Difficulty-Estimated Policy Optimization

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Zhao, Yu, Jiang, Fan, Liu, Tianle, Zeng, Bo, Liu, Yu, Wang, Longyue, Luo, Weihua
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917254134759424
author Zhao, Yu
Jiang, Fan
Liu, Tianle
Zeng, Bo
Liu, Yu
Wang, Longyue
Luo, Weihua
author_facet Zhao, Yu
Jiang, Fan
Liu, Tianle
Zeng, Bo
Liu, Yu
Wang, Longyue
Luo, Weihua
contents Recent advancements in Large Reasoning Models (LRMs), exemplified by DeepSeek-R1, have underscored the potential of scaling inference-time compute through Group Relative Policy Optimization (GRPO). However, GRPO frequently suffers from gradient signal attenuation when encountering problems that are either too trivial or overly complex. In these scenarios, the disappearance of inter-group advantages makes the gradient signal susceptible to noise, thereby jeopardizing convergence stability. While variants like DAPO attempt to rectify gradient vanishing, they do not alleviate the substantial computational overhead incurred by exhaustive rollouts on low-utility samples. In this paper, we propose Difficulty-Estimated Policy Optimization (DEPO), a novel framework designed to optimize the efficiency and robustness of reasoning alignment. DEPO integrates an online Difficulty Estimator that dynamically assesses and filters training data before the rollout phase. This mechanism ensures that computational resources are prioritized for samples with high learning potential. Empirical results demonstrate that DEPO achieves up to a 2x reduction in rollout costs without compromising model performance. Our approach significantly lowers the computational barrier for training high-performance reasoning models, offering a more sustainable path for reasoning scaling. Code and data will be released upon acceptance.
format Preprint
id arxiv_https___arxiv_org_abs_2602_06375
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Difficulty-Estimated Policy Optimization
Zhao, Yu
Jiang, Fan
Liu, Tianle
Zeng, Bo
Liu, Yu
Wang, Longyue
Luo, Weihua
Artificial Intelligence
Recent advancements in Large Reasoning Models (LRMs), exemplified by DeepSeek-R1, have underscored the potential of scaling inference-time compute through Group Relative Policy Optimization (GRPO). However, GRPO frequently suffers from gradient signal attenuation when encountering problems that are either too trivial or overly complex. In these scenarios, the disappearance of inter-group advantages makes the gradient signal susceptible to noise, thereby jeopardizing convergence stability. While variants like DAPO attempt to rectify gradient vanishing, they do not alleviate the substantial computational overhead incurred by exhaustive rollouts on low-utility samples. In this paper, we propose Difficulty-Estimated Policy Optimization (DEPO), a novel framework designed to optimize the efficiency and robustness of reasoning alignment. DEPO integrates an online Difficulty Estimator that dynamically assesses and filters training data before the rollout phase. This mechanism ensures that computational resources are prioritized for samples with high learning potential. Empirical results demonstrate that DEPO achieves up to a 2x reduction in rollout costs without compromising model performance. Our approach significantly lowers the computational barrier for training high-performance reasoning models, offering a more sustainable path for reasoning scaling. Code and data will be released upon acceptance.
title Difficulty-Estimated Policy Optimization
topic Artificial Intelligence
url https://arxiv.org/abs/2602.06375