---
id: 20260501-T0-03
title: "大模型推理新策略：投票vs重写优化方案"
title_en: "Voting vs Rewriting: New LRM Scaling Strategy"
url: https://ai.daily.yangsir.net/daily/20260501-T0-03
issue_date: 2026-05-01
publish_date: 2026-04-30T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2604.26644
---

# 大模型推理新策略：投票vs重写优化方案

研究提出大模型测试时扩展新策略：通过模型内部投票决定重写方案。在数学推理任务中，该方法将准确率从73%提升至89%。实验证明能有效处理复杂推理场景，减少计算资源浪费。

## English Version

**Voting vs Rewriting: New LRM Scaling Strategy**

Research proposes new LRM scaling strategy: internal voting decides rewrite paths. Improves math reasoning accuracy from 73% to 89%. Experiments show it handles complex scenarios effectively, reducing computational waste.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2604.26644)

**详情页**：https://ai.daily.yangsir.net/daily/20260501-T0-03

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*