MARCH：通过内容路由状态锚点扩展循环记忆

AI 摘要

MARCH 是一种新型网络架构，通过周期性缓存循环状态检查点作为状态锚点，并利用内容条件锚键进行注意力聚合，从而在保持计算效率的同时扩展状态空间模型的记忆容量。实验表明，MARCH 在常识推理、LongBench 和上下文检索任务上优于多种线性注意力变体，显著增强了循环模型的长距离记忆能力。

正文节选

MARCH \setheaderlogos \reportnumber MARCH: Scaling Recurrent Memory with Content-Routed State Anchors Abstract Transformers owe much of their strong long-context retrieval capability to a token-level memory that grows with context length. This flexibility, however, incurs a quadratic computation complexity during training and a key–value cache that grows linearly during autoregressive inference. Recurrent alternatives offer efficient decoding by compressing the entire history into a fixed-size s

发布时间：2026-08-14 12:00

抓取时间：2026-08-14 12:12

来源机构：arXiv