Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. ItΒ collects links to all the places you might be looking at while hunting down a tough bug.
And, if youβre still stuck at the end, weβre happy to hop on a call to see how we can help out.
training question about learn rate , map=0
See original GitHub issueDuring training, it was found that the learning step size changed quickly. Within one epoch, it changed from 6.15E-08 to 2.00E-04, and the corresponding total loss was reduced from 14.5 to 10. The subsequent learning gradually increased until the 65th epoch. , And total loss hovering around 9, the ap of each category is about 0
keys β values β ββββββββββββββββββββͺβββββββββββββββββββββββββββββ‘ β seed β None β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β output_dir β './YOLOX_outputs' β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β print_interval β 1 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β eval_interval β 1 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β num_classes β 3 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β depth β 0.33 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β width β 0.5 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β data_num_workers β 4 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β input_size β (640, 640) β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β multiscale_range β 5 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β data_dir β None β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β train_ann β 'instances_train2017.json' β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β val_ann β 'instances_val2017.json' β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β mosaic_prob β 1.0 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β mixup_prob β 1.0 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β hsv_prob β 1.0 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β flip_prob β 0.5 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β degrees β 10.0 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β translate β 0.1 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β mosaic_scale β (0.1, 2) β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β mixup_scale β (0.5, 1.5) β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β shear β 2.0 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β perspective β 0.0 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β enable_mixup β True β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β warmup_epochs β 5 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β max_epoch β 80 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β warmup_lr β 0 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β basic_lr_per_img β 0.00015625 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β scheduler β 'yoloxwarmcos' β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β no_aug_epochs β 15 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β min_lr_ratio β 0.05 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β ema β True β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β weight_decay β 0.0005 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β momentum β 0.9 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β exp_name β 'yolox_voc_s' β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β test_size β (640, 640) β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β test_conf β 0.01 β ββββββββββββββββββββΌβββββββββββββββββββββββββββββ€ β nmsthre β 0.65
2021-10-11 14:38:09 | INFO | yolox.core.trainer:188 - ---> start train epoch1 2021-10-11 14:38:12 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 1/57, mem: 6736Mb, iter_time: 2.817s, data_time: 0.001s, total_loss: 14.5, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 8.9, cls_loss: 1.7, lr: 6.156e-08, size: 640, ETA: 3:34:03 2021-10-11 14:38:12 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 2/57, mem: 6736Mb, iter_time: 0.306s, data_time: 0.012s, total_loss: 14.9, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 9.2, cls_loss: 1.8, lr: 2.462e-07, size: 640, ETA: 1:58:38 2021-10-11 14:38:12 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 3/57, mem: 6736Mb, iter_time: 0.226s, data_time: 0.001s, total_loss: 13.7, iou_loss: 4.0, l1_loss: 0.0, conf_loss: 8.1, cls_loss: 1.6, lr: 5.540e-07, size: 640, ETA: 1:24:47 2021-10-11 14:38:12 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 4/57, mem: 6736Mb, iter_time: 0.199s, data_time: 0.001s, total_loss: 13.5, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 7.8, cls_loss: 1.9, lr: 9.849e-07, size: 640, ETA: 1:07:21 2021-10-11 14:38:13 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 5/57, mem: 6736Mb, iter_time: 0.212s, data_time: 0.001s, total_loss: 14.0, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 8.4, cls_loss: 1.7, lr: 