🎓 Education
- Ph.D. in Computer Science, Tsinghua University (2025–)
- B.Eng. in Computer Science, Tongji University, 2021–2025
🔬 Research Interests
Large Language Model (LLM) post-training, Reinforcement Learning (RL), and GUI / Computer-Use Agents.
💡 Current Research
- ScaleCUA — scaling computer-use agents with verifiable task synthesis and efficient online RL (open-source SOTA on OSWorld / ScienceBoard).
- Agentic RL systems — asynchronous, train–inference-decoupled online RL for multi-task, multi-turn agents.
- One year at Zhipu AI on large-model post-training and GUI Agent RL.
📝 Publications
-
Bowen Lv, Zehan Qi ScaleCUA: Scaling Computer Use Agents with Verifiable Task Synthesis and Efficient Online RL (first author) Preprint, 2026. — 68.7% OSWorld / 54.0% ScienceBoard (open-source SOTA)
-
Xueqiao Sun, Xiao Liu, Bowen Lv, et al. KARL: Reinforcement Learning for LLM Agents on Multi-Turn Knowledge-Intensive Agentic Tasks ACL 2026 main conference.
-
Wenyi Hong, Xiaotao Gu, ... , Bowen Lv, ... , GLM-V Team GLM-5V-Turbo: Toward a Native Foundation Model for Multimodal Agents Preprint, 2026.
[arxiv] -
Hanchen Zhang, Xiao Liu, Bowen Lv, et al.
AgentRL: Scaling Agentic Reinforcement Learning with a Multi-Turn, Multi-Task Framework.
Preprint, 2025.
[arxiv] -
Di Zhang, Bowen Lv, et al.
Focus on What Matters: Separated Models for Visual-Based RL Generalization.
NeurIPS 2024.
[arxiv]
📝 Blogs
🔗 Links
📫 Contact
📧 1486404293@qq.com / lvbowen1228@gmail.com
🔗 GitHub Profile
📄 Google Scholar