---
title: "Policy Split: Dual-Mode Exploration"
created: 2026-04-15
updated: 2026-04-15
type: concept
tags: [rl, exploration, dual-mode, policy-optimization]
sources: [raw/papers/2026/04/2604.11510.md]
---
# Policy Split: Dual-Mode Exploration in LLM RL

## 核心
LLM RL 训练中引入双模式探索策略，分离 exploit 和 explore 的优化路径。

## Related
- [[grpo-rl-training]]
- [[ttvs]]