Seohong Park

Intention-Conditioned Flow Occupancy Models

Decoupled Q-Chunking

Dual Goal Representations

Transitive RL: Value Learning via Divide and Conquer

Scalable Offline Model-Based RL with Action Chunks

Diffusion Guidance Is a Controllable Policy Improvement Operator

Horizon Reduction Makes RL Scalable

OGBench: Benchmarking Offline Goal-Conditioned RL

Flow Q-Learning

Unsupervised-to-Online Reinforcement Learning