analiza aiqbench
This paper addresses critical exposure bias issues in On-Policy Self-Distillation for Large Language Models by introducing a contrastive learning framework to preserve fine-grained optimization signals. The proposed method, Contrastive Reinforced Policy Optimization (CRPO), demonstrates improved training stability and generalization across complex reasoning benchmarks without relying on external reward models.
Contrastive Reinforced Policy Optimization via Privileged Self-Distillation
arXiv:2607.28026v1 Announce Type: new Abstract: Recent advances in post-training Large Language Models (LLMs) increasingly rely on Reinforcement Learning with Verifiable Rewards (RLVR) or On-Policy Self-Distillation (OPSD). While OPSD provides dense, logit-level supervision, it inherently suffers from exposure bias due to the privileged information of the self-teacher. In multi-turn agentic settings, this leads to reasoning route convergence and the loss of clear optimization directions. To tackle these challenges, we introduce Contrastive Reinforced Policy Optimization (CRPO), which reformulates agentic OPSD from a contrastive learning perspective. By leveraging predictive entropy to distinguish between positive positions (reflective exploration) and negative positions (exposure bias), CRPO conducts group-wise contrast to preserve reliable, fine-grained optimization signals. Extensive evaluations across 13 challenging reasoning and deep-search benchmarks demonstrate that CRPO consistently outperforms existing reinforcement learning and self-distillation baselines, significantly enhancing training stability and generalization in long-horizon interactions.
Abstract:Recent advances in post-training Large Language Models (LLMs) increasingly rely on Reinforcement Learning with Verifiable Rewards (RLVR) or On-Policy Self-Distillation (OPSD). While OPSD provides dense, logit-level supervision, it inherently suffers from exposure bias due to the privileged information of the self-teacher. In multi-turn agentic settings, this leads to reasoning route convergence and the loss of clear optimization directions. To tackle these challenges, we introduce Contrastive Reinforced Policy Optimization (CRPO), which reformulates agentic OPSD from a contrastive learning perspective. By leveraging predictive entropy to distinguish between positive positions (reflective exploration) and negative positions (exposure bias), CRPO conducts group-wise contrast to preserve reliable, fine-grained optimization signals. Extensive evaluations across 13 challenging reasoning and deep-search benchmarks demonstrate that CRPO consistently outperforms existing reinforcement learning and self-distillation baselines, significantly enhancing training stability and generalization in long-horizon interactions.
Machine Learning (cs.LG)
arXiv:2607.28026 [cs.LG]
(or arXiv:2607.28026v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2607.28026 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 227 cuvinte
Subiecte conexe
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · 17 sept. 2026
SIPO: Unifying Reinforcement Learning with On-Policy Self-Distillation
arXiv · 1 oct. 2026
RLCSD: Reinforcement Learning with Contrastive On-Policy Self-Distillation
arXiv · 12 iun. 2026
$\beta$-OPSD: Deriving with Policy Optimization, Training with Self-Distillation
arXiv · 31 iul. 2026