| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 20.0, | |
| "eval_steps": 3000, | |
| "global_step": 5400, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.00019962962962962963, | |
| "loss": 2.1858, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00019925925925925927, | |
| "loss": 2.0674, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0001988888888888889, | |
| "loss": 1.8298, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00019851851851851853, | |
| "loss": 1.6136, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.00019814814814814814, | |
| "loss": 1.5421, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00019777777777777778, | |
| "loss": 1.3979, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.00019740740740740743, | |
| "loss": 1.3041, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.00019703703703703704, | |
| "loss": 1.2235, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00019666666666666666, | |
| "loss": 1.3809, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0001962962962962963, | |
| "loss": 1.2073, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00019592592592592594, | |
| "loss": 1.114, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00019555555555555556, | |
| "loss": 1.1675, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.0001951851851851852, | |
| "loss": 1.1497, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00019481481481481482, | |
| "loss": 1.2117, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00019444444444444446, | |
| "loss": 0.9834, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00019407407407407408, | |
| "loss": 1.014, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00019370370370370372, | |
| "loss": 0.9551, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00019333333333333333, | |
| "loss": 0.99, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00019296296296296298, | |
| "loss": 0.9756, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.0001925925925925926, | |
| "loss": 1.0463, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00019222222222222224, | |
| "loss": 1.0912, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00019185185185185185, | |
| "loss": 0.9984, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0001914814814814815, | |
| "loss": 0.8911, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00019111111111111114, | |
| "loss": 0.7515, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00019074074074074075, | |
| "loss": 1.1003, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00019037037037037037, | |
| "loss": 0.9923, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00019, | |
| "loss": 0.8772, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00018962962962962965, | |
| "loss": 0.6926, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.00018925925925925927, | |
| "loss": 0.7489, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 0.6285, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00018851851851851853, | |
| "loss": 0.6881, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0001881851851851852, | |
| "loss": 0.6729, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001878148148148148, | |
| "loss": 0.642, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00018744444444444445, | |
| "loss": 0.6423, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.0001870740740740741, | |
| "loss": 0.72, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.0001867037037037037, | |
| "loss": 0.7628, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.00018633333333333333, | |
| "loss": 0.7165, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00018596296296296297, | |
| "loss": 0.6817, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0001855925925925926, | |
| "loss": 0.6094, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00018522222222222223, | |
| "loss": 0.662, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00018485185185185184, | |
| "loss": 0.594, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.00018448148148148149, | |
| "loss": 0.7857, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00018411111111111113, | |
| "loss": 0.4482, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.00018374074074074074, | |
| "loss": 0.5306, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00018337037037037036, | |
| "loss": 0.5849, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.000183, | |
| "loss": 0.848, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00018262962962962965, | |
| "loss": 0.4931, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00018225925925925926, | |
| "loss": 0.4542, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.0001818888888888889, | |
| "loss": 0.5918, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00018151851851851852, | |
| "loss": 0.5543, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00018114814814814816, | |
| "loss": 0.6295, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.00018077777777777778, | |
| "loss": 0.6341, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.00018040740740740742, | |
| "loss": 0.5688, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00018003703703703704, | |
| "loss": 0.646, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00017966666666666668, | |
| "loss": 0.3895, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.0001792962962962963, | |
| "loss": 0.4572, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.00017892592592592594, | |
| "loss": 0.3104, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.00017855555555555555, | |
| "loss": 0.366, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.0001781851851851852, | |
| "loss": 0.3954, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.00017781481481481484, | |
| "loss": 0.3588, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.00017744444444444445, | |
| "loss": 0.3124, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.00017707407407407407, | |
| "loss": 0.4456, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.0001767037037037037, | |
| "loss": 0.3196, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.00017633333333333335, | |
| "loss": 0.3449, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.00017596296296296297, | |
| "loss": 0.4789, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.00017559259259259259, | |
| "loss": 0.3636, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 0.00017522222222222223, | |
| "loss": 0.4037, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.00017485185185185187, | |
| "loss": 0.3514, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.0001744814814814815, | |
| "loss": 0.3531, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.0001741111111111111, | |
| "loss": 0.3864, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.00017374074074074077, | |
| "loss": 0.3057, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.0001733703703703704, | |
| "loss": 0.3689, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.000173, | |
| "loss": 0.3429, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.00017262962962962962, | |
| "loss": 0.4835, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.0001722592592592593, | |
| "loss": 0.584, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 0.0001718888888888889, | |
| "loss": 0.4278, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.00017151851851851852, | |
| "loss": 0.3338, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.00017114814814814814, | |
| "loss": 0.3246, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.00017077777777777778, | |
| "loss": 0.4246, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.00017040740740740742, | |
