We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
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
Hi @krasserm!
Fabulous the work you have presented. Very good results for scale 4 bicubic, also easy to implement
But when I run the model for scale 8, the datasets load perfectly but in generator pre-training an error happens that I don't understand the reason
Apparently the size of the images are different but I do not understand the reason since the datasets load perfectly
You have tried an 8 scale bicubic and it has not given you problems ???
Your help would be appreciated!
Thank so much!
The text was updated successfully, but these errors were encountered:
SRGAN model and training is hard-coded for scale 4.
Sorry, something went wrong.
No branches or pull requests
Hi @krasserm!
Fabulous the work you have presented. Very good results for scale 4 bicubic, also easy to implement
But when I run the model for scale 8, the datasets load perfectly but in generator pre-training an error happens that I don't understand the reason
Apparently the size of the images are different but I do not understand the reason since the datasets load perfectly
You have tried an 8 scale bicubic and it has not given you problems ???
Your help would be appreciated!
Thank so much!
The text was updated successfully, but these errors were encountered: