Decision MetaMamba: Enhancing Selective SSM in Offline RL with Heterogeneous Sequence Mixing

TL;DR

Decision MetaMamba introduces a novel sequence mixing approach in offline RL that improves performance and preserves local information, outperforming existing models with fewer parameters.

Contribution

The paper proposes Decision MetaMamba, a new structure replacing token mixing with dense layer-based sequence mixing and modified positional encoding, enhancing selective mechanisms in offline RL.

Findings

Achieves state-of-the-art results across diverse RL tasks.

Maintains high performance with a compact model size.

Prevents information loss during sequence processing.

Abstract

Mamba-based models have drawn much attention in offline RL. However, their selective mechanism often detrimental when key steps in RL sequences are omitted. To address these issues, we propose a simple yet effective structure, called Decision MetaMamba (DMM), which replaces Mamba's token mixer with a dense layer-based sequence mixer and modifies positional structure to preserve local information. By performing sequence mixing that considers all channels simultaneously before Mamba, DMM prevents information loss due to selective scanning and residual gating. Extensive experiments demonstrate that our DMM delivers the state-of-the-art performance across diverse RL tasks. Furthermore, DMM achieves these results with a compact parameter footprint, demonstrating strong potential for real-world applications. Code is available at https://github.com/too-z/decision-metamamba