Revisiting the Minimalist Approach to Offline Reinforcement Learning
This work addresses the need for clarity in offline RL design choices, benefiting researchers and practitioners by providing a high-performing, simplified algorithm, though it is incremental as it builds on existing methods.
The authors tackled the problem of understanding the impact of minor design choices in offline reinforcement learning algorithms by proposing ReBRAC, a minimalist algorithm built on TD3+BC, which achieved state-of-the-art performance among ensemble-free methods on 51 datasets across D4RL and V-D4RL benchmarks.
Recent years have witnessed significant advancements in offline reinforcement learning (RL), resulting in the development of numerous algorithms with varying degrees of complexity. While these algorithms have led to noteworthy improvements, many incorporate seemingly minor design choices that impact their effectiveness beyond core algorithmic advances. However, the effect of these design choices on established baselines remains understudied. In this work, we aim to bridge this gap by conducting a retrospective analysis of recent works in offline RL and propose ReBRAC, a minimalistic algorithm that integrates such design elements built on top of the TD3+BC method. We evaluate ReBRAC on 51 datasets with both proprioceptive and visual state spaces using D4RL and V-D4RL benchmarks, demonstrating its state-of-the-art performance among ensemble-free methods in both offline and offline-to-online settings. To further illustrate the efficacy of these design choices, we perform a large-scale ablation study and hyperparameter sensitivity analysis on the scale of thousands of experiments.