2 papers across 2 sessions
revealing This paper analyses the free energy foundations of inverse folding models for protein stability prediction, revealing limitations in current practice and proposing simple improvements for better zero-shot prediction.