| "loss": 0.5964, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.00017003703703703704, | |
| "loss": 0.5198, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 0.00016966666666666668, | |
| "loss": 0.4489, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 0.0001692962962962963, | |
| "loss": 0.204, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 0.00016892592592592594, | |
| "loss": 0.2281, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 0.00016855555555555555, | |
| "loss": 0.2293, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.0001681851851851852, | |
| "loss": 0.1439, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 0.0001678148148148148, | |
| "loss": 0.1872, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 0.00016744444444444445, | |
| "loss": 0.2705, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 0.00016707407407407407, | |
| "loss": 0.2329, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 0.0001667037037037037, | |
| "loss": 0.349, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 0.00016633333333333333, | |
| "loss": 0.2587, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 0.00016596296296296297, | |
| "loss": 0.1563, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 0.00016559259259259261, | |
| "loss": 0.1608, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 0.00016522222222222223, | |
| "loss": 0.264, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 0.00016485185185185185, | |
| "loss": 0.2586, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 0.0001644814814814815, | |
| "loss": 0.2584, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 0.00016411111111111113, | |
| "loss": 0.1729, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 0.00016374074074074075, | |
| "loss": 0.3161, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.00016337037037037036, | |
| "loss": 0.1579, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 0.000163, | |
| "loss": 0.159, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 0.00016262962962962965, | |
| "loss": 0.2138, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 0.00016225925925925926, | |
| "loss": 0.2743, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.00016188888888888888, | |
| "loss": 0.2057, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 0.00016151851851851855, | |
| "loss": 0.2297, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 0.00016114814814814816, | |
| "loss": 0.3396, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 0.00016077777777777778, | |
| "loss": 0.2853, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 0.0001604074074074074, | |
| "loss": 0.2184, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.00016003703703703707, | |
| "loss": 0.2157, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 0.00015966666666666668, | |
| "loss": 0.1986, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 0.0001592962962962963, | |
| "loss": 0.1425, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 0.0001589259259259259, | |
| "loss": 0.1571, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.00015855555555555558, | |
| "loss": 0.1702, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 0.0001581851851851852, | |
| "loss": 0.0456, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 0.0001578148148148148, | |
| "loss": 0.1112, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 0.00015744444444444446, | |
| "loss": 0.0581, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 0.00015707407407407407, | |
| "loss": 0.161, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 0.00015670370370370371, | |
| "loss": 0.2081, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 0.00015633333333333333, | |
| "loss": 0.1118, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 0.00015596296296296297, | |
| "loss": 0.1484, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 0.0001555925925925926, | |
| "loss": 0.1166, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 0.00015522222222222223, | |
| "loss": 0.162, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 0.00015485185185185187, | |
| "loss": 0.1473, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 0.0001544814814814815, | |
| "loss": 0.1251, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 0.0001541111111111111, | |
| "loss": 0.1109, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 0.00015374074074074075, | |
| "loss": 0.1666, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 0.0001533703703703704, | |
| "loss": 0.1386, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 0.000153, | |
| "loss": 0.0548, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 0.00015262962962962962, | |
| "loss": 0.1446, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.00015225925925925926, | |
| "loss": 0.3583, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 0.0001518888888888889, | |
| "loss": 0.3026, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 0.00015151851851851852, | |
| "loss": 0.2623, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 0.00015114814814814814, | |
| "loss": 0.2197, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 0.0001507777777777778, | |
| "loss": 0.1991, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 0.00015040740740740742, | |
| "loss": 0.1439, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.00015003703703703704, | |
| "loss": 0.1856, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 5.04, | |
| "learning_rate": 0.00014966666666666665, | |
| "loss": 0.0815, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 0.00014929629629629632, | |
| "loss": 0.1795, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 0.00014892592592592594, | |
| "loss": 0.0757, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 0.00014855555555555556, | |
| "loss": 0.037, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 0.00014818518518518517, | |
| "loss": 0.0464, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 5.22, | |
| "learning_rate": 0.00014781481481481484, | |
| "loss": 0.0151, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 0.00014744444444444446, | |
| "loss": 0.0441, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 0.00014707407407407407, | |
| "loss": 0.0703, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 0.00014670370370370371, | |
| "loss": 0.1126, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 0.00014633333333333336, | |
| "loss": 0.066, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 0.00014596296296296297, | |
| "loss": 0.1397, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 0.0001455925925925926, | |
| "loss": 0.0641, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 0.00014522222222222223, | |
| "loss": 0.1555, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 0.00014485185185185187, | |
| "loss": 0.164, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 0.0001444814814814815, | |
| "loss": 0.073, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 0.0001441111111111111, | |
| "loss": 0.1236, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 0.00014374074074074075, | |
| "loss": 0.0639, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 0.00014337037037037036, | |
| "loss": 0.1367, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 0.000143, | |
| "loss": 0.0772, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 0.00014262962962962965, | |
| "loss": 0.0283, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 5.78, | |
| "learning_rate": 0.00014225925925925926, | |
| "loss": 0.0523, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 0.00014188888888888888, | |
