How does Masked loss work? #2897
codecomputation
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
It checked Masked loss in the UI and specify this additional parameters:
--conditioning_data_dir "C:\SD\masks"
The masks seems to have an effect on training. But, the content in the black areas are learned regardless once a certain number of steps are reached. Is it because the mask doesn't fully eliminate learning in the blacked out areas?
Beta Was this translation helpful? Give feedback.
All reactions