Open
Description
Why do we need negative sequences?
I think contrastive learning do not need negative sequences, because negative pairs can define as two different frames in a video.
Metadata
Metadata
Assignees
Labels
No labels
{{ message }}
No one assigned
No labels
No projects
No milestone
None yet
No branches or pull requests