Towards Lightweight Transformer via Group-wise Transformation for Vision-and-Language Tasks

TL;DR

This paper introduces LW-Transformer, a lightweight Transformer model using Group-wise Transformation to reduce parameters and computation while maintaining performance on vision-and-language tasks.

Contribution

It proposes a universal lightweight Transformer architecture that preserves key properties of standard Transformers, applicable to various vision-and-language tasks.

Findings

Significant reduction in parameters and computation.

Competitive performance on multiple vision-and-language benchmarks.

Effective generalization to image classification with Swin-Transformer.

Abstract

Despite the exciting performance, Transformer is criticized for its excessive parameters and computation cost. However, compressing Transformer remains as an open problem due to its internal complexity of the layer designs, i.e., Multi-Head Attention (MHA) and Feed-Forward Network (FFN). To address this issue, we introduce Group-wise Transformation towards a universal yet lightweight Transformer for vision-and-language tasks, termed as LW-Transformer. LW-Transformer applies Group-wise Transformation to reduce both the parameters and computations of Transformer, while also preserving its two main properties, i.e., the efficient attention modeling on diverse subspaces of MHA, and the expanding-scaling feature transformation of FFN. We apply LW-Transformer to a set of Transformer-based networks, and quantitatively measure them on three vision-and-language tasks and six benchmark datasets. Experimental results show that while saving a large number of parameters and computations, LW-Transformer achieves very competitive performance against the original Transformer networks for vision-and-language tasks. To examine the generalization ability, we also apply our optimization strategy to a recently proposed image Transformer called Swin-Transformer for image classification, where the effectiveness can be also confirmed