-
Notifications
You must be signed in to change notification settings - Fork 4
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Pretrain Models issues. #1
Comments
@Xuguowei-hub beat me to the punch on this. I dug a little deeper and the trained data contains the dict trans, not net as indicated in the code. Switch to use trans instead introduces other errors. As indicated above clip_best_fid.pth does not exist but clip_best.pth does.
|
Thank you very much for pointing out the issues. Regarding the first error with the file name, we have made the corresponding correction. For the second issue, we may not have clarified in the usage instructions. The weights loaded here should be those of the VQ-VAE, and for this weight, we have used the same weight as t2m-gpt without any adjustments. We have also uploaded the weights to the link |
Thank you very much for your reply! Will you provide additional tutorials on demo visualization in the future? |
Thank you very much for your contribution to the community.
python eval_t2m.py --resume-pth pretrained/net_best_fid.pth --clip_path pretrained/clip_best_fid.pth
2.I'm getting a KeyError, the checkpoint file net_best_fid.pth doesn't seem to contain the key ['net'] that the code is expecting.
print('loading checkpoint from {}'.format(args.resume_pth)) ckpt = torch.load(args.resume_pth, map_location='cpu') net.load_state_dict(ckpt['net'], strict=True) net.eval() net.cuda()
Therefore, it raises a KeyError.
Looking forward to your reply!
The text was updated successfully, but these errors were encountered: