308024e-05, | |
| "loss": 0.5278, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.6566005176876618e-05, | |
| "loss": 0.5114, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.6458153580672994e-05, | |
| "loss": 0.5026, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.635030198446937e-05, | |
| "loss": 0.506, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.6242450388265746e-05, | |
| "loss": 0.5117, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.6134598792062123e-05, | |
| "loss": 0.5018, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.60267471958585e-05, | |
| "loss": 0.5128, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.5918895599654875e-05, | |
| "loss": 0.5117, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.5811044003451255e-05, | |
| "loss": 0.4867, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.5703192407247627e-05, | |
| "loss": 0.5135, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5595340811044004e-05, | |
| "loss": 0.4854, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.548748921484038e-05, | |
| "loss": 0.5213, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5379637618636756e-05, | |
| "loss": 0.5172, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.5271786022433132e-05, | |
| "loss": 0.5093, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.5163934426229509e-05, | |
| "loss": 0.5317, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5056082830025887e-05, | |
| "loss": 0.4774, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.4948231233822261e-05, | |
| "loss": 0.4928, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.4840379637618637e-05, | |
| "loss": 0.4947, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.4732528041415013e-05, | |
| "loss": 0.5077, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.4624676445211391e-05, | |
| "loss": 0.5069, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4516824849007766e-05, | |
| "loss": 0.5152, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4408973252804142e-05, | |
| "loss": 0.5001, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4301121656600518e-05, | |
| "loss": 0.487, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4193270060396893e-05, | |
| "loss": 0.4716, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.408541846419327e-05, | |
| "loss": 0.5083, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.3977566867989647e-05, | |
| "loss": 0.4868, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.3869715271786023e-05, | |
| "loss": 0.478, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.3761863675582398e-05, | |
| "loss": 0.497, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.3654012079378776e-05, | |
| "loss": 0.4859, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.3546160483175152e-05, | |
| "loss": 0.4857, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.343830888697153e-05, | |
| "loss": 0.4758, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.3330457290767903e-05, | |
| "loss": 0.5011, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.322260569456428e-05, | |
| "loss": 0.5234, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.3114754098360657e-05, | |
| "loss": 0.4959, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.3006902502157031e-05, | |
| "loss": 0.4935, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.2899050905953408e-05, | |
| "loss": 0.4949, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.2791199309749785e-05, | |
| "loss": 0.4963, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.2683347713546162e-05, | |
| "loss": 0.501, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.2575496117342536e-05, | |
| "loss": 0.4954, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.2467644521138912e-05, | |
| "loss": 0.504, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.235979292493529e-05, | |
| "loss": 0.4889, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.2251941328731665e-05, | |
| "loss": 0.4794, | |
| "step": 3500 | |
| } | |
| ], | |
| "max_steps": 4636, | |
| "num_train_epochs": 2, | |
| "total_flos": 6.34654862180352e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |