Picture for Wen-Zhuo Liu

Wen-Zhuo Liu

Recoverable Compression: A Multimodal Vision Token Recovery Mechanism Guided by Text Information

Add code
Sep 02, 2024
Viaarxiv icon