Will Liang

(github, linkedin, gscholar, twitter, notes)

I'm a computer science student at the University of Pennsylvania. I'm working on robot learning in the GRASP Laboratory, where I'm fortunate to be mentored by Jason Ma and advised by Dinesh Jayaraman, Osbert Bastani, and Jianbo Shi. My interests are multimodal representations, large language models, and reinforcement learning for intelligent, robust embodied agents.

Previously, I've interned at Fungible, Anduril Industries, and Hudson River Trading. I've also dabbled in traditional art and game development.

News

Oct 2023: Taught a robot hand some pen spinning tricks with evolutionary reward design—Eureka!
Aug 2023: Met awesome people at CMMRS, hosted by MPI-SWS, Cornell, and University of Maryland.
May 2023: Real robot experiments for multimodal representations (LIV) released on arXiv.
Aug 2021: Started school at Penn.
Aug 2020: Played Snake with evolutionary algorithms and tensor decomposition, mentored by Ryan Solgi.

Research

Eureka: Human-Level Reward Design via Coding Large Language Models

Yecheng Jason Ma, William Liang, Guanzhi Wang, De-An Huang, Osbert Bastani, Dinesh Jayaraman, Yuke Zhu, Linxi "Jim" Fan, Anima Anandkumar

International Conference on Learning Representations (ICLR), 2024
CoRL Workshop on Language and Robot Learning (Oral), 2023
CoRL Towards Generalist Robotics Workshop (Oral), 2023
NeurIPS Agent Learning in Open-Endedness Workshop (Oral), 2023
NeurIPS Workshop on Foundation Models for Decision Making, 2023

(paper, website, code)

LIV: Language-Image Representations and Rewards for Robotic Control

Yecheng Jason Ma, William Liang, Vaidehi Som, Vikash Kumar, Amy Zhang, Osbert Bastani, Dinesh Jayaraman

arXiv, 2023 (extended ICML)
RSS Workshop on Utilizing Language for Robot Learning (Oral), 2023

(paper)

Tensor Shape Search for Optimum Data Compression

Ryan Solgi, Zichang He, William Liang, Zheng Zhang

arXiv, 2022

(paper)