HuggingFace Daily Papers（社区热门论文）· HuggingFace Daily Papers（社区热门论文）·· 2026-07-06精选AI 评分73
LLM-as-a-Verifier：一种通用验证框架
AI 导读
LLM-as-a-Verifier 是一种无需额外训练的通用验证框架，通过计算评分 token logits 分布的期望生成连续分数，实现细粒度反馈。该框架在 Terminal-Bench V2（86.5%）、SWE-Bench Verified（78.2%）、RoboRewardBench（87.4%）和 MedAgentBench（73.3%）上取得 SOTA 性能。其细粒度信号可用于 Claude Code 扩展，帮助开发者监控和改进智能体系统，也可为强化学习（如 SAC、GRPO）提供密集反馈，提升机器人学和数学推理基准的样本效率。
推荐理由
把验证当作新的缩放轴，这个思路很扎实，尤其直接给 Claude Code 搭了扩展。做 agent 系统的开发者现在可以试试把评估模块换成连续概率打分，也许比离散判断有效。
正文
Abstract:Scaling pre-training, post-training, and test-time compute have become the central paradigms for improving the capabilities of LLMs. In this work, we identify verification, the ability to determine the correctness of a solution, as a new scaling axis. To unlock this and demonstrate its effectiveness, we introduce LLM-as-a-Verifier, a general-purpose verification framework that provides fine-grained feedback for agentic tasks without requiring additional training. Unlike standard LM judges that prompt LLMs to produce discrete scores for candidate solutions, LLM-as-a-Verifier computes the expectation over the distribution of scoring token logits to generate continuous scores. This probabilistic formulation enables verification to scale along multiple dimensions: (1) score granularity, (2) repeated evaluation, and (3) criteria decomposition. In particular, we show that scaling the scoring granularity leads to better separation between positive and negative solutions, resulting in more calibrated comparisons. Moreover, scaling repeated evaluation and criteria decomposition consistently lead to additional gains in verification accuracy through variance and complexity reduction. We further introduce a cost-efficient ranking algorithm for selecting the best solution among candidates using the verifier's continuous scores. LLM-as-a-Verifier achieves state-of-the-art performance on Terminal-Bench V2 (86.5%), SWE-Bench Verified (78.2%), RoboRewardBench (87.4%), and MedAgentBench (73.3%). Beyond verification, the fine-grained signals from LLM-as-a-Verifier can also serve as a proxy for estimating task progress. We build an extension for Claude Code, enabling developers to monitor and improve their own agentic systems. Finally, we show that LLM-as-a-Verifier can provide dense feedback for RL, improving the sample efficiency of SAC and GRPO on robotics and mathematical reasoning benchmarks.
| Comments: | Code: this https URL Website: this https URL |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA); Robotics (cs.RO) |
| Cite as: | arXiv:2607.05391 [cs.AI] |
| (or arXiv:2607.05391v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.05391 arXiv-issued DOI via DataCite |
Submission history
From: Jacky Kwok [view email]
[v1]
Mon, 6 Jul 2026 17:59:35 UTC (4,216 KB)
[v2]
Tue, 7 Jul 2026 17:26:37 UTC (4,211 KB)
来源：HuggingFace Daily Papers（社区热门论文） · arxiv.org