Skip to content

Pull requests: mlfoundations/open_clip

New pull request New
32 Open 342 Closed
32 Open 342 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Experimenting with alternative siglip loss impl for better dist scaling
#971 opened Oct 26, 2024 by rwightman Loading…
1
Add GLIP (Centered Masking for Language-Image Pre-Training)
#964 opened Oct 19, 2024 by MingliangLiang3 Loading…
Enable passing output_hidden_states
#731 opened Nov 2, 2023 by thepowerfuldeez Loading…
3
Add gen reg tests
#689 opened Oct 22, 2023 by gpucce • Draft
9
[WIP] Add mammut
#641 opened Sep 19, 2023 by gpucce Loading…
CoCa: fix MultimodalTransformer init + Mask CLS token at end of seq
#551 opened Jun 26, 2023 by iejMac Loading…
10
Save intermediate checkpoints when sampling without replacement (take 2)
#535 opened May 17, 2023 by gabrielilharco Loading…
Implement Locking of Text Tower for CLIP Models
#523 opened May 9, 2023 by rsomani95 Loading…
8
[WIP] Allow for attention caching during CoCa generation
#502 opened Apr 20, 2023 by sramshetty • Draft
2 tasks
feat: add meta device initialization for pretrained models, 5x faster load times
#501 opened Apr 19, 2023 by ErwannMillon • Draft
5
Batched CoCa Generation
#498 opened Apr 17, 2023 by sramshetty Loading…
[WIP] Build open_clip as container and publish on push to main
#485 opened Apr 12, 2023 by Jeadie Loading…
4
CoCa: Condition captioning loss on the CLIP similarity
#469 opened Mar 17, 2023 by iejMac • Draft
Adding image decoder to CoCa
#467 opened Mar 15, 2023 by iejMac • Draft
4
Add optional flag in the text model config to return embeddings for all tokens in sequence instead of just the EOS embedding.
#456 opened Mar 2, 2023 by mkaic Loading…
Make coca and HF work together
#447 opened Feb 22, 2023 by gpucce • Draft
4 tasks
1
1
Experiment using autocast dtype in residual path of text transformer
#433 opened Feb 17, 2023 by rwightman • Draft
enable eval test
#431 opened Feb 15, 2023 by gpucce Loading…
Add video support
#430 opened Feb 15, 2023 by iejMac • Draft
11
vqa fine tune script
#424 opened Feb 13, 2023 by andrewwangva Loading…
2
Interpolate positional embeddings for input images with larger sizes
#416 opened Feb 11, 2023 by EeroHeikkinen Loading…
add audio spectrogram transformer, and full audio clip
#406 opened Feb 3, 2023 by lucidrains Loading…
41
[WIP] Support FSDP
#358 opened Jan 17, 2023 by mehdidc • Draft
23
Stop trying to guess the length of webdataset datasets
#339 opened Jan 5, 2023 by gabrielilharco Loading…
3
Support for initializing image tower with pretrained weights
#333 opened Dec 31, 2022 by Ja1Zhou Loading…
ProTip! Adding no:label will show everything without a label.