Why does LengthGroupSampler ensure that all samples in the same batch are from the same modality, either text or image? #1868
Unanswered
UnableToUseGit
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Intuitively, mixing image and text data in the same batch seems like a more reasonable approach.
Beta Was this translation helpful? Give feedback.
All reactions