🏢 Yale University

PHYSICS: A new benchmark reveals foundation models struggle with university-level physics, highlighting needs for improved reasoning and knowledge integration.

MCTS-RAG: Combines Monte Carlo Tree Search with Retrieval-Augmented Generation to enhance small LMs’ reasoning on complex tasks.

MEDAGENTSBENCH: a new benchmark for assessing complex medical reasoning in LLMs, revealing performance gaps and cost-effective strategies.

SynthLight: A novel diffusion model relights portraits realistically by learning to re-render synthetic faces, generalizing remarkably well to real photographs.