| "loss": 0.0502, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 0.00014151851851851852, | |
| "loss": 0.1211, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 0.00014114814814814817, | |
| "loss": 0.0252, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 0.00014077777777777778, | |
| "loss": 0.1756, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 0.0001404074074074074, | |
| "loss": 0.1343, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 0.00014003703703703704, | |
| "loss": 0.111, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 0.00013966666666666668, | |
| "loss": 0.0292, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 0.0001392962962962963, | |
| "loss": 0.033, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 0.00013892592592592591, | |
| "loss": 0.0446, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 0.00013855555555555558, | |
| "loss": 0.0272, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 0.0001381851851851852, | |
| "loss": 0.0359, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 0.00013781481481481481, | |
| "loss": 0.0102, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 0.00013744444444444443, | |
| "loss": 0.0641, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 0.0001370740740740741, | |
| "loss": 0.0454, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 6.33, | |
| "learning_rate": 0.00013670370370370372, | |
| "loss": 0.0134, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.00013633333333333333, | |
| "loss": 0.142, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 0.00013596296296296295, | |
| "loss": 0.0604, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 0.00013559259259259262, | |
| "loss": 0.0709, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 0.00013522222222222223, | |
| "loss": 0.0063, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 0.00013485185185185185, | |
| "loss": 0.0101, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 6.56, | |
| "learning_rate": 0.0001344814814814815, | |
| "loss": 0.0382, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 0.00013411111111111113, | |
| "loss": 0.0049, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 0.00013374074074074075, | |
| "loss": 0.0952, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 0.00013337037037037036, | |
| "loss": 0.049, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 0.000133, | |
| "loss": 0.1004, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 6.74, | |
| "learning_rate": 0.00013262962962962965, | |
| "loss": 0.0295, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 0.00013225925925925927, | |
| "loss": 0.0158, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 6.81, | |
| "learning_rate": 0.0001318888888888889, | |
| "loss": 0.0157, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 0.00013151851851851852, | |
| "loss": 0.1002, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 0.00013114814814814817, | |
| "loss": 0.1301, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 0.00013077777777777778, | |
| "loss": 0.0851, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 0.00013040740740740742, | |
| "loss": 0.0822, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 0.00013003703703703704, | |
| "loss": 0.0388, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 7.04, | |
| "learning_rate": 0.00012966666666666666, | |
| "loss": 0.0098, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 0.0001292962962962963, | |
| "loss": 0.0563, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 7.11, | |
| "learning_rate": 0.00012892592592592594, | |
| "loss": 0.0362, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 7.15, | |
| "learning_rate": 0.00012855555555555556, | |
| "loss": 0.0491, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 0.00012818518518518517, | |
| "loss": 0.0435, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 7.22, | |
| "learning_rate": 0.00012781481481481484, | |
| "loss": 0.0735, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 0.00012744444444444446, | |
| "loss": 0.0654, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 0.00012707407407407407, | |
| "loss": 0.0175, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 0.0001267037037037037, | |
| "loss": 0.1319, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 0.00012633333333333336, | |
| "loss": 0.0109, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 0.00012596296296296297, | |
| "loss": 0.018, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 0.0001255925925925926, | |
| "loss": 0.0495, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 0.0001252222222222222, | |
| "loss": 0.0214, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 0.00012485185185185188, | |
| "loss": 0.0412, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 0.0001244814814814815, | |
| "loss": 0.0253, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 7.59, | |
| "learning_rate": 0.0001241111111111111, | |
| "loss": 0.0485, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 7.63, | |
| "learning_rate": 0.00012374074074074075, | |
| "loss": 0.0292, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 0.0001233703703703704, | |
| "loss": 0.0554, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 0.000123, | |
| "loss": 0.0476, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 0.00012262962962962962, | |
| "loss": 0.0079, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 0.00012225925925925927, | |
| "loss": 0.0417, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 0.0001218888888888889, | |
| "loss": 0.0397, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 0.00012151851851851852, | |
| "loss": 0.08, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 0.00012114814814814814, | |
| "loss": 0.0065, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 0.00012077777777777778, | |
| "loss": 0.0511, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 7.96, | |
| "learning_rate": 0.00012040740740740741, | |
| "loss": 0.1212, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 0.00012003703703703704, | |
| "loss": 0.0828, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 0.00011966666666666668, | |
| "loss": 0.0674, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 8.07, | |
| "learning_rate": 0.0001192962962962963, | |
| "loss": 0.0968, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 0.00011892592592592593, | |
| "loss": 0.0061, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 0.00011855555555555556, | |
| "loss": 0.0225, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 0.0001181851851851852, | |
| "loss": 0.0875, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 0.00011781481481481482, | |
| "loss": 0.0385, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 0.00011744444444444445, | |
| "loss": 0.0058, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 8.3, | |
| "learning_rate": 0.00011707407407407407, | |
| "loss": 0.011, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 0.00011670370370370372, | |
| "loss": 0.0127, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 8.37, | |
| "learning_rate": 0.00011633333333333333, | |
| "loss": 0.0293, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 0.00011596296296296296, | |
| "loss": 0.0221, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 8.44, | |
| "learning_rate": 0.0001155925925925926, | |
| "loss": 0.0031, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 8.48, | |
| "learning_rate": 0.00011522222222222223, | |
| "loss": 0.0393, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 0.00011485185185185185, | |
