Enregistré dans:
Détails bibliographiques
Auteurs principaux: Zhang, Junshuo, Huang, Chengrui, Guo, Feng, Li, Zihan, Shi, Ke, Jiang, Menghua, Yu, Jiguo, Shang, Shuo, Gao, Shen
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2604.24320
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866910169363906560
author Zhang, Junshuo
Huang, Chengrui
Guo, Feng
Li, Zihan
Shi, Ke
Jiang, Menghua
Yu, Jiguo
Shang, Shuo
Gao, Shen
author_facet Zhang, Junshuo
Huang, Chengrui
Guo, Feng
Li, Zihan
Shi, Ke
Jiang, Menghua
Yu, Jiguo
Shang, Shuo
Gao, Shen
contents Large language model (LLM) agents that follow the sequential "reason-then-act" paradigm have achieved superior performance in many complex tasks.However, these methods suffer from limited exploration and incomplete environmental understanding, as they interact with only a single environment per step. In this paper, we first introduce a novel paradigm that enables an agent to interact with multiple environments simultaneously and share cross-trajectory experiences. Building upon this paradigm, we further propose DPEPO, a reinforcement learning (RL) algorithm that encourages the agent to perform diverse parallel exploration. There are two stages in DPEPO: initial supervised fine-tuning (SFT) imparts basic parallel reasoning and action generation, followed by reinforcement learning stage with a hierarchical reward scheme. We design a parallel trajectory-level success reward and two step-level rewards: Diverse Action Reward and Diverse State Transition Reward, which actively penalize behavioral redundancy and promote broad exploration. Extensive experiments on ALFWorld and ScienceWorld show that DPEPO achieves state-of-the-art (SOTA) success rates, while maintaining comparable efficiency to strong sequential baselines. (Code is available at https://github.com/LePanda026/Code-for-DPEPO)
format Preprint
id arxiv_https___arxiv_org_abs_2604_24320
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle DPEPO: Diverse Parallel Exploration Policy Optimization for LLM-based Agents
Zhang, Junshuo
Huang, Chengrui
Guo, Feng
Li, Zihan
Shi, Ke
Jiang, Menghua
Yu, Jiguo
Shang, Shuo
Gao, Shen
Computation and Language
Large language model (LLM) agents that follow the sequential "reason-then-act" paradigm have achieved superior performance in many complex tasks.However, these methods suffer from limited exploration and incomplete environmental understanding, as they interact with only a single environment per step. In this paper, we first introduce a novel paradigm that enables an agent to interact with multiple environments simultaneously and share cross-trajectory experiences. Building upon this paradigm, we further propose DPEPO, a reinforcement learning (RL) algorithm that encourages the agent to perform diverse parallel exploration. There are two stages in DPEPO: initial supervised fine-tuning (SFT) imparts basic parallel reasoning and action generation, followed by reinforcement learning stage with a hierarchical reward scheme. We design a parallel trajectory-level success reward and two step-level rewards: Diverse Action Reward and Diverse State Transition Reward, which actively penalize behavioral redundancy and promote broad exploration. Extensive experiments on ALFWorld and ScienceWorld show that DPEPO achieves state-of-the-art (SOTA) success rates, while maintaining comparable efficiency to strong sequential baselines. (Code is available at https://github.com/LePanda026/Code-for-DPEPO)
title DPEPO: Diverse Parallel Exploration Policy Optimization for LLM-based Agents
topic Computation and Language
url https://arxiv.org/abs/2604.24320