1.539e-06, size: 640, ETA: 0:57:05 2021-10-11 14:38:13 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 6/57, mem: 6736Mb, iter_time: 0.211s, data_time: 0.001s, total_loss: 13.3, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 7.7, cls_loss: 1.7, lr: 2.216e-06, size: 640, ETA: 0:50:13 2021-10-11 14:38:13 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 7/57, mem: 6736Mb, iter_time: 0.210s, data_time: 0.001s, total_loss: 14.0, iou_loss: 4.0, l1_loss: 0.0, conf_loss: 8.3, cls_loss: 1.7, lr: 3.016e-06, size: 640, ETA: 0:45:19 2021-10-11 14:38:13 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 8/57, mem: 6736Mb, iter_time: 0.477s, data_time: 0.001s, total_loss: 14.4, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 8.9, cls_loss: 1.6, lr: 3.940e-06, size: 640, ETA: 0:44:10 2021-10-11 14:38:14 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 9/57, mem: 6736Mb, iter_time: 0.357s, data_time: 0.001s, total_loss: 14.3, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 8.7, cls_loss: 1.9, lr: 4.986e-06, size: 640, ETA: 0:42:15 2021-10-11 14:38:14 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 10/57, mem: 6736Mb, iter_time: 0.195s, data_time: 0.001s, total_loss: 15.6, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 9.9, cls_loss: 2.0, lr: 6.156e-06, size: 640, ETA: 0:39:30 2021-10-11 14:38:16 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 11/57, mem: 7069Mb, iter_time: 1.865s, data_time: 0.232s, total_loss: 14.3, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 8.6, cls_loss: 1.7, lr: 7.448e-06, size: 736, ETA: 0:48:45 2021-10-11 14:38:16 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 12/57, mem: 7069Mb, iter_time: 0.253s, data_time: 0.001s, total_loss: 14.5, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 8.7, cls_loss: 1.8, lr: 8.864e-06, size: 736, ETA: 0:46:17 2021-10-11 14:38:16 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 13/57, mem: 7069Mb, iter_time: 0.265s, data_time: 0.001s, total_loss: 14.4, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 8.7, cls_loss: 1.8, lr: 1.040e-05, size: 736, ETA: 0:44:15 2021-10-11 14:38:17 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 14/57, mem: 7069Mb, iter_time: 0.261s, data_time: 0.001s, total_loss: 13.8, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 8.1, cls_loss: 2.0, lr: 1.207e-05, size: 736, ETA: 0:42:30 2021-10-11 14:38:17 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 15/57, mem: 7069Mb, iter_time: 0.258s, data_time: 0.001s, total_loss: 14.0, iou_loss: 4.0, l1_loss: 0.0, conf_loss: 8.3, cls_loss: 1.7, lr: 1.385e-05, size: 736, ETA: 0:40:57 2021-10-11 14:38:17 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 16/57, mem: 7069Mb, iter_time: 0.321s, data_time: 0.001s, total_loss: 14.3, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 8.6, cls_loss: 1.9, lr: 1.576e-05, size: 736, ETA: 0:39:54 2021-10-11 14:38:17 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 17/57, mem: 7069Mb, iter_time: 0.258s, data_time: 0.001s, total_loss: 13.6, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 8.1, cls_loss: 1.8, lr: 1.779e-05, size: 736, ETA: 0:38:42 2021-10-11 14:38:18 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 18/57, mem: 7069Mb, iter_time: 0.248s, data_time: 0.002s, total_loss: 13.3, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 7.7, cls_loss: 1.8, lr: 1.994e-05, size: 736, ETA: 0:37:35 2021-10-11 14:38:18 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 19/57, mem: 7069Mb, iter_time: 0.264s, data_time: 0.002s, total_loss: 13.9, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 8.2, cls_loss: 1.8, lr: 2.222e-05, size: 736, ETA: 0:36:38 2021-10-11 14:38:19 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 20/57, mem: 7069Mb, iter_time: 0.806s, data_time: 0.001s, total_loss: 13.8, iou_loss: 4.0, l1_loss: 0.0, conf_loss: 8.2, cls_loss: 1.6, lr: 2.462e-05, size: 736, ETA: 0:37:51 2021-10-11 14:38:20 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 21/57, mem: 7069Mb, iter_time: 1.373s, data_time: 0.010s, total_loss: 13.6, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 7.9, cls_loss: 1.9, lr: 2.715e-05, size: 672, ETA: 0:40:59 2021-10-11 14:38:20 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 22/57, mem: 7069Mb, iter_time: 0.291s, data_time: 0.001s, total_loss: 13.0, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 7.5, cls_loss: 1.7, lr: 2.979e-05, size: 672, ETA: 0:40:07 2021-10-11 14:38:21 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 23/57, mem: 7069Mb, iter_time: 0.242s, data_time: 0.001s, total_loss: 13.4, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 7.8, cls_loss: 1.9, lr: 3.256e-05, size: 672, ETA: 0:39:09 2021-10-11 14:38:21 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 24/57, mem: 7069Mb, iter_time: 0.241s, data_time: 