facebookresearch / facebookresearch/SlowFast

Finetune MViT-B 32x3 on Charades

Open
#556 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
7.4k
Forks
1.3k
PR merge metrics
No merged PRs in 30d

Description

Hi teams, appreciate your excellent work!
I'm trying to fine-tune the MViT-B 32x3 on Charades with this codebase.
However, I got a very low mAP result compared with the reported one, although I followed the setting you mentioned in supplementary D.3. And I'm also confused about the statement "A per-class sigmoid output is used to account for the multi-class nature.". I believe there must be some configs/codes missing in my training.
Will it be convenient for your group to share the fine-tune config or related code with me (shoubin@cs.unc.edu)? It will be greatly helpful for our undergoing research. Thanks in advance!

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.