| "loss": 0.0041, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 0.00011448148148148148, | |
| "loss": 0.0314, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 0.00011411111111111112, | |
| "loss": 0.0586, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 0.00011374074074074075, | |
| "loss": 0.0168, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 8.67, | |
| "learning_rate": 0.00011337037037037037, | |
| "loss": 0.0052, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 0.000113, | |
| "loss": 0.0057, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 0.00011262962962962964, | |
| "loss": 0.1525, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 0.00011225925925925927, | |
| "loss": 0.0104, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 8.81, | |
| "learning_rate": 0.00011188888888888888, | |
| "loss": 0.0588, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 0.00011151851851851854, | |
| "loss": 0.0024, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 0.00011114814814814815, | |
| "loss": 0.0084, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 0.00011077777777777778, | |
| "loss": 0.0052, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 0.0001104074074074074, | |
| "loss": 0.0303, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 0.00011003703703703706, | |
| "loss": 0.013, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 0.00010966666666666667, | |
| "loss": 0.0294, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 0.0001092962962962963, | |
| "loss": 0.0257, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 0.00010892592592592592, | |
| "loss": 0.002, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 0.00010855555555555557, | |
| "loss": 0.0087, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 0.00010818518518518519, | |
| "loss": 0.0082, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 0.00010781481481481482, | |
| "loss": 0.0066, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 0.00010744444444444446, | |
| "loss": 0.0604, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 0.00010707407407407408, | |
| "loss": 0.0146, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 0.0001067037037037037, | |
| "loss": 0.0829, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 0.00010633333333333333, | |
| "loss": 0.0169, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 0.00010596296296296298, | |
| "loss": 0.0252, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 0.00010559259259259259, | |
| "loss": 0.0368, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 0.00010522222222222222, | |
| "loss": 0.0042, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 0.00010485185185185186, | |
| "loss": 0.0369, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 9.56, | |
| "learning_rate": 0.0001044814814814815, | |
| "loss": 0.0191, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 0.00010411111111111111, | |
| "loss": 0.0035, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 9.63, | |
| "learning_rate": 0.00010374074074074074, | |
| "loss": 0.0121, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 0.00010337037037037038, | |
| "loss": 0.0019, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 9.7, | |
| "learning_rate": 0.00010300000000000001, | |
| "loss": 0.0376, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 0.00010262962962962963, | |
| "loss": 0.0098, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 9.78, | |
| "learning_rate": 0.00010225925925925925, | |
| "loss": 0.0348, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 9.81, | |
| "learning_rate": 0.0001018888888888889, | |
| "loss": 0.0035, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 9.85, | |
| "learning_rate": 0.00010151851851851853, | |
| "loss": 0.0016, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 9.89, | |
| "learning_rate": 0.00010114814814814814, | |
| "loss": 0.002, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 9.93, | |
| "learning_rate": 0.0001007777777777778, | |
| "loss": 0.0015, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 0.00010040740740740741, | |
| "loss": 0.0017, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.00010003703703703704, | |
| "loss": 0.0491, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 10.04, | |
| "learning_rate": 9.966666666666667e-05, | |
| "loss": 0.0025, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 10.07, | |
| "learning_rate": 9.92962962962963e-05, | |
| "loss": 0.0024, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 10.11, | |
| "learning_rate": 9.892592592592593e-05, | |
| "loss": 0.0198, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 10.15, | |
| "learning_rate": 9.855555555555556e-05, | |
| "loss": 0.0036, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 10.19, | |
| "learning_rate": 9.818518518518519e-05, | |
| "loss": 0.0018, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 10.22, | |
| "learning_rate": 9.781481481481482e-05, | |
| "loss": 0.002, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "learning_rate": 9.744444444444445e-05, | |
| "loss": 0.0051, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 10.3, | |
| "learning_rate": 9.707407407407409e-05, | |
| "loss": 0.0015, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 10.33, | |
| "learning_rate": 9.67037037037037e-05, | |
| "loss": 0.0082, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 9.633333333333335e-05, | |
| "loss": 0.0134, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 10.41, | |
| "learning_rate": 9.596296296296296e-05, | |
| "loss": 0.0016, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 10.44, | |
| "learning_rate": 9.55925925925926e-05, | |
| "loss": 0.0025, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "learning_rate": 9.522222222222222e-05, | |
| "loss": 0.0018, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 10.52, | |
| "learning_rate": 9.485185185185187e-05, | |
| "loss": 0.0017, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 10.56, | |
| "learning_rate": 9.448148148148148e-05, | |
| "loss": 0.0014, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 10.59, | |
| "learning_rate": 9.411111111111111e-05, | |
| "loss": 0.0013, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 10.63, | |
| "learning_rate": 9.374074074074074e-05, | |
| "loss": 0.003, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 10.67, | |
| "learning_rate": 9.34074074074074e-05, | |
| "loss": 0.015, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 10.7, | |
| "learning_rate": 9.303703703703705e-05, | |
| "loss": 0.0303, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 10.74, | |
| "learning_rate": 9.266666666666666e-05, | |
| "loss": 0.0267, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 10.78, | |
| "learning_rate": 9.229629629629631e-05, | |
| "loss": 0.0078, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 10.81, | |
| "learning_rate": 9.192592592592592e-05, | |
| "loss": 0.0054, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 9.155555555555557e-05, | |
| "loss": 0.0102, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 10.89, | |
| "learning_rate": 9.118518518518518e-05, | |
| "loss": 0.0014, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 10.93, | |
| "learning_rate": 9.081481481481482e-05, | |
| "loss": 0.0039, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 10.96, | |
| "learning_rate": 9.044444444444445e-05, | |
| "loss": 0.0015, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "learning_rate": 9.007407407407408e-05, | |
| "loss": 0.1549, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 11.04, | |
| "learning_rate": 8.970370370370371e-05, | |
