Reinforcement Learning with Action Chunking

📰 ArXiv cs.AI

arXiv:2507.07969v4 Announce Type: replace-cross Abstract: We present Q-chunking, a simple yet effective recipe for improving reinforcement learning (RL) algorithms for long-horizon, sparse-reward tasks. Our recipe is designed for the offline-to-online RL setting, where the goal is to leverage an offline prior dataset to maximize the sample-efficiency of online learning. Effective exploration and sample-efficient learning remain central challenges in this setting, as it is not obvious how the off

Published 12 May 2026

Full Article

Title: Reinforcement Learning with Action Chunking

Abstract:
arXiv:2507.07969v4 Announce Type: replace-cross Abstract: We present Q-chunking, a simple yet effective recipe for improving reinforcement learning (RL) algorithms for long-horizon, sparse-reward tasks. Our recipe is designed for the offline-to-online RL setting, where the goal is to leverage an offline prior dataset to maximize the sample-efficiency of online learning. Effective exploration and sample-efficient learning remain central challenges in this setting, as it is not obvious how the off
Read full paper → ← Back to Reads