openai/CLIP
1
0
Fork 0
mirror of https://github.com/openai/CLIP.git synced 2026-11-05 01:37:26 +00:00
Code Issues 274 Projects Packages Wiki Activity

[GH-ISSUE #281] Any plan to add Swin transformer? #244

New issue
Open
opened 2026-10-31 21:12:53 +00:00 by bsakaguchi · 2 comments
bsakaguchi commented 2026-10-31 21:12:53 +00:00
Owner
Copy link

Originally created by @klae01 on GitHub (Aug 14, 2022).
Original GitHub issue: https://github.com/openai/CLIP/issues/281

Swin transformer achieves higher accuracy in model size and computational amount similar to ViT.
I think that using clip's method and dataset will show higher performance.

  • ViT-B/16, 384x384, 86M, 55.4Gflops, 77.9 (imagenet 1k acc)
  • Swin-B, 384x384, 88M, 47.0Gflops, 84.5 (imagenet 1k acc)

reference : Swin Transformer table 1 (a)

It would be great if a Swin transformer could be added to compare performance.

Although Resnet is still powerful, I would like to compare whether the performance that was too poor compared to ViT showed that transformer showed overwhelming performance compared to convolution in vision field. Therefore, it would be nice to add famous convolution-based networks such as Efficientnet and Convnext.

Originally created by @klae01 on GitHub (Aug 14, 2022). Original GitHub issue: https://github.com/openai/CLIP/issues/281 `Swin transformer` achieves higher accuracy in model size and computational amount similar to `ViT`. I think that using clip's method and dataset will show higher performance. - ViT-B/16, 384x384, 86M, 55.4Gflops, 77.9 (imagenet 1k acc) - Swin-B, 384x384, 88M, 47.0Gflops, 84.5 (imagenet 1k acc) reference : [Swin Transformer](https://arxiv.org/pdf/2103.14030.pdf) table 1 (a) It would be great if a `Swin transformer` could be added to compare performance. Although `Resnet` is still powerful, I would like to compare whether the performance that was too poor compared to ViT showed that transformer showed overwhelming performance compared to convolution in vision field. Therefore, it would be nice to add famous convolution-based networks such as `Efficientnet` and `Convnext`.
bsakaguchi commented 2026-10-31 21:12:54 +00:00
Author
Owner
Copy link

@celestialxevermore commented on GitHub (Sep 28, 2022):

It is excactly what I've expected too

<!-- gh-comment-id:1260512673 --> @celestialxevermore commented on GitHub (Sep 28, 2022): It is excactly what I've expected too
bsakaguchi commented 2026-10-31 21:12:54 +00:00
Author
Owner
Copy link

@MaAo commented on GitHub (Jan 13, 2023):

I've expected , too

<!-- gh-comment-id:1382076669 --> @MaAo commented on GitHub (Jan 13, 2023): I've expected , too
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
codex/pin-github-workflow-refs-20260325-132838
jongwook/issue-396
jongwook-patch-1
No results found.
Labels
Clear labels   
pull-request
Mirrored from GitHub Pull Request

No labels
pull-request
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
openai/CLIP#244
No description provided.