MemTrace: Tracing and Attributing Errors in Large Language Model Memory Systems 事件

MemTrace: Tracing and Attributing Errors in Large Language Model Memory Systems arXiv:2605.28732v1 Announce Type: new Abstract: Memory is essential for enabling large language models to support long-horizon reasoning, yet existing memory systems remain unreliable and difficult to debug. Tracing memory's dynamic evolution is crucial to understand how information is synthesized, propagated, or corrupted over time. In this work, we study the new problem of error tracing and attribution in LLM memo