ai.hackcv
论文精选 81arXiv

Top-$k$ Retrieval Leaves Memory Value Undetermined· Top-k检索无法确定记忆价值

A memory-augmented agent must decide which memories to keep, merge and retrieve, and its only evidence of their value is its retrieval log: the memories it retrieved in each episode and whether it succeeded. Existing methods assign every memory a value from this log, even when the log cannot separate one memory's contribution from another's. We treat the log as an experiment that nobody designed and ask which memory values it determines. Two counts answer this before any outcome is observed: how often each memory was retrieved, and how often each pair was retrieved together. From these counts, we prove that top-$k$ retrieval leaves every memory value undetermined, however many episodes are logged, and that a question with its own memories, asked once, determines at most one of their values

领域:cs.AI作者:Arman Behnam
相关推荐
论文精选 65已解读

Strategist: Evolving Reasoning with Agents· 演进推理策略的AI元代理

A model is only half of an AI agent system. The other half is the inference strategy: the …

领域:cs.AI作者:Taiming Lu、Qi Xu、Zhigang Hua
📎 arXiv🕒 11-03 02:51🔗 arxiv.org
论文精选 65已解读

DiabAgent: Toward Agentic Physiological Control· DiabAgent:迈向代理生理控制

As pretraining scales, medical language models have advanced rapidly in clinical question …

领域:cs.AI作者:Chengxi Li、Shaoxuan Li、Deyao Hong
📎 arXiv🕒 11-03 02:41🔗 arxiv.org

本站内容由 LLM 精选聚合,原文版权归 arXiv 所有 · 摘录仅供参考