| { | |
| "best_metric": 0.6497328281402588, | |
| "best_model_checkpoint": "./vit-large-beans-demo-v5/checkpoint-1000", | |
| "epoch": 2.0, | |
| "eval_steps": 1000, | |
| "global_step": 1596, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.937343358395991e-05, | |
| "loss": 1.229, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.87468671679198e-05, | |
| "loss": 1.0071, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.81203007518797e-05, | |
| "loss": 0.9902, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.749373433583961e-05, | |
| "loss": 0.9331, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.68671679197995e-05, | |
| "loss": 0.9112, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.62406015037594e-05, | |
| "loss": 0.904, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.56140350877193e-05, | |
| "loss": 0.8937, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.49874686716792e-05, | |
| "loss": 0.8887, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.436090225563911e-05, | |
| "loss": 0.8668, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.373433583959899e-05, | |
| "loss": 0.8551, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.31077694235589e-05, | |
| "loss": 0.8759, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.24812030075188e-05, | |
| "loss": 0.8171, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.18546365914787e-05, | |
| "loss": 0.8854, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.12280701754386e-05, | |
| "loss": 0.8901, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.06015037593985e-05, | |
| "loss": 0.7488, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.99749373433584e-05, | |
| "loss": 0.8285, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.93483709273183e-05, | |
| "loss": 0.886, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.87218045112782e-05, | |
| "loss": 0.8396, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.80952380952381e-05, | |
| "loss": 0.802, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.7468671679198e-05, | |
| "loss": 0.8519, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.68421052631579e-05, | |
| "loss": 0.8145, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.621553884711779e-05, | |
| "loss": 0.8867, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.55889724310777e-05, | |
| "loss": 0.8109, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.49624060150376e-05, | |
| "loss": 0.8546, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.43358395989975e-05, | |
| "loss": 0.7561, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 8.37092731829574e-05, | |
| "loss": 0.776, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 8.308270676691729e-05, | |
| "loss": 0.7976, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 8.24561403508772e-05, | |
| "loss": 0.8036, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 8.182957393483711e-05, | |
| "loss": 0.7369, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 8.120300751879699e-05, | |
| "loss": 0.881, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 8.05764411027569e-05, | |
| "loss": 0.683, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 7.994987468671679e-05, | |
| "loss": 0.7664, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 7.93233082706767e-05, | |
| "loss": 0.7325, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 7.869674185463659e-05, | |
| "loss": 0.8024, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 7.807017543859649e-05, | |
| "loss": 0.7889, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 7.74436090225564e-05, | |
| "loss": 0.7871, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 7.68170426065163e-05, | |
| "loss": 0.7085, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 7.619047619047618e-05, | |
| "loss": 0.8221, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 7.556390977443609e-05, | |
| "loss": 0.8302, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 7.4937343358396e-05, | |
| "loss": 0.6974, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 7.43107769423559e-05, | |
| "loss": 0.6954, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 7.368421052631579e-05, | |
| "loss": 0.7266, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 7.30576441102757e-05, | |
| "loss": 0.7308, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 7.243107769423559e-05, | |
| "loss": 0.7764, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 7.18045112781955e-05, | |
| "loss": 0.7621, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 7.11779448621554e-05, | |
| "loss": 0.7511, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 7.055137844611529e-05, | |
| "loss": 0.6622, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 6.99248120300752e-05, | |
| "loss": 0.7743, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 6.929824561403509e-05, | |
| "loss": 0.7547, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.867167919799499e-05, | |
| "loss": 0.758, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 6.80451127819549e-05, | |
| "loss": 0.7378, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 6.741854636591479e-05, | |
| "loss": 0.7583, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 6.67919799498747e-05, | |
| "loss": 0.7618, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.616541353383459e-05, | |
| "loss": 0.6799, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 6.553884711779449e-05, | |
| "loss": 0.6106, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 6.49122807017544e-05, | |
| "loss": 0.7618, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 6.428571428571429e-05, | |
| "loss": 0.6979, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 6.365914786967418e-05, | |
| "loss": 0.6811, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 6.303258145363409e-05, | |
| "loss": 0.7091, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 6.240601503759398e-05, | |
| "loss": 0.7377, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 6.177944862155389e-05, | |
| "loss": 0.6556, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.115288220551379e-05, | |
| "loss": 0.6782, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.052631578947369e-05, | |
| "loss": 0.7018, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.989974937343359e-05, | |
| "loss": 0.7544, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.927318295739349e-05, | |
| "loss": 0.7339, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 5.8646616541353386e-05, | |
| "loss": 0.6871, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 5.802005012531329e-05, | |
| "loss": 0.7215, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.739348370927319e-05, | |
| "loss": 0.6803, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.676691729323309e-05, | |
| "loss": 0.6658, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 5.6140350877192984e-05, | |
| "loss": 0.5731, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 5.5513784461152885e-05, | |
| "loss": 0.6081, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 5.4887218045112786e-05, | |
| "loss": 0.7074, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 5.426065162907269e-05, | |
| "loss": 0.7139, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 5.363408521303258e-05, | |
| "loss": 0.6573, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.300751879699248e-05, | |
| "loss": 0.612, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.2380952380952384e-05, | |
| "loss": 0.6389, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.1754385964912286e-05, | |
| "loss": 0.6942, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.112781954887218e-05, | |
| "loss": 0.6322, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.050125313283208e-05, | |
| "loss": 0.6333, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.987468671679198e-05, | |
| "loss": 0.6327, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 4.9248120300751884e-05, | |
| "loss": 0.5053, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 4.862155388471178e-05, | |
| "loss": 0.4566, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 4.799498746867168e-05, | |
