Fetching the paper…

Efficient Vision-Language Models by Summarizing Visual Tokens into Compact Registers · Around