InfoAgent: Advancing Autonomous Information-Seeking Agents

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhang, Gongrui, Zhu, Jialiang, Yang, Ruiqi, Qiu, Kai, Zhang, Miaosen, Wu, Zhirong, Dai, Qi, Liu, Bei, Luo, Chong, Yang, Zhengyuan, Li, Linjie, Wang, Lijuan, Chen, Weizhu, Zhang, Yuan, Li, Xin, Liu, Zhaoyi, Geng, Xin, Guo, Baining
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916977169137664
author Zhang, Gongrui
Zhu, Jialiang
Yang, Ruiqi
Qiu, Kai
Zhang, Miaosen
Wu, Zhirong
Dai, Qi
Liu, Bei
Luo, Chong
Yang, Zhengyuan
Li, Linjie
Wang, Lijuan
Chen, Weizhu
Zhang, Yuan
Li, Xin
Liu, Zhaoyi
Geng, Xin
Guo, Baining
author_facet Zhang, Gongrui
Zhu, Jialiang
Yang, Ruiqi
Qiu, Kai
Zhang, Miaosen
Wu, Zhirong
Dai, Qi
Liu, Bei
Luo, Chong
Yang, Zhengyuan
Li, Linjie
Wang, Lijuan
Chen, Weizhu
Zhang, Yuan
Li, Xin
Liu, Zhaoyi
Geng, Xin
Guo, Baining
contents Building Large Language Model agents that expand their capabilities by interacting with external tools represents a new frontier in AI research and applications. In this paper, we introduce InfoAgent, a deep research agent powered by an innovative data synthesis pipeline and orchestrated web search tools. To construct challenging, hard-to-find queries,we build entity trees and apply sub-tree sampling with entity fuzzification to systematically increase question difficulty. Unlike prior work that relies heavily on commercial search tools, we develop a dedicated self-hosted search infrastructure, enhancing transparency of agent environments and facilitating further advancement of agent capacity. We evaluate the effectiveness of our data pipeline by measuring the average number of tool calls required to correctly answer a question, and also show that our agent yields better performance when equipped with our tools. Our \mbox{InfoAgent} is post-trained from Qwen3-14B using a two-stage recipe: cold-start supervised finetuning to instill long-horizon search behaviors, followed by reinforcement learning which significantly improves reasoning-driven tool use. With our methods, InfoAgent achieves 15.3\% accuracy on BrowseComp, 29.2\% on BrowseComp-ZH, and 40.4\% on Xbench-DS, outperforming prior open-source deep research agents such as WebSailor-72B and DeepDive-32B.
format Preprint
id arxiv_https___arxiv_org_abs_2509_25189
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle InfoAgent: Advancing Autonomous Information-Seeking Agents
Zhang, Gongrui
Zhu, Jialiang
Yang, Ruiqi
Qiu, Kai
Zhang, Miaosen
Wu, Zhirong
Dai, Qi
Liu, Bei
Luo, Chong
Yang, Zhengyuan
Li, Linjie
Wang, Lijuan
Chen, Weizhu
Zhang, Yuan
Li, Xin
Liu, Zhaoyi
Geng, Xin
Guo, Baining
Computation and Language
Artificial Intelligence
Building Large Language Model agents that expand their capabilities by interacting with external tools represents a new frontier in AI research and applications. In this paper, we introduce InfoAgent, a deep research agent powered by an innovative data synthesis pipeline and orchestrated web search tools. To construct challenging, hard-to-find queries,we build entity trees and apply sub-tree sampling with entity fuzzification to systematically increase question difficulty. Unlike prior work that relies heavily on commercial search tools, we develop a dedicated self-hosted search infrastructure, enhancing transparency of agent environments and facilitating further advancement of agent capacity. We evaluate the effectiveness of our data pipeline by measuring the average number of tool calls required to correctly answer a question, and also show that our agent yields better performance when equipped with our tools. Our \mbox{InfoAgent} is post-trained from Qwen3-14B using a two-stage recipe: cold-start supervised finetuning to instill long-horizon search behaviors, followed by reinforcement learning which significantly improves reasoning-driven tool use. With our methods, InfoAgent achieves 15.3\% accuracy on BrowseComp, 29.2\% on BrowseComp-ZH, and 40.4\% on Xbench-DS, outperforming prior open-source deep research agents such as WebSailor-72B and DeepDive-32B.
title InfoAgent: Advancing Autonomous Information-Seeking Agents
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2509.25189