Skip-Layer Attention: Bridging Abstract and Detailed Dependencies in Transformers

TL;DR

This paper proposes Skip-Layer Attention, a novel mechanism that allows direct attention between non-adjacent Transformer layers, improving the model's ability to capture complex dependencies and enhancing language modeling performance.

Contribution

The paper introduces Skip-Layer Attention, enabling direct inter-layer attention in Transformers, which improves dependency modeling without extra computational cost.

Findings

Enhanced language modeling performance

Better capture of abstract and detailed dependencies

Improved multi-head attention diversity

Abstract

The Transformer architecture has significantly advanced deep learning, particularly in natural language processing, by effectively managing long-range dependencies. However, as the demand for understanding complex relationships grows, refining the Transformer's architecture becomes critical. This paper introduces Skip-Layer Attention (SLA) to enhance Transformer models by enabling direct attention between non-adjacent layers. This method improves the model's ability to capture dependencies between high-level abstract features and low-level details. By facilitating direct attention between these diverse feature levels, our approach overcomes the limitations of current Transformers, which often rely on suboptimal intra-layer attention. Our implementation extends the Transformer's functionality by enabling queries in a given layer to interact with keys and values from both the current layer and one preceding layer, thus enhancing the diversity of multi-head attention without additional computational burden. Extensive experiments demonstrate that our enhanced Transformer model achieves superior performance in language modeling tasks, highlighting the effectiveness of our skip-layer attention mechanism.