Simple Ingredients for Offline Reinforcement Learning

TL;DR

This paper investigates the challenges of offline reinforcement learning with heterogeneous data sources, revealing that increasing model scale with simple methods like AWAC and IQL improves performance and outperforms state-of-the-art algorithms.

Contribution

It demonstrates that model scale, rather than algorithm complexity, is crucial for handling diverse data in offline RL, and shows simple scaled methods can outperform complex algorithms.

Findings

Scaling up models improves performance with diverse data

Simple methods like AWAC and IQL outperform complex algorithms

Increased data diversity can cause performance deterioration in existing methods

Abstract

Offline reinforcement learning algorithms have proven effective on datasets highly connected to the target downstream task. Yet, leveraging a novel testbed (MOOD) in which trajectories come from heterogeneous sources, we show that existing methods struggle with diverse data: their performance considerably deteriorates as data collected for related but different tasks is simply added to the offline buffer. In light of this finding, we conduct a large empirical study where we formulate and test several hypotheses to explain this failure. Surprisingly, we find that scale, more than algorithmic considerations, is the key factor influencing performance. We show that simple methods like AWAC and IQL with increased network size overcome the paradoxical failure modes from the inclusion of additional data in MOOD, and notably outperform prior state-of-the-art algorithms on the canonical D4RL benchmark.