| "loss": 0.0015, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 11.07, | |
| "learning_rate": 8.933333333333334e-05, | |
| "loss": 0.0504, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 11.11, | |
| "learning_rate": 8.896296296296297e-05, | |
| "loss": 0.0087, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 11.11, | |
| "eval_accuracy": 0.8972222222222223, | |
| "eval_loss": 0.5567966103553772, | |
| "eval_runtime": 12.8044, | |
| "eval_samples_per_second": 84.346, | |
| "eval_steps_per_second": 10.543, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 11.15, | |
| "learning_rate": 8.85925925925926e-05, | |
| "loss": 0.0011, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "learning_rate": 8.822222222222223e-05, | |
| "loss": 0.0015, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 11.22, | |
| "learning_rate": 8.785185185185186e-05, | |
| "loss": 0.0012, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 11.26, | |
| "learning_rate": 8.748148148148149e-05, | |
| "loss": 0.0011, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 11.3, | |
| "learning_rate": 8.711111111111112e-05, | |
| "loss": 0.0129, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 11.33, | |
| "learning_rate": 8.674074074074074e-05, | |
| "loss": 0.0011, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 11.37, | |
| "learning_rate": 8.637037037037037e-05, | |
| "loss": 0.0048, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 11.41, | |
| "learning_rate": 8.6e-05, | |
| "loss": 0.0012, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 11.44, | |
| "learning_rate": 8.562962962962963e-05, | |
| "loss": 0.0103, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "learning_rate": 8.525925925925926e-05, | |
| "loss": 0.0457, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 11.52, | |
| "learning_rate": 8.488888888888889e-05, | |
| "loss": 0.0012, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 11.56, | |
| "learning_rate": 8.451851851851852e-05, | |
| "loss": 0.0079, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 8.414814814814815e-05, | |
| "loss": 0.0326, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 8.377777777777778e-05, | |
| "loss": 0.0017, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 11.67, | |
| "learning_rate": 8.340740740740741e-05, | |
| "loss": 0.001, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 8.303703703703705e-05, | |
| "loss": 0.0184, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 11.74, | |
| "learning_rate": 8.266666666666667e-05, | |
| "loss": 0.002, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 11.78, | |
| "learning_rate": 8.229629629629631e-05, | |
| "loss": 0.0011, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 11.81, | |
| "learning_rate": 8.192592592592592e-05, | |
| "loss": 0.0026, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 11.85, | |
| "learning_rate": 8.155555555555557e-05, | |
| "loss": 0.0038, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 11.89, | |
| "learning_rate": 8.118518518518518e-05, | |
| "loss": 0.0238, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 11.93, | |
| "learning_rate": 8.081481481481483e-05, | |
| "loss": 0.0394, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 11.96, | |
| "learning_rate": 8.044444444444444e-05, | |
| "loss": 0.0328, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 8.007407407407408e-05, | |
| "loss": 0.0078, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 12.04, | |
| "learning_rate": 7.97037037037037e-05, | |
| "loss": 0.0012, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "learning_rate": 7.933333333333334e-05, | |
| "loss": 0.0517, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 12.11, | |
| "learning_rate": 7.896296296296297e-05, | |
| "loss": 0.0011, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 12.15, | |
| "learning_rate": 7.85925925925926e-05, | |
| "loss": 0.0011, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 12.19, | |
| "learning_rate": 7.822222222222223e-05, | |
| "loss": 0.0066, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 7.785185185185186e-05, | |
| "loss": 0.0793, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "learning_rate": 7.748148148148149e-05, | |
| "loss": 0.0305, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 12.3, | |
| "learning_rate": 7.711111111111112e-05, | |
| "loss": 0.0012, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 12.33, | |
| "learning_rate": 7.674074074074075e-05, | |
| "loss": 0.001, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 12.37, | |
| "learning_rate": 7.637037037037038e-05, | |
| "loss": 0.0017, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 12.41, | |
| "learning_rate": 7.6e-05, | |
| "loss": 0.0029, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 12.44, | |
| "learning_rate": 7.562962962962963e-05, | |
| "loss": 0.0114, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 12.48, | |
| "learning_rate": 7.525925925925926e-05, | |
| "loss": 0.001, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 12.52, | |
| "learning_rate": 7.488888888888889e-05, | |
| "loss": 0.0009, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 7.451851851851852e-05, | |
| "loss": 0.0039, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 12.59, | |
| "learning_rate": 7.414814814814815e-05, | |
| "loss": 0.008, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 12.63, | |
| "learning_rate": 7.377777777777778e-05, | |
| "loss": 0.0038, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 12.67, | |
| "learning_rate": 7.340740740740741e-05, | |
| "loss": 0.029, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 12.7, | |
| "learning_rate": 7.303703703703704e-05, | |
| "loss": 0.0019, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 12.74, | |
| "learning_rate": 7.266666666666667e-05, | |
| "loss": 0.0018, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 12.78, | |
| "learning_rate": 7.22962962962963e-05, | |
| "loss": 0.0211, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 12.81, | |
| "learning_rate": 7.192592592592592e-05, | |
| "loss": 0.0192, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 12.85, | |
| "learning_rate": 7.155555555555555e-05, | |
| "loss": 0.0181, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 12.89, | |
| "learning_rate": 7.118518518518518e-05, | |
| "loss": 0.0226, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 12.93, | |
| "learning_rate": 7.081481481481483e-05, | |
| "loss": 0.0083, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 12.96, | |
| "learning_rate": 7.044444444444444e-05, | |
| "loss": 0.013, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "learning_rate": 7.007407407407408e-05, | |
| "loss": 0.0008, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 6.97037037037037e-05, | |
| "loss": 0.0701, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 13.07, | |
| "learning_rate": 6.933333333333334e-05, | |
| "loss": 0.0032, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 13.11, | |
| "learning_rate": 6.896296296296296e-05, | |
| "loss": 0.0021, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 13.15, | |
| "learning_rate": 6.85925925925926e-05, | |
| "loss": 0.001, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 13.19, | |
| "learning_rate": 6.822222222222222e-05, | |
| "loss": 0.0289, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 13.22, | |
| "learning_rate": 6.785185185185186e-05, | |
| "loss": 0.001, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 13.26, | |
| "learning_rate": 6.748148148148149e-05, | |
| "loss": 0.0358, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 13.3, | |
