search

google-research / big_vision

Official codebase used to develop Vision Transformer, SigLIP, MLP-Mixer, LiT and more.
Apache License 2.0
2.16k stars 147 forks source link

Question: Will SigLIP / SigLiT be added to this codebase? #38

Closed linzhiqiu closed 10 months ago

linzhiqiu commented 1 year ago

I mean both model checkpoints and training/inference code

lucasb-eyer commented 11 months ago

Hi, yes, we are working on the code release (though there's a chance we'll only finish it slightly after ICCV). Meanwhile, we just released some SigLIP models and colab in #47 and @rwightman has (independently) reimplemented SigLIP in PyTorch OpenCLIP here: https://github.com/mlfoundations/open_clip/pull/634

inkzk commented 10 months ago

Hi, yes, we are working on the code release (though there's a chance we'll only finish it slightly after ICCV). Meanwhile, we just released some SigLIP models and colab in #47 and @rwightman has (independently) reimplemented SigLIP in PyTorch OpenCLIP here: mlfoundations/open_clip#634

@lucasb-eyer hey there, offerring some pressure hhha. Whether the code will be released soon since ICCV ended for two weeks already

lucasb-eyer commented 10 months ago

Since you're pressuring me: The comment that you quoted contains the PR, which was created and merged three weeks ago, i.e. before ICCV started. I just forgot to close this issue I guess. Closing :)

(Or is there something else you need?)

lucasb-eyer commented 10 months ago

Oops, I was too bold, that was only the colabs (i.e. inference code). The training code was added in #51 which has also been there for three weeks (i.e. before ICCV) but only merged last week.

inkzk commented 10 months ago

@lucasb-eyer My bad, great appreciating!~