1
MemTrace: Tracing and Attributing Errors in Large Language Model Memory Systems
新论文提出MemTrace，追踪大模型内存中的错误归因，揭示信息随时间如何被合成或破坏，为长程推理可靠性提供诊断方法。
arXiv:2605.28732v1 Announce Type: cross Abstract: Memory is essential for enabling large language models to support long-horizon reasoning, yet existi…