ByteDance-Seed / ByteDance-Seed/Depth-Anything-3

Misleading statement about L_s : L_g

Open
#48 0 comments 2 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
6.3k
Forks
702
PR merge metrics
No merged PRs in 30d

Description

In the paper it says L_s : L_g should be 2:1
Image

Image

However in the released code alternative attention starts at 1/3 which indicates L_s : L_g should be 1:2.
Maybe in the paper what you want to say is that num_of_frame_attention : num_of_global_attention = 2 : 1?

Contributor guide

No contributing guide indexed for this repository

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.