arXiv cs.AI· Ali Afoud, Jie JW Wu·· 3 天前AI 评分34
将验证作为 LLM 智能体的架构层：V-Model 设计及确定性核心的试点研究
Verification as an Architectural Layer for LLM Agents: A V-Model Design, and a Pilot Study of Its Deterministic Core
AI 导读
研究提出将验证作为 LLM 智能体的架构层，借鉴软件工程 V-Model，让规范层级逐级对应专用验证器，由确定性控制器执行判定，仅验证结果写入记忆。在 MuSiQue 四跳任务、8B 参数骨干上对比五种配置：两个无验证配置十题全未答对，均因步数上限或 token 限制终止；带验证但无规划器的配置答对八题，其余弃答。确定性 gate 以零边际成本贡献了九次纠正中的八次，且验证存在时规划反而降低表现。
正文 · 原文
Abstract:Large language model (LLM) agents built on the ReAct pattern concentrate four responsibilities in one model: selecting a strategy, choosing each action, formatting it, and judging whether the result is adequate. Nothing outside the generative loop can reject its output, so an agent that cannot make progress does not report failure; it runs until an external budget stops it. We propose treating verification as an architectural layer by adapting the V-model from software engineering: specification levels descend from requirements to individual steps, each level is paired with a dedicated verifier, a deterministic controller enforces every verdict, and only verification outcomes write to memory, so a rejection localizes the level that introduced the fault and an agent halts by declining rather than by exhaustion. Each verifier separates a zero-cost deterministic \emph{gate} from an optional LLM \emph{judge}, so the contribution and cost of each can be measured independently. We report a pilot implementing the acceptance- and unit-level verifier pairs, comparing five configurations that share one executor, tool set, and scorer and differ only in verification, on the four-hop stratum of MuSiQue with an 8B-parameter backbone. Across 47 executions, the two unverified configurations answered none of ten questions, every run ending at a step cap or provider token limit; the verified configuration without a planner answered eight and abstained on the rest. Deterministic gates produced eight of the nine observed corrections at zero marginal cost, and planning degraded performance once verification was present. These results characterize termination behavior, not accuracy at scale; we outline a twelve-month plan to complete and evaluate the full architecture, including the integration-level pair the pilot omits.
| Subjects: | Software Engineering (cs.SE); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.31937 [cs.SE] |
| (or arXiv:2609.31937v2 [cs.SE] for this version) | |
| https://doi.org/10.48550/arXiv.2609.31937 arXiv-issued DOI via DataCite |
Submission history
From: Ali Afoud [view email]
[v1]
Fri, 25 Sep 2026 19:30:58 UTC (26 KB)
[v2]
Sat, 10 Oct 2026 04:25:59 UTC (26 KB)
来源：arXiv cs.AI · arxiv.org