MossNet: Mixture of State-Space Experts is a Multi-Head Attention

TL;DR

MossNet introduces a mixture-of-experts architecture that emulates multi-head attention using state-space models, achieving superior performance and efficiency in language modeling tasks.

Contribution

It presents a novel mixture-of-experts approach in state-space models to emulate multi-head attention, enhancing expressiveness and scalability over existing models.

Findings

Outperforms transformer- and SSM-based architectures of similar size.

Scales effectively to trillions of tokens with superior performance.

Demonstrates favorable runtime and resource efficiency on real devices.

Abstract

Large language models (LLMs) have significantly advanced generative applications in natural language processing (NLP). Recent trends in model architectures revolve around efficient variants of transformers or state-space/gated-recurrent models (SSMs, GRMs). However, prevailing SSM/GRM-based methods often emulate only a single attention head, potentially limiting their expressiveness. In this work, we propose MossNet, a novel mixture-of-state-space-experts architecture that emulates a linear multi-head attention (MHA). MossNet leverages a mixture-of-experts (MoE) implementation not only in channel-mixing multi-layered perceptron (MLP) blocks but also in the time-mixing SSM kernels to realize multiple "attention heads." Extensive experiments on language modeling and downstream evaluations show that MossNet outperforms both transformer- and SSM-based architectures of similar model size and data budgets. Larger variants of MossNet, trained on trillions of tokens, further confirm its scalability and superior performance. In addition, real-device profiling on a Samsung Galaxy S24 Ultra and an Nvidia A100 GPU demonstrate favorable runtime speed and resource usage compared to similarly sized baselines. Our results suggest that MossNet is a compelling new direction for efficient, high-performing recurrent LLM architectures.