Evidence-Augmented Policy Optimization with Reward Co-Evolution for Long-Context Reasoning

arXiv:2601.10306v2 Announce Type: replace-cross Abstract: While Reinforcement Learning (RL) has advanced LLM reasoning, applying it to long-context scenarios is hindered by sparsity of outcome rewards. This limitation fails to penalize ungrounded "lucky guesses," leaving the critical process of needle-in-a-haystack evidence retrieval largely unsupervised. To address this, we propose EAPO (Evidence-Augmented Policy Optimization). We first establish the Evidence-Augmented Reasoning paradigm, validating via Tree-Structured Evidence Sampling that precise evidence extraction is the decisive bottleneck for long-context reasoning. Guided by this insight, EAPO introduces a specialized RL algorithm where a reward model computes a Group-Relative Evidence Reward, providing dense process supervision to explicitly improve evidence quality. To sustain accurate supervision throughout training, we further incorporate an Adaptive Reward-Policy Co-Evolution mechanism. This mechanism iteratively refines the reward model using outcome-consistent rollouts, sharpening its discriminative capability to ensure precise process guidance. Comprehensive evaluations across eight benchmarks demonstrate that EAPO significantly enhances long-context reasoning performance compared to SOTA baselines.

Related

• Less Noise, More Voice: Reinforcement Learning for Reasoning via Instruction Purification
• EEPO: Exploration-Enhanced Policy Optimization via Sample-Then-Forget
• IG-Search: Step-Level Information Gain Rewards for Search-Augmented Reasoning
• OPSDL: On-Policy Self-Distillation for Long-Context Language Models
• Graph-Based Chain-of-Thought Pruning for Reducing Redundant Reflections in Reasoning LLMs

Source: arXiv cs.CL | 2026-04-21