Rui Yang | Foundation Models for Agents & Reinforcement Learning

I am a Ph.D. candidate in Computer Science at UIUC , advised by Professors Tong Zhang and Huan Zhang. Previously, I received bachelor’s and master’s degrees from Tsinghua University and HKUST .

My research focuses on building reliable interactive agents powered by foundation models that can perceive multimodal inputs, reason about complex tasks, evaluate outcomes, and act effectively in dynamic environments. I also study robust learning under distribution shift and imperfect feedback, spanning LLM/VLM alignment, reward modeling, and robust offline and goal-conditioned RL.

Multimodal agents

Scalable post-training and evaluation for multimodal agents that perceive accurately, assess their current state, and reason and act effectively over long horizons.

OpenWebRL · GUI-Libra · GUI-Actor · MeMento · ERA · EmbodiedBench

Trustworthy alignment & evaluation

Aligning LLMs/VLMs with heterogeneous human preferences, together with robust reasoning evaluation under changing inputs.

RiC · GRM · MiCRo · DRMs · DynaMath

Robust and generalizable RL

Robust offline and goal-conditioned RL under data corruption, observation shifts, and unseen goals.

RORL · RIQL · GOAT · WGCSL

I am actively seeking full-time opportunities in foundation models, AI agents, and reinforcement learning.

News

🌟 2026.06 We released OpenWebRL, an open framework for training visual web agents with online multi-turn RL on live websites, and Orchard, an open-source framework with scalable training recipes across diverse agent domains.
🎉 2026.04 ReCAP was accepted to ICML 2026.
🌟 2026.02 We released GUI-Libra, a data-efficient post-training recipe for GUI agents that achieves strong online performance using 81K open-source examples.
🎉 2026.01 BEAT and DROCO were accepted to ICLR 2026.
🎉 2025.11 MiCRo received an EMNLP 2025 Outstanding Paper Award.
🎉 2025.09 GUI-Actor and ADG were accepted to NeurIPS 2025, and MergeBench to the Datasets & Benchmarks Track.
🎉 2025.05 EmbodiedBench was accepted to ICML 2025 as an oral paper.

Selected Publications

^* Equal contribution. Selected papers I led or co-led; see Google Scholar for the full list.

Foundation Models for Interactive Agents

OpenWebRL: Demystifying Online Multi-turn Reinforcement Learning for Visual Web Agents.
Preprint 2026 [Code] [Project] [Models & Data]
Rui Yang^*, Qianhui Wu^*, Yuxi Chen, Hao Bai, Wenlin Yao, Hao Cheng, Baolin Peng, Huan Zhang, Tong Zhang, Jianfeng Gao.
GUI-Libra: Training Native GUI Agents to Reason and Act with Action-aware Supervision and Partially Verifiable RL.
Preprint 2026 [Code] [Project]
Rui Yang, Qianhui Wu, Zhaoyang Wang, Hanyang Chen, Ke Yang, Hao Cheng, Huaxiu Yao, Baolin Peng, Huan Zhang, Jianfeng Gao, Tong Zhang.
EmbodiedBench: Comprehensive Benchmarking Multi-modal Large Language Models for Vision-Driven Embodied Agents.
ICML 2025 Oral [Code] [Project]
Rui Yang^*, Hanyang Chen^*, Junyu Zhang^*, Mark Zhao^*, Cheng Qian, Kangrui Wang, Qineng Wang, Teja Venkat Koripella, Marziyeh Movahedi, Manling Li, Heng Ji, Huan Zhang, Tong Zhang.
GUI-Actor: Coordinate-Free Visual Grounding for GUI Agents.
NeurIPS 2025 [Code] [Project]
Qianhui Wu^*, Kanzhi Cheng^*, Rui Yang^*, Chaoyun Zhang, Jianwei Yang, Huiqiang Jiang, Jian Mu, Baolin Peng, Bo Qiao, Reuben Tan, Si Qin, Lars Liden, Qingwei Lin, Huan Zhang, Tong Zhang, Jianbing Zhang, Dongmei Zhang, Jianfeng Gao.

Reliable Reasoning and Alignment

DynaMath: A Dynamic Visual Benchmark for Evaluating Mathematical Reasoning Robustness of Vision Language Models.
ICLR 2025 [Code] [Project]
Chengke Zou^*, Xingang Guo^*, Rui Yang^*, Junyu Zhang, Bin Hu, Huan Zhang.
Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs.
NeurIPS 2024 [Code]
Rui Yang, Ruomeng Ding, Yong Lin, Huan Zhang, Tong Zhang.
Rewards-in-Context: Multi-objective Alignment of Foundation Models with Dynamic Preference Adjustment.
ICML 2024 [Code]
Rui Yang^*, Xiaoman Pan^*, Feng Luo^*, Shuang Qiu^*, Han Zhong, Dong Yu, Jianshu Chen.
MiCRo: Mixture Modeling and Context-aware Routing for Personalized Preference Learning.
EMNLP 2025 Outstanding Paper
Jingyan Shen^*, Jiarui Yao^*, Rui Yang^*, Yifan Sun, Feng Luo, Rui Pan, Tong Zhang, Han Zhao.

Robust Offline RL and Goal-Conditioned RL

Towards Robust Offline Reinforcement Learning under Diverse Data Corruption.
ICLR 2024 Spotlight [Code]
Rui Yang^*, Han Zhong^*, Jiawei Xu^*, Amy Zhang, Chongjie Zhang, Lei Han, Tong Zhang.
RORL: Robust Offline Reinforcement Learning via Conservative Smoothing.
NeurIPS 2022 Spotlight [Code]
Rui Yang^*, Chenjia Bai^*, Xiaoteng Ma, Zhaoran Wang, Chongjie Zhang, Lei Han.
What Is Essential for Unseen Goal Generalization of Offline Goal-conditioned RL?.
ICML 2023 [Code]
Rui Yang, Yong Lin, Xiaoteng Ma, Hao Hu, Chongjie Zhang, Tong Zhang.
Rethinking Goal-conditioned Supervised Learning and Its Connection to Offline RL.
ICLR 2022 [Code]
Rui Yang, Yiming Lu, Wenzhe Li, Hao Sun, Meng Fang, Yali Du, Xiu Li, Lei Han, Chongjie Zhang.

Experience

Research Intern, Apple Foundation Models, 2026.
Research Intern, Microsoft Research, Deep Learning Group, 2025.
Research Intern, Tencent AI Lab and Robotics X Lab, 2020–2022 (multiple internship terms).
Machine Learning Intern, Meituan, 2019.

Service

Conference Reviewer: ICML, ICLR, NeurIPS (NeurIPS 2023 Top Reviewer), ACL/ARR, ICRA, AAMAS.

Journal Reviewer: IEEE Robotics and Automation Letters (RA-L), IEEE Transactions on Neural Networks and Learning Systems (TNNLS), IEEE Transactions on Artificial Intelligence (TAI), Machine Learning, Journal of Artificial Intelligence Research.

Teaching Assistant: CS 441 Applied Machine Learning, UIUC; COMP 4211 Machine Learning, HKUST; COMP 1021 Introduction to Computer Science, HKUST.

Hobbies

Outside research, I enjoy 🏃 running, 🏓 table tennis, and 🏊 swimming. My personal bests are 1 h 30 min in the half marathon and 3 h 36 min in the marathon.