paper-with-me

Papers

LLM-as-a-Verifier: A General-Purpose Verification Framework

2026-07-06 · Jacky Kwok, Shulu Li, Pranav Atreya, Yuejiang Liu, Yixing Jiang, Chelsea Finn, Marco Pavone, Ion Stoica, Azalia Mirhoseini arxiv

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.

📄 PDF Abstract BibTeX arXiv:2607.05391

Code (0)

등록된 구현이 없습니다.

Tasks

Mathematical Reasoning

Similar Papers 제목 키워드 기반

DelBugV: Delta-Debugging Neural Network Verifiers

2023-05-29 · Raya Elsaleh, Guy Katz

Deep neural networks (DNNs) are becoming a key component in diverse systems across the board. However, despite their success, they often err miserably; and this has triggered significant interest in formally verifying th…

OmniVerifier-M1: Multimodal Meta-Verifier with Explicit Structured Recalibration

2026-05-27 · Xinchen Zhang, Bowei Liu, Jiale Liu, Chufan Shi 외 arxiv

Visual outcomes are increasingly central to multimodal large language models, making reliable and fine-grained verification essential for scaling generalist foundation models. In this work, we investigate multimodal meta…

Reinforcement Learning

CompassVerifier: A Unified and Robust Verifier for LLMs Evaluation and Outcome Reward

2025-08-05 · Shudong Liu, Hongwei Liu, Junnan Liu, Linchen Xiao 외 arxiv

Answer verification is crucial not only for evaluating large language models (LLMs) by matching their unstructured outputs against standard answers, but also serves as the reward model to guide LLM optimization. Most eva…

Reinforcement Learning

Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignment

2026-02-12 · Jacky Kwok, Xilun Zhang, Mengdi Xu, Yuejiang Liu 외 arxiv

The long-standing vision of general-purpose robots hinges on their ability to understand and act upon natural language instructions. Vision-Language-Action (VLA) models have made remarkable progress toward this goal, yet…

Instruction Following

VIA-SD: Verification via Intra-Model Routing for Speculative Decoding

2026-06-10 · Yuchen Xian, Yang He, Yunqiu Xu, Yi Yang arxiv

Speculative decoding (SD) addresses the high inference costs of LLMs by having lightweight drafters generate candidates for large verifiers to validate in parallel. Existing draft-verify methods use binary decisions: acc…