Lichang Chen

I am a PhD candidate at the Computer Science Department, University of Maryland, College Park, where I work closely with Dr. Heng Huang, Dr. Tianyi Zhou, and Dr. Tom Goldstein. I obtained my bachelor's degree from Zhejiang University. I can be reached at {bob}{my-last-name}@cs.umd.edu.

My name in Chinese: 陈力ē•…

Google Scholar  /  Twitter  /  LinkedIn  /  Github

profile photo
Research Interests

My research interests are in steering the Foundation Models(including Large Large Models and Vision Languge Models) effectively and efficiently. šŸ“š Iā€™m currently working on multimodal alignment, dynamic evaluations of LLMs, and mitigating hackings in RLHF.šŸ“½ļø Below is my selected publications.

Experience

Google Deepmind, 2024.9 - present, Foundational Research on Gemini Games.
Google Deepmind, 2024.5 - 2024.8, The evaluation/alignment of Omni-Modality Language Models.
Google Research & Cloud AI Research, 2024.2 - 2024.5, The self-improvement of multimodal LLMs.
NVIDIA ADLR team, 2023.9 - 2024.1, Mitigate hackings in RLHF/More Robust Reward Models
Samsung AI Research, 2023.5 - 2023.8, Data Filter for Instruction Tuning.

PontTuset Figure 2 Figure 3 Figure 4
Selected Publications
PontTuset OPTune: Efficient Online Preference Tuning
Lichang Chen, Jiuhai Chen, Chenxi Liu, John Kirchenbauer, Davit Soselio, Chen Zhu, Tom Goldstein, Tianyi Zhou, Heng Huang
Arxiv, 2024

PontTuset OmnixR: Evaluating Omni-modality Language Models on Reasoning across Modalities
Lichang Chen, Hexiang Hu, Pranav Shyam, Ming-Hsuan Yang, Boqing Gong, et al.
Google Deepmind, 2024

PontTuset From Lists to Emojis: How Format Bias Affects Model Alignment
Lichang Chen*, Xuanchang Zhang*, Wei Xiong*, Tianyi Zhou, Heng Huang, Tong Zhang.
Arxiv, 2024

PontTuset ODIN: Disentangled Reward Mitigates Hacking in RLHF
Lichang Chen*, Chen Zhu*, Davit Soselio, Tianyi Zhou, Tom Goldstein, Heng Huang, Mohammad Shoeybi, Bryan Catanzaro
ICML, 2024.

PontTuset InstructZero: Efficient Instruction Optimization for Black-Box Large Language Models
Lichang Chen*, Jiuhai Chen*, Tom Goldstein, Heng Huang, Tianyi Zhou
ICML, 2024

PontTuset AlpaGasus: Training a Better Alpaca with Fewer Data
Lichang Chen*, Shiyang Li*, Jun Yan, Hai Wang, Kalpa Gunaratna, Vikas Yadav, Zheng Tang, Vijay Srinivasan, Tianyi Zhou, Heng Huang, Hongxia Jin
ICLR, 2024

PontTuset Advanced PPO finetuning tricks
Wei Shen, Jian Hu, Pengyu Zhao, Xiaonan He, Lichang Chen (Last Author)
Blogs, 2024

PontTuset Backdooring Instruction-Tuned Large Language Models with Virtual Prompt Injection
Jun Yan, Vikas Yadav, Shiyang Li, Lichang Chen, Zheng Tang, Hai Wang, Vijay Srinivasan, Xiang Ren, Hongxia Jin
NAACL , 2024

PontTuset HallusionBench: an image-context reasoning benchmark challenging for multi-modality models
Fuxiao Liu, Tianrui Guan, Zongxia Li, Lichang Chen, et al.
CVPR, 2024

PontTuset How Many Demonstrations Do You Need for In-context Learning?
Jiuhai Chen, Lichang Chen, Chen Zhu, Tianyi Zhou
EMNLP, 2023

PontTuset PTP: Boosting Stability and Performance of Prompt Tuning with Perturbation-Based Regularizer
Lichang Chen, Jiuhai Chen, Heng Huang, Minhao Cheng
EMNLP, 2023