PQV-Mobile: A Combined Pruning and Quantization Toolkit to Optimize Vision Transformers for Mobile Applications
This addresses the need for efficient ViT deployment on resource-constrained mobile systems, though it is incremental as it builds on existing compression techniques.
The paper tackles the problem of optimizing Vision Transformers for mobile deployment by introducing PQV-Mobile, a toolkit that combines pruning and quantization, achieving a 7.18x latency reduction with only a 2.24% accuracy loss on a DeiT model.
While Vision Transformers (ViTs) are extremely effective at computer vision tasks and are replacing convolutional neural networks as the new state-of-the-art, they are complex and memory-intensive models. In order to effectively run these models on resource-constrained mobile/edge systems, there is a need to not only compress these models but also to optimize them and convert them into deployment-friendly formats. To this end, this paper presents a combined pruning and quantization tool, called PQV-Mobile, to optimize vision transformers for mobile applications. The tool is able to support different types of structured pruning based on magnitude importance, Taylor importance, and Hessian importance. It also supports quantization from FP32 to FP16 and int8, targeting different mobile hardware backends. We demonstrate the capabilities of our tool and show important latency-memory-accuracy trade-offs for different amounts of pruning and int8 quantization with Facebook Data Efficient Image Transformer (DeiT) models. Our results show that even pruning a DeiT model by 9.375% and quantizing it to int8 from FP32 followed by optimizing for mobile applications, we find a latency reduction by 7.18X with a small accuracy loss of 2.24%. The tool is open source.