AI RESEARCH
ReDiPrune: Relevance-Diversity Pre-Projection Token Pruning for Efficient Multimodal LLMs
arXiv CS.CV
•
ArXi:2603.24680v1 Announce Type: new Recent multimodal large language models are computationally expensive because Transformers must process a large number of visual tokens. We present \textbf{ReDiPrune}, a