When should we prefer Decision Transformers for Offline Reinforcement Learning?

TL;DR

This paper empirically compares Decision Transformers with CQL and BC in offline RL, revealing their strengths and weaknesses across various data qualities, task complexities, and stochastic environments, with insights on scaling and architecture.

Contribution

It provides a comprehensive empirical analysis of when to prefer Decision Transformers over other offline RL algorithms, including design and scaling recommendations.

Findings

DT requires more data than CQL but is more robust.

DT outperforms in sparse-reward and low-quality data scenarios.

Scaling data for DT improves Atari scores significantly.

Abstract

Offline reinforcement learning (RL) allows agents to learn effective, return-maximizing policies from a static dataset. Three popular algorithms for offline RL are Conservative Q-Learning (CQL), Behavior Cloning (BC), and Decision Transformer (DT), from the class of Q-Learning, Imitation Learning, and Sequence Modeling respectively. A key open question is: which algorithm is preferred under what conditions? We study this question empirically by exploring the performance of these algorithms across the commonly used D4RL and Robomimic benchmarks. We design targeted experiments to understand their behavior concerning data suboptimality, task complexity, and stochasticity. Our key findings are: (1) DT requires more data than CQL to learn competitive policies but is more robust; (2) DT is a substantially better choice than both CQL and BC in sparse-reward and low-quality data settings; (3) DT and BC are preferable as task horizon increases, or when data is obtained from human demonstrators; and (4) CQL excels in situations characterized by the combination of high stochasticity and low data quality. We also investigate architectural choices and scaling trends for DT on Atari and D4RL and make design/scaling recommendations. We find that scaling the amount of data for DT by 5x gives a 2.5x average score improvement on Atari.