| "learning_rate": 6.711111111111112e-05, | |
| "loss": 0.001, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 13.33, | |
| "learning_rate": 6.674074074074075e-05, | |
| "loss": 0.008, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 13.37, | |
| "learning_rate": 6.637037037037038e-05, | |
| "loss": 0.0375, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 6.6e-05, | |
| "loss": 0.0009, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 6.562962962962963e-05, | |
| "loss": 0.0039, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 13.48, | |
| "learning_rate": 6.525925925925926e-05, | |
| "loss": 0.0033, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 13.52, | |
| "learning_rate": 6.488888888888889e-05, | |
| "loss": 0.0008, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 6.451851851851852e-05, | |
| "loss": 0.0111, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 13.59, | |
| "learning_rate": 6.414814814814815e-05, | |
| "loss": 0.0189, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 13.63, | |
| "learning_rate": 6.377777777777778e-05, | |
| "loss": 0.0238, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 13.67, | |
| "learning_rate": 6.340740740740741e-05, | |
| "loss": 0.0315, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 13.7, | |
| "learning_rate": 6.303703703703704e-05, | |
| "loss": 0.0007, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 13.74, | |
| "learning_rate": 6.266666666666667e-05, | |
| "loss": 0.0013, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 13.78, | |
| "learning_rate": 6.22962962962963e-05, | |
| "loss": 0.0079, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 13.81, | |
| "learning_rate": 6.192592592592593e-05, | |
| "loss": 0.0008, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 6.155555555555555e-05, | |
| "loss": 0.0008, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 13.89, | |
| "learning_rate": 6.118518518518518e-05, | |
| "loss": 0.0007, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 13.93, | |
| "learning_rate": 6.081481481481481e-05, | |
| "loss": 0.0046, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 13.96, | |
| "learning_rate": 6.044444444444445e-05, | |
| "loss": 0.0009, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "learning_rate": 6.007407407407407e-05, | |
| "loss": 0.0007, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 5.970370370370371e-05, | |
| "loss": 0.0007, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 14.07, | |
| "learning_rate": 5.9333333333333343e-05, | |
| "loss": 0.0008, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 14.11, | |
| "learning_rate": 5.8962962962962966e-05, | |
| "loss": 0.0007, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 5.85925925925926e-05, | |
| "loss": 0.0008, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 14.19, | |
| "learning_rate": 5.8222222222222224e-05, | |
| "loss": 0.0028, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 14.22, | |
| "learning_rate": 5.785185185185186e-05, | |
| "loss": 0.0112, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 14.26, | |
| "learning_rate": 5.748148148148148e-05, | |
| "loss": 0.0007, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 14.3, | |
| "learning_rate": 5.711111111111112e-05, | |
| "loss": 0.0076, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 14.33, | |
| "learning_rate": 5.674074074074074e-05, | |
| "loss": 0.0007, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 14.37, | |
| "learning_rate": 5.637037037037037e-05, | |
| "loss": 0.0038, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 14.41, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 0.0007, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 14.44, | |
| "learning_rate": 5.562962962962963e-05, | |
| "loss": 0.0006, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 14.48, | |
| "learning_rate": 5.5259259259259264e-05, | |
| "loss": 0.0094, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 14.52, | |
| "learning_rate": 5.488888888888889e-05, | |
| "loss": 0.0068, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 14.56, | |
| "learning_rate": 5.451851851851852e-05, | |
| "loss": 0.0007, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 14.59, | |
| "learning_rate": 5.4148148148148145e-05, | |
| "loss": 0.0007, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 14.63, | |
| "learning_rate": 5.377777777777778e-05, | |
| "loss": 0.0006, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 14.67, | |
| "learning_rate": 5.34074074074074e-05, | |
| "loss": 0.0006, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 5.303703703703704e-05, | |
| "loss": 0.0229, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 14.74, | |
| "learning_rate": 5.266666666666666e-05, | |
| "loss": 0.0292, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 14.78, | |
| "learning_rate": 5.22962962962963e-05, | |
| "loss": 0.0006, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 14.81, | |
| "learning_rate": 5.1925925925925933e-05, | |
| "loss": 0.0006, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 5.1555555555555556e-05, | |
| "loss": 0.0097, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 14.89, | |
| "learning_rate": 5.118518518518519e-05, | |
| "loss": 0.0006, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 14.93, | |
| "learning_rate": 5.0814814814814814e-05, | |
| "loss": 0.0007, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 14.96, | |
| "learning_rate": 5.044444444444445e-05, | |
| "loss": 0.0006, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 5.007407407407407e-05, | |
| "loss": 0.0006, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 15.04, | |
| "learning_rate": 4.970370370370371e-05, | |
| "loss": 0.0006, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 15.07, | |
| "learning_rate": 4.933333333333334e-05, | |
| "loss": 0.0006, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 15.11, | |
| "learning_rate": 4.896296296296297e-05, | |
| "loss": 0.0006, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 15.15, | |
| "learning_rate": 4.8592592592592596e-05, | |
| "loss": 0.0055, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 15.19, | |
| "learning_rate": 4.8222222222222225e-05, | |
| "loss": 0.0006, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 15.22, | |
| "learning_rate": 4.7851851851851854e-05, | |
| "loss": 0.0006, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 4.7481481481481483e-05, | |
| "loss": 0.0006, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 15.3, | |
| "learning_rate": 4.711111111111111e-05, | |
| "loss": 0.0008, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 15.33, | |
| "learning_rate": 4.674074074074074e-05, | |
| "loss": 0.0044, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 15.37, | |
| "learning_rate": 4.637037037037038e-05, | |
| "loss": 0.0006, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 15.41, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.0006, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 15.44, | |
| "learning_rate": 4.5629629629629636e-05, | |
| "loss": 0.0076, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 15.48, | |
| "learning_rate": 4.5259259259259265e-05, | |
| "loss": 0.0006, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 15.52, | |
| "learning_rate": 4.4888888888888894e-05, | |
| "loss": 0.0006, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 15.56, | |
| "learning_rate": 4.4518518518518523e-05, | |
| "loss": 0.0053, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 15.59, | |
| "learning_rate": 4.414814814814815e-05, | |
