今日速览
| 序号 | 标题 | 来源 | 日期 | 主题 | 推荐等级 |
|---|---|---|---|---|---|
| 1 | Evidence Blindness in Direct Corpus Interaction: Persistent Navigation with AtlasNav | arXiv | 2026-08-25 | RAG | 高 |
| 2 | SkillForge: Evolving Verifiable Skills for Reinforcement Learning Agents | arXiv | 2026-08-25 | AI-Agent | 高 |
重点论文与技术动态
1. Evidence Blindness in Direct Corpus Interaction: Persistent Navigation with AtlasNav
- 来源：arXiv
- 日期：2026-08-25
- 作者/机构：Hongyu Guo, Zhiyu Zheng, Zhao Cao
- 主题标签：
RAG,arXiv - 推荐等级：高
- 分类：cs.AI
一句话结论
AtlasNav通过持久化语料库结构有效解决直接语料库交互中的证据盲点问题，显著提升证据获取效率与计算性能。
一段话：在有限交互预算下，传统方法常因证据无法浮现、未打开或关键片段未暴露导致性能下降；AtlasNav构建Corpus Atlas实现自适应导航，避免重复构建结构，使证据获取更早、更高效，同时降低推理成本。
核心内容
- 证据盲点（Evidence Blindness）指证据在交互中无法有效获取、打开或暴露关键片段，被量化为阶段式证据实现。
- AtlasNav将语料库一次性组织为Corpus Atlas，支持各查询自适应导航，避免动态重建共享结构。
- 在BrowseComp-Plus上实现92.05%严格准确率，降低30.21%在线推理成本，且在PhantomWiki和企业知识系统中有效迁移。
方法与数据
- 方法：AtlasNav框架，持久化多视图语料库导航。
- 数据：在BrowseComp-Plus和PhantomWiki数据集上验证，支持10K-1M规模缩放。
价值判断
- 值得关注：证据盲点对agentic search效率的核心制约机制。
- 可复用点：Corpus Atlas表示原则可迁移至不同语料库组织及企业知识系统。
- 局限/待核查：摘要未明确提及局限。
摘要
Large language model agents are moving beyond conventional retrieval-augmented generation toward direct interaction with external corpora. Direct Corpus Interaction (DCI) keeps the full corpus accessible, yet reachable evidence can remain unusable under finite interaction budgets. Required evidence may fail to surface, a surfaced supporting document may remain unopened, or an opened document may fail to expose its decisive fragment. We call this progressive silent loss Evidence Blindness and quantify it through stage-wise evidence realization. Within the DCI paradigm, raw interaction adds little reusable corpus organization, while dynamic-workspace methods reconstruct a query-conditioned interaction space from each query and trajectory. In both cases, useful structure is recovered largely online. We instead formulate large-scale agentic search as finite-budget navigation over reusable corpus structure. We introduce AtlasNav, a persistent multi-view corpus-navigation framework that retains direct corpus interaction but organizes the corpus once into a Corpus Atlas, allowing each query to navigate adaptively rather than reconstruct shared structure. On BrowseComp-Plus, AtlasNav achieves 92.05% strict accuracy while reducing recorded online inference cost by 30.21% relative to the prior dynamic-workspace state of the art. Under matched budgets, it realizes the complete required evidence earlier and approaches the same model's evidence-supplied empirical reference more rapidly. The same representation principle remains effective under PhantomWiki's distinct corpus organization and controlled 10K-1M scaling, and transfers competitively to heterogeneous enterprise knowledge. These results show that agentic search depends not only on accessible evidence, but also on how the corpus is represented so that limited interaction becomes effective navigation.2. SkillForge: Evolving Verifiable Skills for Reinforcement Learning Agents
- 来源：arXiv
- 日期：2026-08-25
- 作者/机构：Shidong Yang, Ziyu Ma, Tongwen Huang, Xucong Wang, Renda Li, Yiming Hu
- 主题标签：
AI-Agent,arXiv - 推荐等级：高
- 分类：cs.CL
一句话结论
SkillForge通过环境交互验证和精炼技能，实现LLM代理的连续技能进化。
核心内容
- 解决技能库仅追加且未验证的问题，使技能通过环境交互被验证和精炼。
- 显式技能使用使RL直接优化环境动作和技能调用决策。
- 引入证据基技能验证和多路径技能归纳，确保技能库持续增长并保持高质量。
方法与数据
- 方法包括证据基技能验证和多路径技能归纳。
- 实验在ALFWorld、WebShop和AppWorld数据集上进行。
价值判断
- 值得关注：在ALFWorld、WebShop和AppWorld等复杂决策任务上，SkillForge持续优于SkillRL，证明连续验证技能对训练更强LLM代理的有效性。
- 可复用点：技能验证与精炼机制可被其他强化学习框架复用，以提升知识累积能力。
- 局限/待核查：摘要未明确提及局限。