| "loss": 0.503, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 4.736842105263158e-05, | |
| "loss": 0.5174, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 4.674185463659148e-05, | |
| "loss": 0.4601, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 4.6115288220551377e-05, | |
| "loss": 0.5172, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 4.548872180451128e-05, | |
| "loss": 0.4515, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 4.486215538847118e-05, | |
| "loss": 0.4852, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 4.423558897243108e-05, | |
| "loss": 0.5067, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.3609022556390975e-05, | |
| "loss": 0.4752, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 4.298245614035088e-05, | |
| "loss": 0.4933, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 4.235588972431078e-05, | |
| "loss": 0.4793, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 4.172932330827068e-05, | |
| "loss": 0.4608, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 4.110275689223057e-05, | |
| "loss": 0.43, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 4.047619047619048e-05, | |
| "loss": 0.4715, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.9849624060150376e-05, | |
| "loss": 0.4629, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.922305764411028e-05, | |
| "loss": 0.4755, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.859649122807018e-05, | |
| "loss": 0.5432, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.796992481203008e-05, | |
| "loss": 0.5436, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.7343358395989974e-05, | |
| "loss": 0.581, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "eval_accuracy": 0.7335398915569326, | |
| "eval_loss": 0.6497328281402588, | |
| "eval_runtime": 159.466, | |
| "eval_samples_per_second": 24.287, | |
| "eval_steps_per_second": 3.041, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.6716791979949875e-05, | |
| "loss": 0.4371, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.6090225563909776e-05, | |
| "loss": 0.4285, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.546365914786968e-05, | |
| "loss": 0.455, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.483709273182957e-05, | |
| "loss": 0.4088, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.421052631578947e-05, | |
| "loss": 0.5105, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.3583959899749374e-05, | |
| "loss": 0.4458, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.2957393483709276e-05, | |
| "loss": 0.4961, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.233082706766917e-05, | |
| "loss": 0.5237, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.170426065162908e-05, | |
| "loss": 0.4457, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.107769423558897e-05, | |
| "loss": 0.4075, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 3.0451127819548874e-05, | |
| "loss": 0.477, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.9824561403508772e-05, | |
| "loss": 0.3818, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.9197994987468673e-05, | |
| "loss": 0.4805, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.857142857142857e-05, | |
| "loss": 0.4489, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.7944862155388472e-05, | |
| "loss": 0.4282, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.731829573934837e-05, | |
| "loss": 0.4145, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.6691729323308275e-05, | |
| "loss": 0.3852, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.606516290726817e-05, | |
| "loss": 0.4014, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.550125313283208e-05, | |
| "loss": 0.4444, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.487468671679198e-05, | |
| "loss": 0.4709, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.424812030075188e-05, | |
| "loss": 0.4012, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.362155388471178e-05, | |
| "loss": 0.4174, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.2994987468671682e-05, | |
| "loss": 0.4309, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.236842105263158e-05, | |
| "loss": 0.4216, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.174185463659148e-05, | |
| "loss": 0.3635, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.111528822055138e-05, | |
| "loss": 0.4032, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.048872180451128e-05, | |
| "loss": 0.3429, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.9862155388471178e-05, | |
| "loss": 0.3645, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 1.923558897243108e-05, | |
| "loss": 0.4039, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 1.860902255639098e-05, | |
| "loss": 0.393, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 1.7982456140350878e-05, | |
| "loss": 0.4526, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 1.735588972431078e-05, | |
| "loss": 0.4054, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 1.6729323308270677e-05, | |
| "loss": 0.3888, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 1.610275689223058e-05, | |
| "loss": 0.4117, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 1.5476190476190476e-05, | |
| "loss": 0.404, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 1.4849624060150378e-05, | |
| "loss": 0.434, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 1.4223057644110277e-05, | |
| "loss": 0.4473, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 1.3596491228070177e-05, | |
| "loss": 0.3732, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 1.2969924812030076e-05, | |
| "loss": 0.3367, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 1.2343358395989974e-05, | |
| "loss": 0.3256, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 1.1716791979949874e-05, | |
| "loss": 0.347, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 1.1090225563909775e-05, | |
| "loss": 0.4329, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 1.0463659147869675e-05, | |
| "loss": 0.348, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.837092731829574e-06, | |
| "loss": 0.3269, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 9.210526315789474e-06, | |
| "loss": 0.3124, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 8.583959899749373e-06, | |
| "loss": 0.3806, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 7.957393483709273e-06, | |
| "loss": 0.4165, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 7.330827067669173e-06, | |
| "loss": 0.3784, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 6.704260651629073e-06, | |
| "loss": 0.3618, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 6.077694235588972e-06, | |
| "loss": 0.3688, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 5.451127819548872e-06, | |
| "loss": 0.3775, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 4.824561403508772e-06, | |
| "loss": 0.3975, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 4.197994987468672e-06, | |
| "loss": 0.3594, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 3.5714285714285714e-06, | |
| "loss": 0.3771, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 2.9448621553884713e-06, | |
| "loss": 0.3319, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 2.318295739348371e-06, | |
| "loss": 0.3762, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.6917293233082707e-06, | |
| "loss": 0.3299, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.0651629072681704e-06, | |
| "loss": 0.3763, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 4.385964912280702e-07, | |
| "loss": 0.3268, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 1596, | |
| "total_flos": 1.5287325828267258e+19, | |
| "train_loss": 0.5975370544538761, | |
| "train_runtime": 3758.5978, | |
| "train_samples_per_second": 14.848, | |
| "train_steps_per_second": 0.425 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1596, | |
| "num_train_epochs": 2, | |
| "save_steps": 1000, | |
| "total_flos": 1.5287325828267258e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |