Tensorized Embedding Layers for Efficient Model Compression

TL;DR

This paper proposes a tensor train decomposition method to compress large embedding layers in NLP models, significantly reducing size with minimal performance loss, enabling deployment in resource-constrained environments.

Contribution

The paper introduces a novel tensor train-based parametrization for embedding layers, achieving high compression ratios while maintaining or improving performance.

Findings

Significant model size reduction with negligible performance drop.

Effective across various architectures including MLPs, LSTMs, and Transformers.

Trade-offs between compression ratio and accuracy are analyzed.

Abstract

The embedding layers transforming input words into real vectors are the key components of deep neural networks used in natural language processing. However, when the vocabulary is large, the corresponding weight matrices can be enormous, which precludes their deployment in a limited resource setting. We introduce a novel way of parametrizing embedding layers based on the Tensor Train (TT) decomposition, which allows compressing the model significantly at the cost of a negligible drop or even a slight gain in performance. We evaluate our method on a wide range of benchmarks in natural language processing and analyze the trade-off between performance and compression ratios for a wide range of architectures, from MLPs to LSTMs and Transformers.