| "loss": 0.0006, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 15.63, | |
| "learning_rate": 4.377777777777778e-05, | |
| "loss": 0.0006, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 4.340740740740741e-05, | |
| "loss": 0.0032, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 4.303703703703704e-05, | |
| "loss": 0.015, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 15.74, | |
| "learning_rate": 4.266666666666667e-05, | |
| "loss": 0.0006, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 15.78, | |
| "learning_rate": 4.22962962962963e-05, | |
| "loss": 0.0046, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 15.81, | |
| "learning_rate": 4.192592592592593e-05, | |
| "loss": 0.0006, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 4.155555555555556e-05, | |
| "loss": 0.0006, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 15.89, | |
| "learning_rate": 4.1185185185185186e-05, | |
| "loss": 0.0006, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 15.93, | |
| "learning_rate": 4.0814814814814815e-05, | |
| "loss": 0.0006, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 15.96, | |
| "learning_rate": 4.0444444444444444e-05, | |
| "loss": 0.0006, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "learning_rate": 4.007407407407407e-05, | |
| "loss": 0.0006, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 16.04, | |
| "learning_rate": 3.97037037037037e-05, | |
| "loss": 0.0006, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 16.07, | |
| "learning_rate": 3.933333333333333e-05, | |
| "loss": 0.0074, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 16.11, | |
| "learning_rate": 3.896296296296296e-05, | |
| "loss": 0.0006, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 16.15, | |
| "learning_rate": 3.85925925925926e-05, | |
| "loss": 0.0005, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 16.19, | |
| "learning_rate": 3.8222222222222226e-05, | |
| "loss": 0.0006, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 16.22, | |
| "learning_rate": 3.7851851851851855e-05, | |
| "loss": 0.006, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 16.26, | |
| "learning_rate": 3.7481481481481484e-05, | |
| "loss": 0.0006, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 16.3, | |
| "learning_rate": 3.7111111111111113e-05, | |
| "loss": 0.0006, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 16.33, | |
| "learning_rate": 3.674074074074074e-05, | |
| "loss": 0.0005, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 16.37, | |
| "learning_rate": 3.637037037037037e-05, | |
| "loss": 0.0104, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 16.41, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.0005, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 16.44, | |
| "learning_rate": 3.562962962962963e-05, | |
| "loss": 0.0056, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 16.48, | |
| "learning_rate": 3.525925925925926e-05, | |
| "loss": 0.0108, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 16.52, | |
| "learning_rate": 3.4888888888888895e-05, | |
| "loss": 0.0005, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 16.56, | |
| "learning_rate": 3.4518518518518524e-05, | |
| "loss": 0.0005, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 16.59, | |
| "learning_rate": 3.4148148148148153e-05, | |
| "loss": 0.0005, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 16.63, | |
| "learning_rate": 3.377777777777778e-05, | |
| "loss": 0.0005, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 3.340740740740741e-05, | |
| "loss": 0.0005, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 16.7, | |
| "learning_rate": 3.303703703703704e-05, | |
| "loss": 0.0005, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 16.74, | |
| "learning_rate": 3.266666666666667e-05, | |
| "loss": 0.0005, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 16.78, | |
| "learning_rate": 3.22962962962963e-05, | |
| "loss": 0.0005, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 3.192592592592593e-05, | |
| "loss": 0.0005, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 16.85, | |
| "learning_rate": 3.155555555555556e-05, | |
| "loss": 0.0005, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 16.89, | |
| "learning_rate": 3.118518518518519e-05, | |
| "loss": 0.0043, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 16.93, | |
| "learning_rate": 3.0814814814814816e-05, | |
| "loss": 0.0005, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 16.96, | |
| "learning_rate": 3.044444444444445e-05, | |
| "loss": 0.0005, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "learning_rate": 3.0074074074074078e-05, | |
| "loss": 0.0005, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 17.04, | |
| "learning_rate": 2.9703703703703707e-05, | |
| "loss": 0.0005, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 17.07, | |
| "learning_rate": 2.9333333333333336e-05, | |
| "loss": 0.0068, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 17.11, | |
| "learning_rate": 2.8962962962962965e-05, | |
| "loss": 0.0005, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 17.15, | |
| "learning_rate": 2.8592592592592594e-05, | |
| "loss": 0.0048, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 17.19, | |
| "learning_rate": 2.8222222222222223e-05, | |
| "loss": 0.0005, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 17.22, | |
| "learning_rate": 2.7851851851851853e-05, | |
| "loss": 0.0005, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 17.26, | |
| "learning_rate": 2.7481481481481482e-05, | |
| "loss": 0.0065, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 17.3, | |
| "learning_rate": 2.7111111111111114e-05, | |
| "loss": 0.0005, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 17.33, | |
| "learning_rate": 2.6740740740740743e-05, | |
| "loss": 0.0048, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 17.37, | |
| "learning_rate": 2.6370370370370373e-05, | |
| "loss": 0.0005, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 17.41, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.0005, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 17.44, | |
| "learning_rate": 2.562962962962963e-05, | |
| "loss": 0.0005, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 17.48, | |
| "learning_rate": 2.525925925925926e-05, | |
| "loss": 0.004, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 17.52, | |
| "learning_rate": 2.488888888888889e-05, | |
| "loss": 0.0005, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 17.56, | |
| "learning_rate": 2.451851851851852e-05, | |
| "loss": 0.0005, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 17.59, | |
| "learning_rate": 2.414814814814815e-05, | |
| "loss": 0.0005, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 17.63, | |
| "learning_rate": 2.377777777777778e-05, | |
| "loss": 0.0005, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 17.67, | |
| "learning_rate": 2.340740740740741e-05, | |
| "loss": 0.0005, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 17.7, | |
| "learning_rate": 2.303703703703704e-05, | |
| "loss": 0.0005, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 17.74, | |
| "learning_rate": 2.2666666666666668e-05, | |
| "loss": 0.0032, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 17.78, | |
| "learning_rate": 2.2296296296296297e-05, | |
| "loss": 0.0005, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 17.81, | |
| "learning_rate": 2.1925925925925926e-05, | |
| "loss": 0.0005, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 17.85, | |
| "learning_rate": 2.1555555555555555e-05, | |
