search

JKBox / YOLOv3-quadrangle

YOLOv3 detector with quadrangle in PyTorch
89 stars 29 forks source link

RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn #2

Closed ygean closed 5 years ago

ygean commented 5 years ago

It seems that torch version does not match, which version do you use? I used torch 0.4.1 and met this problem.

ygean commented 5 years ago

@JKBox When I print loss, I got tensor([0.], device='cuda:0')

JKBox commented 5 years ago

It seems that torch version does not match, which version do you use? I used torch 0.4.1 and met this problem.

I forget which torch version I used, you can try a higher version, which is downward compatible.

ygean commented 5 years ago

@JKBox Hey! I solved it by update torch version to 1.1.0, and I had trained it successfully on my dataset, the result looks good enough. Thank you very much and thanks for the code sharing, too!

vineeth357 commented 3 years ago

@JKBox Hey! I solved it by update the torch version to 1.1.0, and I had trained it successfully on my dataset, the result looks good enough. Thank you very much and thanks for the code sharing, too!

hi, can u help me by answering some questions??

  1. format of the dataset to use?? (ex: Coco, JSON etc)
  2. In which directory we have to place our dataset?
  3. format of annotations?
  4. any problem in training for u and suggestions for that?? thanks before :)