Offline Actor-Critic Reinforcement Learning Scales to Large Models

TL;DR

This paper demonstrates that offline actor-critic reinforcement learning can effectively scale to large models like transformers, outperforming behavioral cloning on multi-task control tasks and enabling multi-domain mastery from sub-optimal data.

Contribution

It introduces a scalable offline actor-critic method with a Perceiver-based architecture, showing its effectiveness across large models and complex multi-task environments.

Findings

Offline actor-critic scales similarly to supervised learning.

Outperforms behavioral cloning on 132 control tasks.

Enables multi-task learning from sub-optimal demonstrations.

Abstract

We show that offline actor-critic reinforcement learning can scale to large models - such as transformers - and follows similar scaling laws as supervised learning. We find that offline actor-critic algorithms can outperform strong, supervised, behavioral cloning baselines for multi-task training on a large dataset containing both sub-optimal and expert behavior on 132 continuous control tasks. We introduce a Perceiver-based actor-critic model and elucidate the key model features needed to make offline RL work with self- and cross-attention modules. Overall, we find that: i) simple offline actor critic algorithms are a natural choice for gradually moving away from the currently predominant paradigm of behavioral cloning, and ii) via offline RL it is possible to learn multi-task policies that master many domains simultaneously, including real robotics tasks, from sub-optimal demonstrations or self-generated data.