0.001s, total_loss: 13.5, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 7.9, cls_loss: 1.8, lr: 3.546e-05, size: 672, ETA: 0:38:17 2021-10-11 14:38:21 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 25/57, mem: 7069Mb, iter_time: 0.254s, data_time: 0.001s, total_loss: 13.4, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 7.9, cls_loss: 1.7, lr: 3.847e-05, size: 672, ETA: 0:37:30 2021-10-11 14:38:21 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 26/57, mem: 7069Mb, iter_time: 0.233s, data_time: 0.002s, total_loss: 13.0, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 7.5, cls_loss: 1.6, lr: 4.161e-05, size: 672, ETA: 0:36:44 2021-10-11 14:38:22 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 27/57, mem: 7069Mb, iter_time: 0.254s, data_time: 0.001s, total_loss: 12.6, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 7.1, cls_loss: 1.9, lr: 4.488e-05, size: 672, ETA: 0:36:04 2021-10-11 14:38:22 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 28/57, mem: 7069Mb, iter_time: 0.243s, data_time: 0.001s, total_loss: 12.5, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 7.0, cls_loss: 1.6, lr: 4.826e-05, size: 672, ETA: 0:35:26 2021-10-11 14:38:22 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 29/57, mem: 7069Mb, iter_time: 0.251s, data_time: 0.001s, total_loss: 12.2, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 6.7, cls_loss: 1.7, lr: 5.177e-05, size: 672, ETA: 0:34:51 2021-10-11 14:38:22 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 30/57, mem: 7069Mb, iter_time: 0.233s, data_time: 0.001s, total_loss: 12.5, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 7.0, cls_loss: 1.8, lr: 5.540e-05, size: 672, ETA: 0:34:16 2021-10-11 14:38:24 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 31/57, mem: 7069Mb, iter_time: 1.456s, data_time: 0.002s, total_loss: 12.6, iou_loss: 4.0, l1_loss: 0.0, conf_loss: 7.0, cls_loss: 1.5, lr: 5.916e-05, size: 704, ETA: 0:36:42 2021-10-11 14:38:24 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 32/57, mem: 7069Mb, iter_time: 0.250s, data_time: 0.002s, total_loss: 12.2, iou_loss: 4.0, l1_loss: 0.0, conf_loss: 6.8, cls_loss: 1.5, lr: 6.303e-05, size: 704, ETA: 0:36:08 2021-10-11 14:38:24 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 33/57, mem: 7069Mb, iter_time: 0.284s, data_time: 0.001s, total_loss: 11.7, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 6.3, cls_loss: 1.7, lr: 6.704e-05, size: 704, ETA: 0:35:41 2021-10-11 14:38:25 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 34/57, mem: 7069Mb, iter_time: 0.254s, data_time: 0.001s, total_loss: 11.6, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 6.2, cls_loss: 1.5, lr: 7.116e-05, size: 704, ETA: 0:35:11 2021-10-11 14:38:25 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 35/57, mem: 7069Mb, iter_time: 0.254s, data_time: 0.001s, total_loss: 11.9, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 6.5, cls_loss: 1.8, lr: 7.541e-05, size: 704, ETA: 0:34:44 2021-10-11 14:38:25 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 36/57, mem: 7069Mb, iter_time: 0.250s, data_time: 0.001s, total_loss: 11.8, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 6.4, cls_loss: 1.5, lr: 7.978e-05, size: 704, ETA: 0:34:17 2021-10-11 14:38:26 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 37/57, mem: 7069Mb, iter_time: 0.549s, data_time: 0.002s, total_loss: 11.7, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 6.4, cls_loss: 1.7, lr: 8.427e-05, size: 704, ETA: 0:34:28 2021-10-11 14:38:26 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 38/57, mem: 7069Mb, iter_time: 0.257s, data_time: 0.001s, total_loss: 11.8, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 6.5, cls_loss: 1.6, lr: 8.889e-05, size: 704, ETA: 0:34:03 2021-10-11 14:38:27 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 39/57, mem: 7069Mb, iter_time: 0.486s, data_time: 0.001s, total_loss: 11.6, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 6.3, cls_loss: 1.6, lr: 9.363e-05, size: 704, ETA: 0:34:07 2021-10-11 14:38:27 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 40/57, mem: 7069Mb, iter_time: 0.245s, data_time: 0.001s, total_loss: 11.2, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 6.0, cls_loss: 1.6, lr: 9.849e-05, size: 704, ETA: 0:33:43 2021-10-11 14:38:27 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 41/57, mem: 7069Mb, iter_time: 0.694s, data_time: 0.001s, total_loss: 11.5, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 6.3, cls_loss: 1.6, lr: 1.035e-04, size: 704, ETA: 0:34:10 2021-10-11 14:38:28 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 42/57, mem: 7069Mb, iter_time: 0.253s, data_time: 0.002s, total_loss: 10.3, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 5.3, cls_loss: 1.5, lr: 1.086e-04, size: 704, ETA: 0:33:48 2021-10-11 