Accelerating the Low-Rank Decomposed Models

TL;DR

This paper explores modifications to low-rank tensor decomposition techniques to enable efficient AI model compression, balancing accuracy, memory reduction, and speed without increasing model depth.

Contribution

It proposes new methods to adapt low-rank decomposition for AI models, reducing redundancy while maintaining performance and avoiding increased latency.

Findings

Achieved significant parameter reduction without increasing model depth

Improved training and inference speed in decomposed models

Maintained high accuracy with compressed models

Abstract

Tensor decomposition is a mathematically supported technique for data compression. It consists of applying some kind of a Low Rank Decomposition technique on the tensors or matrices in order to reduce the redundancy of the data. However, it is not a popular technique for compressing the AI models duo to the high number of new layers added to the architecture after decomposition. Although the number of parameters could shrink significantly, it could result in the model be more than twice deeper which could add some latency to the training or inference. In this paper, we present a comprehensive study about how to modify low rank decomposition technique in AI models so that we could benefit from both high accuracy and low memory consumption as well as speeding up the training and inference