Cache-to-Cache: Direct Semantic Communication Between Large Language Models

Fu, Tianyu, Min, Zihan, Zhang, Hanling · arXiv · 2025