Hassaan

[Peter Tong @TongPetersb]

Train Beyond Language. We bet on the visual world as the critical next step alongside and beyond language modeling. So, we studied building foundation models from scratch with vision.
We share our exploration: visual representations, data, world modeling, architecture, and scaling behavior! [1/9]

[Hassaan @urboioi]

@szq0214 Congratulations sir

[Zhiqiang Shen @szq0214]

After two strong rating ICML submissions were unexpectedly rejected, we had a nice rebound with two ACL acceptances (out of three submissions) and one KDD acceptance (one for one)—a clear case of “when one door closes, another opens” 😀. One ACL main conference paper was completed by three undergraduates and two of my master students. In it, we propose a novel RAG framework tailored for SLMs: Distilling RAG! The paper will be released soon, stay tuned!