Welcome! The Princeton RL lab aims to develop effective and principled reinforcement learning algorithms. This includes studying/analyzing existing methods, building new methods, applying methods to new applications, and understanding what happens when you scale up these methods. Currently, there is a focus on understanding unsupervised/self-supervised methods, including methods that can propose their own goals and autonomously collect data.
Code for our projects is available at https://github.com/Princeton-RL.
Recent Publications
-
A Single Goal is All You Need: Skills and Exploration Emerge from Contrastive RL without Rewards, Demonstrations, or Subgoals
Grace Liu∗, Michael Tang, Benjamin Eysenbach
Princeton University
Released at ICLR 2025 -
Accelerating Goal-Conditioned Reinforcement Learning Algorithms and Research
Michał Bortkiewicz¹, Władysław Pałucki², Vivek Myers³, Tadeusz Dziarmaga⁴, Tomasz Arczewski⁴, Łukasz Kucinski²⁵⁶, Benjamin Eysenbach⁷
¹Warsaw University of Technology, ²University of Warsaw, ³UC Berkeley, ⁴Jagiellonian University, ⁵Polish Academy of Sciences, ⁶IDEAS NCBR, ⁷Princeton University
Released at ICLR 2025 -
Horizon Generalization in Reinforcement Learning
Vivek Myers∗¹, Catherine Ji∗², Benjamin Eysenbach²
¹UC Berkeley, ²Princeton University
Released at ICLR 2025 -
The “Law” of the Unconscious Contrastive Learner: Probabilistic Alignment of Unpaired Modalities
Yongwei Che, Benjamin Eysenbach
Princeton University
Released at ICLR 2025 -
Can a MISL Fly? Analysis and Ingredients for Mutual Information Skill Learning
Chongyi Zheng∗, Jens Tuyls∗, Joanne Peng, Benjamin Eysenbach
Princeton University
Released at ICLR 2025 -
OGBENCH: Benchmarking Offline Goal-Conditioned RL
Seohong Park¹, Kevin Frans¹, Benjamin Eysenbach², Sergey Levine¹
¹University of California, Berkeley, ²Princeton University
Released at ICLR 2025
