Efficient vector similarity search is critical for Retrieval-Augmented Generation (RAG) systems and other real-time AI applications. However, most existing methods are optimized for isolated queries and fail to leverage the continuity and correlation inherent in real-world query streams, such as those in multi-turn dialogues and multi-hop question answering. We formalize this problem as Continuous Vector Similarity Search (CVSS). While recent efforts attempt to reuse prior results, they either sacrifice accuracy through semantic caching or yield only marginal efficiency gains. To address this, we propose Reuse, an end-to-end framework that decomposes CVSS into two synergistic components: (1) Reuse Trigger that decides when to reuse prior search results, and (2) Reuse Searcher that addresses how to reuse them effectively. Together, these components significantly reduce redundant computation while maintaining near-identical recall. Extensive experiments on four real-world datasets show that Reuse achieves 1.6--3.0× higher throughput (QPS) than state-of-the-art methods at the same recall.
Zhuanglin Zheng, Yuxiang Zeng, Yunzhen Chi et al.· Proceedings of the 32nd ACM...· 0 citations
Recently, memory management has become a key infrastructure for LLM-based agents, as it directly affects long-horizon reasoning, personalized responses, and knowledge reuse. However, existing LLM memory systems typically adopt a coarse-grained (utility-agnostic) manner that treats heterogeneous user-LLM interaction records uniformly, leading to redundant and low-impact records persisting in the memory repository. To address this challenge, we present MemLens, a value-aware memory management system that takes memory records as first-class data objects. MemLens provides an end-to-end interactive analytics dashboard that exposes the complete memory lifecycle, including Shapley-style memory evaluation, value-aware storage, and memory-assisted response. Through a study-copilot application, the system enables users to inspect memory values, visualize hierarchical memory structures, and compare various memory management strategies in terms of response quality, retrieval latency, and token consumption. Therefore, our MemLens can serve as an efficient, interpretable, and personalized long-term memory management system for LLM-based agents.
Agent skills are reusable procedural artifacts that extend language agents with specialized workflows, tool conventions, and domain behaviors at inference time. However, creating reliable skills still depends largely on human authorship, model priors, or execution traces. These sources are often unavailable for unfamiliar tasks, suggesting the need to create skills from open-world materials. In this paper, we study open-world skill creation: given an underspecified skill brief and a source-access specification, a creator must discover behavior-relevant requirements omitted by the brief and determine how broadly each source-derived procedure is justified. We propose SkillAlchemy, an admission-centered framework for source-grounded skill creation. SkillAlchemy identifies implicit requirements through contrastive evidence, admits candidate procedures based on evidence-supported scope, and compiles the admitted content into a grammar-guided skill package. Extensive experiments across 87 SkillsBench v1.1 tasks demonstrate that SkillAlchemy improves pass rate over no-skill execution by 19.9 percentage points and the strongest automated baseline by 8.6 percentage points, while achieving performance comparable to human-curated skills.
Heng Wang, Shuyue Wei, Boyi Liu et al.· 0 citations
X-FED is proposed, a novel Conflict-Aware Cross-Client Federated Exit Distillation framework that jointly addresses both client- and depth-wise conflicts while extending PFL to early-exit networks and introduces a client-decoupled formulation that reduces communication overhead with theoretical soundness.
Boyi Liu, Zimu Zhou, Cheng Fang et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
