salesforce/LAVIS
View on GitHub ↗ ★ 11,256 Jupyter Notebook archived
LAVIS - A One-stop Library for Language-Vision Intelligence
5 mentions on Hacker News by 5 people: 3 stories, 2 comments · first 2022-09-27 · latest 2023-05-06
20082026
-
…to what BLIP2 is a good start https://huggingface.co/docs/transformers/main/en/model_doc/b... https://github.com/salesforce/LAVIS
comment by sitkack · 2023-05-06 · in MaMMUT: A simple vision-encoder text-decoder architecture for multimodal tasks -
It's possible to take a text only model and ground it with images. examples are blip-2( https://github.com/salesforce/LAVIS/tree/main/projects/blip2 ) fromage( https://github.com/kohjingyu/fromage ) prismer(…
comment by og_kalu · 2023-04-16 · in Sam Altman: OpenAI is not training GPT-5 and "won't for some time"