ChristmasLee

Results 4 issues of ChristmasLee

b导好,我在下载预训练权值时给我报错如下,这是为什么呢? RuntimeError: invalid hash value (expected "b0", got "73f3a3d3c70508a1dfc1fcb58f8ba0edb1a5aaf2f0aaa2ce4dcd34b18b1a97df")

b导你好,我在使用Faster-RCNN训练时,分别使用vgg和resnet作为主干网络预训练权重,vgg是正常的,但是resnet在前50个冻结训练正常结束之后,loss变成nan,然后每10个epoch的验证集mAP也降成了0,这是为什么呢? ![image](https://user-images.githubusercontent.com/57053404/181866504-53173c68-9ca1-41fe-aa84-2c8912477946.png) ![image](https://user-images.githubusercontent.com/57053404/181866515-5f54337f-8ad9-43f1-8993-f87bc6fed76b.png)

b导,我在使用retinaNet训练时,使用主干网络作为预训练权重,然后pretrain设置为true,结果显示: Only tensors or tuples of tensors can be output from traced functions ...Error occurs, No graph saved 然后训练的每10代验证集mAP一直是0.00几,这是哪里出现问题如何解决呢?

b导,我在跑您的detr代码时,训练10个epoch没问题,但是在每十个epoch计算一次mAP的时候报错: File "/opt/data/private/detr/utils/utils_bbox.py", line 31, in forward boxes = boxes * scale_fct[:, None, :] RuntimeError: expected device cuda:0 and dtype Float but got device cuda:0 and dtype Long 这是怎么回事呢?