RePreM: Representation Pre-training with Masked Model for Reinforcement Learning

TL;DR

RePreM introduces a masked sequence modeling approach for pre-training in reinforcement learning, improving long-term dynamics understanding and transferability across tasks without complex algorithms.

Contribution

It presents a simple, effective pre-training method using masked modeling in RL that captures long-term dynamics and scales well with data and model size.

Findings

RePreM enhances dynamic prediction accuracy.

It improves transfer learning performance.

RePreM enables sample-efficient RL with various algorithms.

Abstract

Inspired by the recent success of sequence modeling in RL and the use of masked language model for pre-training, we propose a masked model for pre-training in RL, RePreM (Representation Pre-training with Masked Model), which trains the encoder combined with transformer blocks to predict the masked states or actions in a trajectory. RePreM is simple but effective compared to existing representation pre-training methods in RL. It avoids algorithmic sophistication (such as data augmentation or estimating multiple models) with sequence modeling and generates a representation that captures long-term dynamics well. Empirically, we demonstrate the effectiveness of RePreM in various tasks, including dynamic prediction, transfer learning, and sample-efficient RL with both value-based and actor-critic methods. Moreover, we show that RePreM scales well with dataset size, dataset quality, and the scale of the encoder, which indicates its potential towards big RL models.