研究库 论文知识库

内容库 / 论文

arXiv:2609.05339 · RAG 检索增强

Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability

Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability

类型：arxiv
标识：2609.05339
链接：http://arxiv.org/abs/2609.05339v1
主分类：rag
形态：method
TLDR：Model upgrades are routine; memory migrations are not. An agent can keep the same memory store and still forget: a new model may interpret old notes differently, mixed embedding versions may break retrieval, and repair may fail without the original evidence. We compare memory as the same history is preserved verbatim for long-context reading (LC-RAW), divided into chunks for retrieval-augmented generation (RAG), compressed by a model into natural-language notes (NOTES), or normalized into a fixed-schema knowledge graph (KG-fixed). The study uses 48 synthetic histories with randomized answer co
副分类：agent
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-07-agent-rag-longcontext-candidates.json

相关 · 按语义匹配

仓库 djasdh/interest-memory

仓库 kolegadev/Katra-Agentic-Memory

笔记 flyP 轻量精读 · Beyond Memory: A Transactional Continuity Kernel for Long-Lived AI Agents（arXiv:2608.11632 · cs.MA · 8-12 v1）

仓库 HmZ9874/umd-planetary-memory

笔记知识库草稿 · Jay · 2026-08-01 19:50（傍晚高频）

笔记工程实践筛选 · Jay · 2026-07-23 19:50 (UTC+8)

arXiv

2609.05339

TLDR

Model upgrades are routine; memory migrations are not. An agent can keep the same memory store and still forget: a new model may interpret old notes differently, mixed embedding versions may break retrieval, and repair may fail without the original evidence. We compare memory as the same history is preserved verbatim for long-context reading (LC-RAW), divided into chunks for retrieval-augmented generation (RAG), compressed by a model into natural-language notes (NOTES), or normalized into a fixed-schema knowledge graph (KG-fixed). The study uses 48 synthetic histories with randomized answer co

arXiv ↗ 中文全文