Salvato in:
Dettagli Bibliografici
Autori principali: Li, Wanli, Qu, Bince, Pan, Bo, Zhang, Jianyu, Liu, Zheng, Zhang, Pan, Chen, Wei, Zhang, Bo
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2604.17931
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917427061719040
author Li, Wanli
Qu, Bince
Pan, Bo
Zhang, Jianyu
Liu, Zheng
Zhang, Pan
Chen, Wei
Zhang, Bo
author_facet Li, Wanli
Qu, Bince
Pan, Bo
Zhang, Jianyu
Liu, Zheng
Zhang, Pan
Chen, Wei
Zhang, Bo
contents Reinforcement Learning (RL) has emerged as a powerful training paradigm for LLM-based agents. However, scaling agentic RL for deep research remains constrained by two coupled challenges: hand-crafted synthetic data fails to elicit genuine real-world search capabilities, and real-world search dependency during RL training introduces instability and prohibitive cost, which limits the scalability of Agentic RL. LiteResearcher is a training framework that makes Agentic RL scalable: by constructing a lite virtual world that mirrors real-world search dynamics, we enable a continuously improving training recipe that empowers a tiny search agent to outperform large-scale open-source and commercial models (e.g., Tongyi DeepResearch and Claude-4.5 Sonnet). Specifically, on common benchmarks such as GAIA and Xbench, our LiteResearcher-4B achieves open-source state-of-the-art results of 71.3% and 78.0% respectively, demonstrating that scalable RL training is a key enabler for Deep Research Agents.
format Preprint
id arxiv_https___arxiv_org_abs_2604_17931
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LiteResearcher: A Scalable Agentic RL Training Framework for Deep Research Agent
Li, Wanli
Qu, Bince
Pan, Bo
Zhang, Jianyu
Liu, Zheng
Zhang, Pan
Chen, Wei
Zhang, Bo
Artificial Intelligence
Reinforcement Learning (RL) has emerged as a powerful training paradigm for LLM-based agents. However, scaling agentic RL for deep research remains constrained by two coupled challenges: hand-crafted synthetic data fails to elicit genuine real-world search capabilities, and real-world search dependency during RL training introduces instability and prohibitive cost, which limits the scalability of Agentic RL. LiteResearcher is a training framework that makes Agentic RL scalable: by constructing a lite virtual world that mirrors real-world search dynamics, we enable a continuously improving training recipe that empowers a tiny search agent to outperform large-scale open-source and commercial models (e.g., Tongyi DeepResearch and Claude-4.5 Sonnet). Specifically, on common benchmarks such as GAIA and Xbench, our LiteResearcher-4B achieves open-source state-of-the-art results of 71.3% and 78.0% respectively, demonstrating that scalable RL training is a key enabler for Deep Research Agents.
title LiteResearcher: A Scalable Agentic RL Training Framework for Deep Research Agent
topic Artificial Intelligence
url https://arxiv.org/abs/2604.17931