LV-XAttn: Distributed Cross-Attention for Long Visual Inputs in Multimodal Large Language Models

Devs

LV-XAttn: Distributed Cross-Attention for Long Visual Inputs in Multimodal Large Language Models | Read Paper on Bytez