Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Details of training base feature extractor (src/utils/train_image.py) #33

Open
Qingzhang1994 opened this issue Mar 13, 2023 · 0 comments

Comments

@Qingzhang1994
Copy link

When I trained the code, I found that the function required by the following code were missing. Could you please provide the code to help me study?
weights = torch.DoubleTensor(utils.class_weights(tr_clss)) in line 116
acc = utils.accuracy(op.data, cl.data, topk=(1,)) in line 276.

I found some functions to calculate accuracy, but after training an epoch it appears that the training set has 99% accuracy and the verification set has 0% accuracy.
image

# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant