Unlocking the Secrets of Linear Complexity Sequence Model from A Unified Perspective

TL;DR

This paper introduces the Linear Complexity Sequence Model (LCSM), unifying various sequence modeling techniques with a three-stage process to improve understanding and performance in language modeling and retrieval tasks.

Contribution

The paper proposes a unified framework for sequence models based on linear complexity, analyzing the impact of each stage and setting to enhance comprehension and performance.

Findings

Data-driven methods improve language modeling.

Hand-crafted methods enhance retrieval tasks.

Comprehensive analysis of stage settings impacts.

Abstract

We present the Linear Complexity Sequence Model (LCSM), a comprehensive solution that unites various sequence modeling techniques with linear complexity, including linear attention, state space model, long convolution, and linear RNN, within a single framework. The goal is to enhance comprehension of these models by analyzing the impact of each component from a cohesive and streamlined viewpoint. Specifically, we segment the modeling processes of these models into three distinct stages: Expand, Oscillation, and Shrink (EOS), with each model having its own specific settings. The Expand stage involves projecting the input signal onto a high-dimensional memory state. This is followed by recursive operations performed on the memory state in the Oscillation stage. Finally, the memory state is projected back to a low-dimensional space in the Shrink stage. We perform comprehensive experiments to analyze the impact of different stage settings on language modeling and retrieval tasks. Our results show that data-driven methods are crucial for the effectiveness of the three stages in language modeling, whereas hand-crafted methods yield better performance in retrieval tasks.