# VeriEvol： 通过可验证进化指令扩展多模态数学推理

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-06-22 08:00
- AIHOT 分数：60
- AIHOT 链接：https://aihot.virxact.com/items/cmqsr0g2805iyslfuj3o0rtm0
- 原文链接：https://arxiv.org/abs/2606.23543

## AI 摘要

VeriEvol 是一个迭代框架，将多模态数学推理的奖励可靠性问题转化为可验证的数据构造问题。其类型感知进化模块将低难度图像-问题种子改写为更难的图像化提示；HTV-Agent 验证器在多项反证失败后才接受答案。在五个视觉数学基准上，将进化 SFT 数据从 10K 扩展至 250K 样本，平均准确率从 35.42 升至 54.73；固定 backbone、SFT 初始化和 GRPO 配方后，VeriEvol 相比未进化 RL 基线累积提升 +3.88，其中进化提示贡献 +1.82，验证器贡献 +2.06。项目开源全部提示、数据、模型、代码及验证轨迹。

## 正文

Scaling reinforcement learning for visual mathematical reasoning requires more than generating harder questions: as data volume grows, the reward labels themselves must remain reliable. Yet existing data pipelines scale supervision while trusting the labeller, and policy-side methods assume the underlying answers are already correct. We instead treat scaling as a verifiable data-construction problem and decouple two axes before any policy update: prompt difficulty, expanded by route-specific evolution operators, and answer reliability, enforced by offline hypothesis-test falsification. We instantiate this as VeriEvol, an iterative framework with two extensible components: a type-aware evolution module that rewrites low-difficulty image-question seeds into harder, image-grounded prompts; and HTV-Agent, a verifier that accepts an answer only after multi-source counter-evidence has failed to refute it. The resulting verified data scales in volume, extends by adding evolution routes or verifier channels, and plugs directly into existing GRPO-style RL recipes. On a five-benchmark visual-math suite, scaling evolved SFT data from 10K to 250K samples raises the mean accuracy from 35.42 to 54.73; then, with backbone, SFT initialization, and GRPO recipe held fixed, VeriEvol adds a cumulative +3.88 over an un-evolved RL baseline, of which +1.82 comes from evolved prompts and +2.06 from the HTV-Agent verifier. We release the prompts, data, models, code, and the full verifier trace of every sample, so that downstream work can scale and audit the pipeline rather than only inspect its outputs.