facebookresearch / facebookresearch/moco
Question about queue dimension
Open
- Dominant language
- No language data
- Stars
- 5.1k
- Forks
- 802
- PR merge metrics
- No merged PRs in 30d
Description
https://github.com/facebookresearch/moco/blob/5a429c00bb6d4efdf511bf31b6f01e064bf929ab/moco/builder.py#L50
Hello. I have a simple question about queue dimension in moco/builder.py
I noticed that `K` is a length of queue, and `dim` is a feature dimension.
Isn't it faster accessing each feature with row major than accessing with column major?
Can anybody explain me why queue designed with dim [dim(feature), queue_length]?
Contributor guide
Assessment
This issue has not been assessed yet.