Language Models Can See Better: Visual Contrastive Decoding For LLM Multimodal Reasoning.

Yuqi Pang, Bowen Yang, Haoqin Tu, Yun Cao, Zeyu Zhang

Browse the full ICASSP paper archive.