Fetching the paper…

PuMer: Pruning and Merging Tokens for Efficient Vision Language Models · Around