Efficient Long-Range Transformers: You Need to Attend More, but Not Necessarily at Every Layer

TL;DR

This paper introduces MASFormer, a transformer variant that combines full and sparse attention across layers to efficiently model long sequences, achieving high performance with reduced computational costs.

Contribution

MASFormer innovatively integrates full and sparse attention in different layers, enabling efficient long-range dependency modeling with lower computational costs.

Findings

Achieves comparable performance to full attention transformers on NLP tasks.

Reduces computational cost by up to 75%.

Effective in long sequence modeling and generation tasks.

Abstract

Pretrained transformer models have demonstrated remarkable performance across various natural language processing tasks. These models leverage the attention mechanism to capture long- and short-range dependencies in the sequence. However, the (full) attention mechanism incurs high computational cost - quadratic in the sequence length, which is not affordable in tasks with long sequences, e.g., inputs with 8k tokens. Although sparse attention can be used to improve computational efficiency, as suggested in existing work, it has limited modeling capacity and often fails to capture complicated dependencies in long sequences. To tackle this challenge, we propose MASFormer, an easy-to-implement transformer variant with Mixed Attention Spans. Specifically, MASFormer is equipped with full attention to capture long-range dependencies, but only at a small number of layers. For the remaining layers, MASformer only employs sparse attention to capture short-range dependencies. Our experiments on natural language modeling and generation tasks show that a decoder-only MASFormer model of 1.3B parameters can achieve competitive performance to vanilla transformers with full attention while significantly reducing computational cost (up to 75%). Additionally, we investigate the effectiveness of continual training with long sequence data and how sequence length impacts downstream generation performance, which may be of independent interest.