| "loss": 0.0005, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 17.89, | |
| "learning_rate": 2.1185185185185184e-05, | |
| "loss": 0.0005, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 17.93, | |
| "learning_rate": 2.0814814814814813e-05, | |
| "loss": 0.0005, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 17.96, | |
| "learning_rate": 2.0444444444444446e-05, | |
| "loss": 0.0081, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "learning_rate": 2.0074074074074075e-05, | |
| "loss": 0.0064, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 18.04, | |
| "learning_rate": 1.9703703703703704e-05, | |
| "loss": 0.0005, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 18.07, | |
| "learning_rate": 1.9333333333333333e-05, | |
| "loss": 0.0005, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 18.11, | |
| "learning_rate": 1.8962962962962963e-05, | |
| "loss": 0.0005, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 18.15, | |
| "learning_rate": 1.8592592592592595e-05, | |
| "loss": 0.0049, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 18.19, | |
| "learning_rate": 1.8222222222222224e-05, | |
| "loss": 0.0005, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 1.7851851851851853e-05, | |
| "loss": 0.0005, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 18.26, | |
| "learning_rate": 1.7481481481481483e-05, | |
| "loss": 0.0039, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 18.3, | |
| "learning_rate": 1.7111111111111112e-05, | |
| "loss": 0.0034, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 18.33, | |
| "learning_rate": 1.674074074074074e-05, | |
| "loss": 0.0033, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 18.37, | |
| "learning_rate": 1.6370370370370374e-05, | |
| "loss": 0.0005, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 18.41, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.0005, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 18.44, | |
| "learning_rate": 1.5629629629629632e-05, | |
| "loss": 0.0005, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 18.48, | |
| "learning_rate": 1.5259259259259258e-05, | |
| "loss": 0.0005, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 18.52, | |
| "learning_rate": 1.4888888888888888e-05, | |
| "loss": 0.0076, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 18.56, | |
| "learning_rate": 1.4518518518518521e-05, | |
| "loss": 0.0076, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 18.59, | |
| "learning_rate": 1.4148148148148148e-05, | |
| "loss": 0.0005, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 18.63, | |
| "learning_rate": 1.3777777777777778e-05, | |
| "loss": 0.0005, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 18.67, | |
| "learning_rate": 1.3407407407407407e-05, | |
| "loss": 0.0005, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 18.7, | |
| "learning_rate": 1.3037037037037036e-05, | |
| "loss": 0.0005, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 18.74, | |
| "learning_rate": 1.2666666666666668e-05, | |
| "loss": 0.0005, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 18.78, | |
| "learning_rate": 1.2296296296296298e-05, | |
| "loss": 0.0005, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 18.81, | |
| "learning_rate": 1.1925925925925927e-05, | |
| "loss": 0.0061, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 18.85, | |
| "learning_rate": 1.1555555555555556e-05, | |
| "loss": 0.0005, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 18.89, | |
| "learning_rate": 1.1185185185185187e-05, | |
| "loss": 0.0005, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 18.93, | |
| "learning_rate": 1.0814814814814814e-05, | |
| "loss": 0.0056, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 18.96, | |
| "learning_rate": 1.0444444444444445e-05, | |
| "loss": 0.0005, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "learning_rate": 1.0074074074074074e-05, | |
| "loss": 0.0004, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 9.703703703703703e-06, | |
| "loss": 0.0004, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 19.07, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.0005, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 19.11, | |
| "learning_rate": 8.962962962962963e-06, | |
| "loss": 0.0049, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 19.15, | |
| "learning_rate": 8.592592592592593e-06, | |
| "loss": 0.005, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 8.222222222222223e-06, | |
| "loss": 0.0005, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 19.22, | |
| "learning_rate": 7.851851851851853e-06, | |
| "loss": 0.0043, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 19.26, | |
| "learning_rate": 7.481481481481483e-06, | |
| "loss": 0.0051, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 19.3, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.0005, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 19.33, | |
| "learning_rate": 6.74074074074074e-06, | |
| "loss": 0.0048, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 19.37, | |
| "learning_rate": 6.370370370370371e-06, | |
| "loss": 0.0051, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 19.41, | |
| "learning_rate": 6e-06, | |
| "loss": 0.0048, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 19.44, | |
| "learning_rate": 5.62962962962963e-06, | |
| "loss": 0.0055, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 19.48, | |
| "learning_rate": 5.259259259259259e-06, | |
| "loss": 0.0005, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 19.52, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.0005, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 19.56, | |
| "learning_rate": 4.5185185185185185e-06, | |
| "loss": 0.0005, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 19.59, | |
| "learning_rate": 4.1481481481481485e-06, | |
| "loss": 0.0005, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 19.63, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.0004, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 19.67, | |
| "learning_rate": 3.4074074074074077e-06, | |
| "loss": 0.0004, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 19.7, | |
| "learning_rate": 3.0370370370370372e-06, | |
| "loss": 0.0005, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 19.74, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0005, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 19.78, | |
| "learning_rate": 2.2962962962962964e-06, | |
| "loss": 0.0005, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 19.81, | |
| "learning_rate": 1.925925925925926e-06, | |
| "loss": 0.0004, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 19.85, | |
| "learning_rate": 1.5555555555555556e-06, | |
| "loss": 0.0004, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 19.89, | |
| "learning_rate": 1.1851851851851852e-06, | |
| "loss": 0.0004, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 19.93, | |
| "learning_rate": 8.148148148148147e-07, | |
| "loss": 0.0005, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 19.96, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0005, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 7.407407407407407e-08, | |
| "loss": 0.0005, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 5400, | |
| "total_flos": 6.69572795818967e+18, | |
| "train_loss": 0.14805768244520381, | |
| "train_runtime": 1678.3865, | |
| "train_samples_per_second": 51.478, | |
| "train_steps_per_second": 3.217 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 5400, | |
| "num_train_epochs": 20, | |
| "save_steps": 6000, | |
| "total_flos": 6.69572795818967e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |