Zechu (Steven) Li

I received my master's degree from TU-Darmstadt from Oct. 2023 to May. 2024, advised by Prof. Georgia Chalvatzaki. My research interest lies in reinforcement learning, especially its applications (e.g., robotics, finance, and transportation) and high-performance and scalable systems.

Prior to this, I was a research assistant at MIT CSAIL, advised by Prof. Pulkit Agrawal, where I conducted research on massively parallel simulation and sim-to-real in robotics.

I received my bachelor's degree from Columbia University in May 2022, majoring in computer science. During my undergraduate studies, I was fortunate to work with Prof. Xiaodong Wang, Prof. Anwar Walid and Prof. Sharon (Xuan) Di.

Email  /  Google Scholar  /  GitHub /  LinkedIn

Selected Publications [* Equal contribution]
sym Learning Multimodal Behaviors from Scratch with Diffusion Policy Gradient
Zechu Li, Rickmer Krohn, Tao Chen, Anurag Ajay, Pulkit Agrawal, Georgia Chalvatzaki
NeurIPS, 2024
paper / website / code

A novel actor-critic algorithm that learns multimodal policies as diffusion models from scratch while maintaining versatile behaviors.

sym Reconciling Reality through Simulation: A Real-to-Sim-to-Real Approach for Robust Manipulation
Marcel Torne, Anthony Simeonov, Zechu Li, April Chan, Tao Chen, Abhishek Gupta*, Pulkit Agrawal*
Robotics: Science and Systems (RSS), 2024
paper / website / code

A system for robustifying real-world imitation learning policies via reinforcement learning in "digital twin" simulation environments constructed on the fly from small amounts of real-world data.

sym Parallel Q-Learning: Scaling Off-policy Reinforcement Learning under Massively Parallel Simulation
Zechu Li*, Tao Chen*, Zhang-Wei Hong, Anurag Ajay, Pulkit Agrawal
ICML, 2023
paper / code

A novel parallel Q-learning framework that scales off-policy learning to 10000+ parallel environments.

sym Homomorphic Matrix Completion
Xiao-Yang Liu*, Zechu Li*, Xiaodong Wang
NeurIPS, 2022
paper

A homomorphic matrix completion algorithm that satisfies the differential privacy property and reduces the best-known error bound to EXACT recovery at a price of more samples.

Open Source Projects
sym FinRL: Financial Reinforcement Learning
project page / code / GitHub Star

The first open-source framework to show the great potential of financial reinforcement learning.

sym ElegantRL “小雅”: Massively Parallel Library for Cloud-native Deep Reinforcement Learning
project page / code / GitHub Star

A massively parallel library for cloud-native deep reinforcement learning (DRL) applications.

As a leader of this project, I have been contributing to

  • develop a series of large-scale training frameworks,
  • implemente SOTA algorithms and techniques,
  • build the documentation website.
Starting from Mar. 2021, I started to write tutorial blogs for the community,

Book Chapter
sym High-performance Tensor Decompositions for Compressing and Accelerating Deep Neural Networks
Xiao-Yang Liu, Yiming Fang, Liuqing Yang, Zechu Li, Anwar Walid
Tensors for Data Processing, Elsevier, 2021
chapter / book

This chapter takes a practical approach to seek a better efficiency-accuracy trade-off, which utilizes high performance tensor decompositions to compress and accelerate neural networks by exploiting low-rank structures of the network weight matrix.


This guy makes a nice webpage.