About Me

I am a first-year Master’s student at the Gaoling School of Artificial Intelligence (GSAI), Renmin University of China (RUC), supervised by Prof. Xin Zhao. I am expected to graduate in June 2029.

Research Motivation

I’m excited to build AGI.

Research Interests

  • Large Language Models
  • Agentic Systems

Education

Renmin University of China

Master’s Student · Gaoling School of Artificial Intelligence

Renmin University of China

Bachelor’s Student · Gaoling School of Artificial Intelligence

News

Recent milestones

Released Evo-Bench, a benchmark for evaluating language models’ harness-evolving capabilities.

Released Nanbeige4.2-3B, a compact general agentic model.

Released SWE-World and the SWE-Master technical report.

Joined the LLM Lab at BOSS Zhipin as a research intern.

Released ManuSearch, an open multi-agent deep-search framework.

Selected Publications

Publications and technical reports · Citations updated Aug 2026

* Equal contribution

View all on Scholar
Evo-Bench leaderboard homepage
ARXIV 2026

Research paper

Evo-Bench: Can Language Models Improve Agent Harness?

Lisheng Huang*, Chen Yang*, Hao Zhou, Huatong Song, Zongchao Chen, Ran Le, Yang Song, Wayne Xin Zhao, and Tao Zhang

A benchmark for evaluating whether language models can autonomously improve executable agent harnesses across search, office, and general-agent tasks.

Google ScholarCitations4
Paper Code Dataset Leaderboard
ManuSearch multi-agent framework diagram
EMNLP 2025

Published paper

ManuSearch: Democratizing Deep Search in Large Language Models with a Transparent and Open Multi-Agent Framework

Lisheng Huang*, Yichen Liu*, Jinhao Jiang*, Rongxiang Zhang, Jiahao Yan, Junyi Li, and Wayne Xin Zhao

A transparent multi-agent framework for open-web deep search and long-tail reasoning.

Google ScholarCitations28
Paper ACL Code
Overview of the SWE-World framework
EMNLP 2026

Published paper

SWE-World: Building Software Engineering Agents in Docker-Free Environments

Shuang Sun*, Huatong Song*, Lisheng Huang*, Jinhao Jiang, Ran Le, Zhihao Lv, et al.

A learned surrogate environment for scalable training and evaluation of software engineering agents.

Google ScholarCitations22
Paper Code
Overview of the SWE-Master post-training framework
TECH REPORT

Technical report

SWE-Master: Unleashing the Potential of Software Engineering Agents via Post-Training

Huatong Song*, Lisheng Huang*, Shuang Sun*, Jinhao Jiang, Ran Le, Daixuan Cheng, et al.

An open and reproducible post-training recipe for long-horizon software engineering agents.

Google ScholarCitations18
Paper Code
Nanbeige4.2-3B model performance comparison table
TECH REPORT

Technical report & model

Nanbeige4.2-3B: Unlocking Agentic Capabilities in a Compact Model

A compact general agentic model with 3B non-embedding parameters, trained for code, office, reasoning, and complex tool-use tasks.

Google ScholarCitations1
Paper Model

Experience

Nov 2025 — Jul 2026

LLM Lab · BOSS Zhipin

Research Intern