EVAF：一种选择性参数整合的测试-重测协议
EVAF: A Test-Retest Protocol for Selective Parametric Consolidation
AI总结：
提出EVAF机制和测试-重测协议，通过门控LoRA整合高价值、高意外经验，在GPT-2和TinyLlama上实现比基线更强的抗干扰行为持久性，同时保持低参数漂移和低跨角色污染。
摘要：
长时间运行的语言代理需要机制来决定哪些经验在工作上下文消失后应持续存在。检索系统可以重新插入过去的文本，但它们本身并不能表明经验已被选择性地整合到模型自身的行为中。我们引入了EVAF（回声-效价吸引子场）机制，用于门控LoRA整合，以及一种测试-重测协议，用于在受控干扰下测量选择性参数整合。在GPT-2和TinyLlama上，EVAF优先整合高效价、高意外的经验，同时通过互补的路由记忆路径保留可检索的事实记忆。测试-重测测量显示，与冻结、仅检索和无门控持续更新基线相比，EVAF在干扰后具有更强的行为持久性，同时保持低参数漂移和低跨角色污染。结果支持记忆访问与记忆深度之间的分离：检索事实和内化经验是不同的计算操作。
英文摘要：
Long-running language agents need mechanisms for deciding which experiences should persist after the working context is gone. Retrieval systems can reinsert past text, but they do not by themselves show that an experience has been selectively consolidated into the model's own behavior. We introduce EVAF, an Echo-Valence Attractor Field mechanism for gated LoRA consolidation, and a test-retest protocol for measuring selective parametric consolidation under controlled interference. Across GPT-2 and TinyLlama, EVAF preferentially consolidates high-valence, high-surprise experiences while preserving retrieval-accessible factual memory through a complementary routed memory path. Test-retest measurements show stronger post-interference behavioral persistence than frozen, retrieval-only, and ungated continual-update baselines, while keeping parameter drift and cross-persona contamination low. The results support a separation between memory access and memory depth: retrieving a fact and internalizing an experience are distinct computational operations.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
