Skip to content
OpenTrain AIFor AI Companies

HFEPX · Eval paper review

ReCode: Reinforcing Code Generation with Reasoning-Process Rewards

Lishui Fan, Yu Zhang, Mouxiang Chen, Zhongxin Liu

Published

Aug 7, 2025

Citations

0

Trust level

Moderate

Usefulness score

50/100 (Medium)

Extraction confidence

55% (Moderate)

Derived from extracted protocol signals and abstract evidence.

Rater population

Not reported

Signals refreshed

May 5, 2026

Open arXiv Find implementation Shortlist (0) RSS feed

Tags

Coding Math Pairwise Preference

Should you rely on this paper?

This paper has useful evaluation signal, but protocol completeness is partial; pair it with related papers before deciding implementation strategy.

Use this for comparison and orientation, not as your only source.

Best use

Secondary protocol comparison source

Use if you need

Background context only.

What to verify

Validate the evaluation procedure and quality controls in the full paper before operational use.

Main weakness

The abstract does not clearly describe the evaluation setup.

Human feedback signal
Detected
From extracted signals
Evaluation signal
Weak or implicit
Validate from full paper
Usefulness for eval research
50/100
Moderate-confidence candidate

Useful as a secondary reference; validate protocol details against neighboring papers.

If you are doing eval pipeline work, start here

Human Eval Papers LLM-as-Judge Papers Pairwise Preference Papers Tool-Use Evaluation Search

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 outcomes, using execution correctness as a hard gate to mitigate reward hacking. Additionally, to assess the reward model's discriminative capability in assessing reasoning-process quality, we introduce LiveCodeBench-RewardBench (LCB-RB), a new benchmark comprising preference pairs of superior and inferior reasoning processes tailored for code generation. Experimental results across HumanEval(+), MBPP(+), LiveCodeBench, and BigCodeBench show that a 7B model trained with ReCode outperforms the base version by 16.1% and reaches performance comparable to GPT-4-Turbo. We further demonstrate the generalizability of ReCode by extending it to the math domain.

What we could verify

These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.

Human Feedback Types

strong

Pairwise Preference

Directly usable for protocol triage.

"In practice, rigorous reasoning is often a key driver of correct code, while Reinforcement Learning (RL) for code generation often neglects optimizing reasoning quality."

Evaluation Modes

missing

None explicit

Validate eval design from full paper text.

"In practice, rigorous reasoning is often a key driver of correct code, while Reinforcement Learning (RL) for code generation often neglects optimizing reasoning quality."

Quality Controls

missing

Not reported

No explicit QC controls found.

"In practice, rigorous reasoning is often a key driver of correct code, while Reinforcement Learning (RL) for code generation often neglects optimizing reasoning quality."

Benchmarks / Datasets

strong

LiveCodeBench, HumanEval+, MBPP+, Livecodebench Rewardbench, Bigcodebench

Useful for quick benchmark comparison.

"Additionally, to assess the reward model's discriminative capability in assessing reasoning-process quality, we introduce LiveCodeBench-RewardBench (LCB-RB), a new benchmark comprising preference pairs of superior and inferior reasoning processes tailored for code generation."

Reported Metrics

missing

Not extracted

No metric anchors detected.

"In practice, rigorous reasoning is often a key driver of correct code, while Reinforcement Learning (RL) for code generation often neglects optimizing reasoning quality."

Benchmarks and datasets

LiveCodeBenchHumanEval+MBPP+livecodebench-rewardbenchBigcodebench

Reported metrics

No metric terms were extracted from the available abstract.

Human feedback details
Uses human feedback
Yes
Feedback types
Pairwise Preference
Rater population
Not reported
Expertise required
Math, Coding
Evaluation details
Evaluation modes
None
Agentic eval
None
Quality controls
Not reported
Evidence quality
Moderate
Use this page as
Secondary protocol comparison source

Research brief

Metadata summary

In practice, rigorous reasoning is often a key driver of correct code, while Reinforcement Learning (RL) for code generation often neglects optimizing reasoning quality.

Based on abstract + metadata only. Check the source paper before making high-confidence protocol decisions.

Key takeaways

  • 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.

Researcher actions

  • Compare this paper against others mentioning LiveCodeBench and MATH.
  • Check the full text for explicit evaluation design choices (raters, protocol, and metrics).
  • Use related-paper links to find stronger protocol-specific references.

Caveats

  • Generated from abstract + metadata only; no PDF parsing.
  • Signals below are heuristic and may miss details reported outside the abstract.

Recommended queries

Pairwise preference evaluation LiveCodeBench MATH

Contribution summary

  • First, training reliable reward models to assess reasoning quality is bottlenecked by the scarcity of fine-grained preference data.
  • Additionally, to assess the reward model's discriminative capability in assessing reasoning-process quality, we introduce LiveCodeBench-RewardBench (LCB-RB), a new benchmark comprising preference pairs of superior and inferior reasoning…
  • Experimental results across HumanEval(+), MBPP(+), LiveCodeBench, and BigCodeBench show that a 7B model trained with ReCode outperforms the base version by 16.1% and reaches performance comparable to GPT-4-Turbo.

Why it matters for eval

  • Additionally, to assess the reward model's discriminative capability in assessing reasoning-process quality, we introduce LiveCodeBench-RewardBench (LCB-RB), a new benchmark comprising preference pairs of superior and inferior reasoning…
  • Experimental results across HumanEval(+), MBPP(+), LiveCodeBench, and BigCodeBench show that a 7B model trained with ReCode outperforms the base version by 16.1% and reaches performance comparable to GPT-4-Turbo.

Researcher checklist

  • Human feedback protocol is explicit

    Detected: Pairwise Preference

  • Evaluation mode is explicit

    No clear evaluation mode extracted.

  • Quality control reporting appears

    No calibration/adjudication/IAA control explicitly detected.

  • Benchmark or dataset anchors are present

    Detected: LiveCodeBench, HumanEval+, MBPP+, livecodebench-rewardbench

  • Metric reporting is present

    No metric terms extracted.