Salesforce AI Research — Random Attention: KV Cache Eviction for Efficient Reasoning
中文摘要
Salesforce AI 研究推出 Random Attention，通过 KV 缓存剔除机制提升推理效率。
English Summary
Salesforce AI Research introduces Random Attention, utilizing KV cache eviction to enhance reasoning efficiency.
Original Excerpt
1. 한눈에 보는 핵심 요약 (Executive Summary) Continue reading on Medium »