Swarat Chaudhuri
@swarat.bsky.social
Professor of Computer Science at UT Austin and Visiting Researcher at Google Deepmind, London. Automated Reasoning + Machine Learning + Formal Methods. https://www.cs.utexas.edu/~swarat
Congrats to @amitayush.bsky.social for leading this effort. And thanks to my student George Tsoukalas and collaborator extraordinaire @gregdnlp.bsky.social, who made critical contributions to the work. (3/3)
February 22, 2025 at 9:32 PM
Congrats to @amitayush.bsky.social for leading this effort. And thanks to my student George Tsoukalas and collaborator extraordinaire @gregdnlp.bsky.social, who made critical contributions to the work. (3/3)
It also has built-in machinery for large-scale, neurally guided proof search. We show that Proofwala's multilingual capabilities can enable transfer across proof assistants. Specifically, our multilingual model can outperform Coq- and Lean-only models at standard proof synthesis metrics. (2/3)
February 22, 2025 at 9:32 PM
It also has built-in machinery for large-scale, neurally guided proof search. We show that Proofwala's multilingual capabilities can enable transfer across proof assistants. Specifically, our multilingual model can outperform Coq- and Lean-only models at standard proof synthesis metrics. (2/3)
You make a good point. Alphaproof will evolve just as the informal approaches have, though.
December 24, 2024 at 8:56 AM
You make a good point. Alphaproof will evolve just as the informal approaches have, though.