Lang Feng

PhD Student @ NTU Singapore

Reinforcement Learning
Large Language Models
Agentic RL

lang005@e.ntu.edu.sg

Google Scholar | Github | Twitter

About Me

I am currently pursuing my Ph.D. degree at the College of Computing and Data Science at Nanyang Technological University, fortunate to be advised by Prof. Bo An. Previously, I received my M.Eng. from Zhejiang University under the supervision of Prof. Gang Pan, and my B.Eng. (Hons.) from Southeast University under the supervision of Prof. Liang Wu.

I am leading the verl-agent project, an open-source framework for training LLM/VLM agents through reinforcement learning. I am scaling RL at ByteDance Seed, mentored by Yujia Qin. If you are seeking any form of academic collaboration, please feel free to email me.

Research Interests:

Scaling RL for LLMs: Scaling reinforcement learning for LLMs through data, models, systems, and training recipes
Agentic RL: Training long-horizon LLM agents through robust and efficient RL optimization
LLM-based Agents: Building general-purpose agents powered by LLMs for interactive and tool-use tasks

News

2026.05: Four papers accepted by ICML 2026!
2026.04: AgentOCR accepted by ACL 2026 (Main)! Check out our project!
2026.02: Dr. MAS released! Stable end-to-end RL for multi-agent LLM systems. Check out our project!
2026.01: One paper accepted by ICLR 2026!
2025.11: One paper accepted by AAAI 2026 (Oral)!
2025.09: GiGPO accepted at NeurIPS 2025! Check out our project!
2025.05: CoSo accepted by ICML 2025! An entropy-based agentic RL algorithm for VLM agent training. Check out our project!
2025.05: ASOR accepted by ICML 2025 (Spotlight)!
2025.05: One paper accepted by KDD 2025!

Selected Publications

For a complete list of publications, please refer to my Google Scholar.

* Equal contribution

Dr. MAS: Stable Reinforcement Learning for Multi-Agent LLM Systems

Lang Feng, Longtao Zheng, Shuo He, Fuxiang Zhang, Bo An

Preprint

AgentOCR: Reimagining Agent History via Optical Self-Compression

Lang Feng*, Fuchao Yang*, Feng Chen, Xin Cheng, Haiyang Xu, Zhenglin Wan, Ming Yan, Bo An

ACL 2026

Hierarchy-of-Groups Policy Optimization for Long-Horizon Agentic Tasks

Shuo He, Lang Feng, Qi Wei, Xin Cheng, Lei Feng, Bo An

ICLR 2026

From Stimuli to Minds: Enhancing Psychological Reasoning in LLMs via Bilateral Reinforcement Learning

Yichao Feng, Haoran Luo, Lang Feng, Shuai Zhao, Anh Tuan Luu

AAAI 2026 Oral

Group-in-Group Policy Optimization for LLM Agent Training

Lang Feng, Zhenghai Xue, Tingcong Liu, Bo An

NeurIPS 2025

TimeMaster: Training Time-Series Multimodal LLMs to Reason via Reinforcement Learning

Junru Zhang*, Lang Feng*, Xu Guo, Yuhan Wu, Yabo Dong, Duanqing Xu

BERT2S@NeurIPS 2025

Towards Efficient Online Tuning of VLM Agents via Counterfactual Soft Reinforcement Learning

Lang Feng, Weihao Tan, Zhiyi Lyu, Longtao Zheng, Haiyang Xu, Ming Yan, Fei Huang, Bo An

ICML 2025

Policy Regularization on Globally Accessible States in Cross-Dynamics Reinforcement Learning

Zhenghai Xue, Lang Feng, Jiacheng Xu, Kang Kang, Xiang Wen, Bo An, Shuicheng Yan

ICML 2025 Spotlight

Diffusion-Guided Diversity for Single Domain Generalization in Time Series Classification

Junru Zhang, Lang Feng, Xu Guo, Han Yu, Yabo Dong, Duanqing Xu

KDD 2025

Resisting Stochastic Risk in Diffusion Planners with the Trajectory Aggregation Tree

Lang Feng, Pengjie Gu, Bo An, Gang Pan

ICML 2024 Spotlight

Diverse Intra- and Inter-Domain Activity Style Fusion for Cross-Person Generalization in Activity Recognition

Junru Zhang, Lang Feng, Zhidan Liu, Yuhan Wu, Yang He, Yabo Dong, Duanqing Xu

KDD 2024

Multi-Level Firing with Spiking DS-ResNet: Enabling Better and Deeper Directly-Trained Spiking Neural Networks

Lang Feng, Qianhui Liu, Huajin Tang, De Ma, Gang Pan

IJCAI 2022

Invited Talks

2026.02: AgentOCR: Reimagining Agent History via Optical Self-Compression. Presented at the China Society of Image and Graphics (CSIG).
2025.07: Toward Long-Horizon RL Training for LLM Agents. Presented at Agent for SWE, hosted by LF AI & Data Singapore.
2025.06: Group-in-Group Policy Optimization for LLM Agent Training. Presented at the ROLL Team, Alibaba Group.

Academic Services

Program Committee/reviewer: ICML (2024, 2025, 2026), NeurIPS (2025, 2026), ICLR (2026), KDD (2025, 2026), AAAI (2026)
Journal reviewer: TNNLS