🏢 Hong Kong Polytechnic University

Aligned LLMs’ safety often anchors in the template region, creating vulnerabilities. Detaching safety mechanisms shows promise in mitigation.

Multimodal LLMs can now evaluate art aesthetics with human-level accuracy using a novel dataset (MM-StyleBench) and prompt method (ArtCoT), significantly improving AI alignment in artistic evaluation.

MolReFlect achieves state-of-the-art molecule-text alignment by using a teacher-student LLM framework that generates fine-grained alignments, improving accuracy and explainability.