Recode: Reinforcing Code Generation With Reasoning-process Rewards

Abstract

arXiv:2508.05170v3 Announce Type: replace-cross Abstract: In practice, rigorous reasoning is often a key driver of correct code, while Reinforcement Learning (RL) for code generation often neglects optimizing reasoning quality. Bringing process-level supervision into RL is appealing, but it faces two challenges. First, training reliable reward models to assess reasoning quality is bottlenecked by the scarcity of fine-grained preference data. Second, naively incorporating such neural rewards may suffer from reward hacking. This work proposes ReCode (Reasoning-Reinforced Code Generation), a novel RL training framework comprising: (1) Contrastive Reasoning-Process Reward Learning (CRPL), which trains a reward model with synthesized optimized and degraded reasoning variants to assess the quality of reasoning process; and (2) Consistency-Gated GRPO (CG-GRPO), which integrates the reasoning-process reward model into RL by gating neural reasoning-process rewards with strict execution outcome

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keyfan2026recode

Related papers

• The Implicit Curriculum: Learning Dynamics In RL With Verifiable Rewards (2026)0.00
• Stepwise Guided Policy Optimization: Coloring Your Incorrect Reasoning In GRPO (2025)0.00
• Policy Improvement Reinforcement Learning (2026)0.00
• Scheduling Your LLM Reinforcement Learning With Reasoning Trees (2026)0.00
• Portool: Importance-aware Policy Optimization With Rewarded Tree For Multi-tool-integrated Reasoning (2026)0.00
• Adapt To Thrive! Adaptive Power-mean Policy Optimization For Improved LLM Reasoning (2026)0.00
• Rewarding The Scientific Process: Process-level Reward Modeling For Agentic Data Analysis (2026)5.07
• Rl-star: Theoretical Analysis Of Reinforcement Learning Frameworks For Self-taught Reasoner (2024)0.00