发生了什么
This Paper 发表 Self-Pruned Key-Value Attention(2605.14037)
为什么值得关注
Quantifies memory reduction (10%-33.7%) and speedup (2.1-4.6x) with specific methodology for KV pruning, offering actionable insights for efficient LLM deployment.
信息来源
以下内容来自公开来源，可打开原文继续核验。
01
Self-Pruned Key-Value Attention: Learning
This paper introduces a method for LLMs to save memory by pruning key-value cache using a predictor that scores token usefulness. The model keeps recent tokens and older tokens only when their score exceeds a threshold. Trained via next-token prediction, it achieves 10%-33.7% KV entry retention while matching performance and delivering 2.1-4.6x decoding speedup in long-context scenarios.
打开原始来源 ↗