Beyond Entropy: Correctness-Aware Advantage Shaping via Contrastive Policy Optimization

Hugging Face Daily Papers Papers

Summary

This paper introduces Contrastive Policy Optimization (CPO), which uses token-level contrastive disagreement between reference-guided and vanilla generation distributions for correctness-aware advantage shaping in reinforcement learning with verifiable rewards. CPO outperforms entropy-based RLVR methods on both in-domain and out-of-domain benchmarks.

Reinforcement learning with verifiable rewards (RLVR) commonly uses entropy for advantage shaping. However, entropy cannot distinguish useful uncertainty from detrimental confusion, limiting its effectiveness as a correctness signal. We propose Contrastive Policy Optimization (CPO), which uses token-level contrastive disagreement between reference-guided and vanilla generation distributions for correctness-aware advantage shaping. Both theoretical and empirical results show that this disagreement reliably indicates token-level correctness. We further show that On-policy Distillation is a special case of CPO, where the posterior distribution is instantiated by an external teacher model. CPO also resolves the zero-advantage problem. Experiments on in-domain and out-of-domain benchmarks demonstrate that CPO substantially outperforms entropy-based RLVR methods while maintaining strong generalization. Further analysis shows that correct and incorrect responses naturally support exploration and exploitation respectively, and balancing both leads to the best performance.
Original Article
View Cached Full Text

Cached at: 07/20/26, 09:39 AM

Paper page - Beyond Entropy: Correctness-Aware Advantage Shaping via Contrastive Policy Optimization

Source: https://huggingface.co/papers/2607.14614

Abstract

Reinforcementlearningwithverifiablerewards(RLVR)commonlyusesentropyforadvantageshaping.However,entropycannotdistinguishusefuluncertaintyfromdetrimentalconfusion,limitingitseffectivenessasacorrectnesssignal.WeproposeContrastivePolicyOptimization(CPO),whichusestoken-levelcontrastivedisagreementbetweenreference-guidedandvanillagenerationdistributionsforcorrectness-awareadvantageshaping.Boththeoreticalandempiricalresultsshowthatthisdisagreementreliablyindicatestoken-levelcorrectness.WefurthershowthatOn-policyDistillationisaspecialcaseofCPO,wheretheposteriordistributionisinstantiatedbyanexternalteachermodel.CPOalsoresolvesthezero-advantageproblem.Experimentsonin-domainandout-of-domainbenchmarksdemonstratethatCPOsubstantiallyoutperformsentropy-basedRLVRmethodswhilemaintainingstronggeneralization.Furtheranalysisshowsthatcorrectandincorrectresponsesnaturallysupportexplorationandexploitationrespectively,andbalancingbothleadstothebestperformance.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2607\.14614

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.14614 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.14614 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.14614 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Guidance Contrastive Token Credit Assignment for Discrete Policy Optimization

Hugging Face Daily Papers

This paper introduces Guidance Contrastive Policy Optimization (GCPO), a novel algorithm that enables per-token credit assignment in reinforcement learning by contrasting model predictions under positive and negative prompts, consistently outperforming GRPO and DAPO baselines on text-to-image generation and chain-of-thought reasoning benchmarks.

Group Entropy-Controlled Policy Optimization

Hugging Face Daily Papers

This paper proposes Group Entropy-Controlled Policy Optimization (GEPO), a lightweight extension to GRPO that uses group entropy to perform entropy-conditioned asymmetric advantage shaping, addressing heterogeneous entropy regimes across tasks during RL-based alignment of LLMs. Experiments show consistent improvements over GRPO and recent entropy-controlled methods across multiple benchmarks.