14:38:29 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 43/57, mem: 7069Mb, iter_time: 0.880s, data_time: 0.002s, total_loss: 10.9, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 5.6, cls_loss: 1.6, lr: 1.138e-04, size: 704, ETA: 0:34:32 2021-10-11 14:38:29 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 44/57, mem: 7069Mb, iter_time: 0.252s, data_time: 0.002s, total_loss: 11.1, iou_loss: 3.5, l1_loss: 0.0, conf_loss: 5.9, cls_loss: 1.6, lr: 1.192e-04, size: 704, ETA: 0:34:11 2021-10-11 14:38:29 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 45/57, mem: 7069Mb, iter_time: 0.372s, data_time: 0.003s, total_loss: 11.3, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 6.2, cls_loss: 1.3, lr: 1.247e-04, size: 704, ETA: 0:34:02 2021-10-11 14:38:30 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 46/57, mem: 7069Mb, iter_time: 0.305s, data_time: 0.001s, total_loss: 11.3, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 6.1, cls_loss: 1.5, lr: 1.303e-04, size: 704, ETA: 0:33:47 2021-10-11 14:38:30 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 47/57, mem: 7069Mb, iter_time: 0.883s, data_time: 0.002s, total_loss: 10.5, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 5.4, cls_loss: 1.4, lr: 1.360e-04, size: 704, ETA: 0:34:28 2021-10-11 14:38:31 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 48/57, mem: 7069Mb, iter_time: 0.246s, data_time: 0.004s, total_loss: 11.2, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 6.0, cls_loss: 1.5, lr: 1.418e-04, size: 704, ETA: 0:34:08 2021-10-11 14:38:31 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 49/57, mem: 7069Mb, iter_time: 0.410s, data_time: 0.001s, total_loss: 10.8, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 5.7, cls_loss: 1.3, lr: 1.478e-04, size: 704, ETA: 0:34:03 2021-10-11 14:38:31 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 50/57, mem: 7069Mb, iter_time: 0.249s, data_time: 0.006s, total_loss: 10.6, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 5.6, cls_loss: 1.4, lr: 1.539e-04, size: 704, ETA: 0:33:45 2021-10-11 14:38:33 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 51/57, mem: 7069Mb, iter_time: 1.222s, data_time: 0.001s, total_loss: 10.6, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 5.5, cls_loss: 1.5, lr: 1.601e-04, size: 704, ETA: 0:34:52 2021-10-11 14:38:33 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 52/57, mem: 7069Mb, iter_time: 0.273s, data_time: 0.001s, total_loss: 11.0, iou_loss: 3.8, l1_loss: 0.0, conf_loss: 5.8, cls_loss: 1.3, lr: 1.665e-04, size: 704, ETA: 0:34:35 2021-10-11 14:38:33 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 53/57, mem: 7069Mb, iter_time: 0.245s, data_time: 0.001s, total_loss: 10.6, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 5.5, cls_loss: 1.3, lr: 1.729e-04, size: 704, ETA: 0:34:17 2021-10-11 14:38:33 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 54/57, mem: 7069Mb, iter_time: 0.239s, data_time: 0.001s, total_loss: 10.4, iou_loss: 3.9, l1_loss: 0.0, conf_loss: 5.3, cls_loss: 1.2, lr: 1.795e-04, size: 704, ETA: 0:33:58 2021-10-11 14:38:34 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 55/57, mem: 7069Mb, iter_time: 1.115s, data_time: 0.001s, total_loss: 10.2, iou_loss: 3.5, l1_loss: 0.0, conf_loss: 5.2, cls_loss: 1.5, lr: 1.862e-04, size: 704, ETA: 0:34:52 2021-10-11 14:38:35 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 56/57, mem: 7069Mb, iter_time: 0.246s, data_time: 0.002s, total_loss: 10.3, iou_loss: 3.7, l1_loss: 0.0, conf_loss: 5.2, cls_loss: 1.4, lr: 1.930e-04, size: 704, ETA: 0:34:34 2021-10-11 14:38:35 | INFO | yolox.core.trainer:246 - epoch: 1/80, iter: 57/57, mem: 7069Mb, iter_time: 0.256s, data_time: 0.002s, total_loss: 10.4, iou_loss: 3.6, l1_loss: 0.0, conf_loss: 5.5, cls_loss: 1.3, lr: 2.000e-04, size: 704, ETA: 0:34:
Issue Analytics
- State:
- Created 2 years ago
- Comments:7 (1 by maintainers)
Top Results From Across the Web
Top Related Medium Post
Top Related StackOverflow Question
Troubleshoot Live Code
Top Related Reddit Thread
Top Related Hackernoon Post
Top Related Tweet
Top Related Dev.to Post
Top Related Hashnode Post
I think I already know what the problem is. The image size in the xml file is inconsistent with the corresponding image.
There is almost no learning in the first 100 epochs. Although the total loss drops from 14 to about 9, the mp is almost 0. After training to 300 epochs, the total loss is still about 9, but the mp is about 0.1, but the mp is unstable.οΌNeed to train more than 200 epochs, ap reaches about 0.1, the whole process is very slow, epoch=300, ap is still about 0.1, because my training set data is very small, only more than 1000 pictures, but the same data The mAP on yolov4 can reach 0.6. I want to know why the yolox-s is so low, do I need to adjust the parameters?