mirror of
https://github.com/openai/CLIP.git
synced 2026-11-05 01:37:26 +00:00
[GH-ISSUE #5] Non-jit version of model? #5
Labels
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
openai/CLIP#5
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @jmhessel on GitHub (Jan 7, 2021).
Original GitHub issue: https://github.com/openai/CLIP/issues/5
Hi!
Thanks for the code release, and for the great work!! I was wondering if it might be possible to release a non-jit version of the model. This might make it easier to convert to other platforms, e.g., tensorflow. What do you think?
Jack
@jongwook commented on GitHub (Jan 7, 2021):
Hello, thanks for your interest! We wanted to keep the release volume minimal and decided to release the JIT model only.
That being said, I guess it'll be relatively straightforward to recover the Python code from the JIT model via
model.code,model.visual.code, etc. (I wonder if there's an automated tool for this)I also expect there will be implementations like lucidrains' or others that will support importing the CLIP model weights.
@jmhessel commented on GitHub (Jan 7, 2021):
Gotcha, that makes sense! It's definitely possible to get a non-JIT version using
model.code, just a bit tedious :) Thanks!@jongwook commented on GitHub (Jan 12, 2021):
Hello! As written in #9 as well, we are preparing to release the model code this week, which will allow users to inspect and modify the architecture more conveniently.
@jongwook commented on GitHub (Jan 12, 2021):
The model code is now available in model.py
SimpleTokenizerfor SAM3 text tokenizer convenience #518