The official implementation for "FQ-ViT: Fully Quantized Vision Transformer without Retraining".

Overview

FQ-ViT [arXiv]

This repo contains the official implementation of "FQ-ViT: Fully Quantized Vision Transformer without Retraining".

Table of Contents

Introduction

Transformer-based architectures have achieved competitive performance in various CV tasks. Compared to the CNNs, Transformers usually have more parameters and higher computational costs, presenting a challenge when deployed to resource-constrained hardware devices.

Most existing quantization approaches are designed and tested on CNNs and lack proper handling of Transformer-specific modules. Previous work found there would be significant accuracy degradation when quantizing LayerNorm and Softmax of Transformer-based architectures. As a result, they left LayerNorm and Softmax unquantized with floating-point numbers. We revisit these two exclusive modules of the Vision Transformers and discover the reasons for degradation. In this work, we propose the FQ-ViT, the first fully quantized Vision Transformer, which contains two specific modules: Powers-of-Two Scale (PTS) and Log-Int-Softmax (LIS).

Layernorm quantized with Powers-of-Two Scale (PTS)

These two figures below show that there exists serious inter-channel variation in Vision Transformers than CNNs, which leads to unacceptable quantization errors with layer-wise quantization.

Taking the advantages of both layer-wise and channel-wise quantization, we propose PTS for LayerNorm's quantization. The core idea of PTS is to equip different channels with different Powers-of-Two Scale factors, rather than different quantization scales.

Softmax quantized with Log-Int-Softmax (LIS)

The storage and computation of attention map is known as a bottleneck for transformer structures, so we want to quantize it to extreme lower bit-width (e.g. 4-bit). However, if directly implementing 4-bit uniform quantization, there will be severe accuracy degeneration. We observe a distribution centering at a fairly small value of the output of Softmax, while only few outliers have larger values close to 1. Based on the following visualization, Log2 preserves more quantization bins than uniform for the small value interval with dense distribution.

Combining Log2 quantization with i-exp, which is a polynomial approximation of exponential function presented by I-BERT, we propose LIS, an integer-only, faster, low consuming Softmax.

The whole process is visualized as follow.

Getting Started

Install

  • Clone this repo.
git clone https://github.com/linyang-zhh/FQ-ViT.git
cd FQ-ViT
  • Create a conda virtual environment and activate it.
conda create -n fq-vit python=3.7 -y
conda activate fq-vit
  • Install PyTorch and torchvision. e.g.,
conda install pytorch=1.7.1 torchvision cudatoolkit=10.1 -c pytorch

Data preparation

You should download the standard ImageNet Dataset.

├── imagenet
│   ├── train
|
│   ├── val

Run

Example: Evaluate quantized DeiT-S with MinMax quantizer and our proposed PTS and LIS

python test_quant.py deit_small <YOUR_DATA_DIR> --quant --pts --lis --quant-method minmax
  • deit_small: model architecture, which can be replaced by deit_tiny, deit_base, vit_base, vit_large, swin_tiny, swin_small and swin_base.

  • --quant: whether to quantize the model.

  • --pts: whether to use Power-of-Two Scale Integer Layernorm.

  • --lis: whether to use Log-Integer-Softmax.

  • --quant-method: quantization methods of activations, which can be chosen from minmax, ema, percentile and omse.

Results on ImageNet

This paper employs several current post-training quantization strategies together with our methods, including MinMax, EMA , Percentile and OMSE.

  • MinMax uses the minimum and maximum values of the total data as the clipping values;

  • EMA is based on MinMax and uses an average moving mechanism to smooth the minimum and maximum values of different mini-batch;

  • Percentile assumes that the distribution of values conforms to a normal distribution and uses the percentile to clip. In this paper, we use the 1e-5 percentile because the 1e-4 commonly used in CNNs has poor performance in Vision Transformers.

  • OMSE determines the clipping values by minimizing the quantization error.

The following results are evaluated on ImageNet.

Method W/A/Attn Bits ViT-B ViT-L DeiT-T DeiT-S DeiT-B Swin-T Swin-S Swin-B
Full Precision 32/32/32 84.53 85.81 72.21 79.85 81.85 81.35 83.20 83.60
MinMax 8/8/8 23.64 3.37 70.94 75.05 78.02 64.38 74.37 25.58
MinMax w/ PTS 8/8/8 83.31 85.03 71.61 79.17 81.20 80.51 82.71 82.97
MinMax w/ PTS, LIS 8/8/4 82.68 84.89 71.07 78.40 80.85 80.04 82.47 82.38
EMA 8/8/8 30.30 3.53 71.17 75.71 78.82 70.81 75.05 28.00
EMA w/ PTS 8/8/8 83.49 85.10 71.66 79.09 81.43 80.52 82.81 83.01
EMA w/ PTS, LIS 8/8/4 82.57 85.08 70.91 78.53 80.90 80.02 82.56 82.43
Percentile 8/8/8 46.69 5.85 71.47 76.57 78.37 78.78 78.12 40.93
Percentile w/ PTS 8/8/8 80.86 85.24 71.74 78.99 80.30 80.80 82.85 83.10
Percentile w/ PTS, LIS 8/8/4 80.22 85.17 71.23 78.30 80.02 80.46 82.67 82.79
OMSE 8/8/8 73.39 11.32 71.30 75.03 79.57 79.30 78.96 48.55
OMSE w/ PTS 8/8/8 82.73 85.27 71.64 78.96 81.25 80.64 82.87 83.07
OMSE w/ PTS, LIS 8/8/4 82.37 85.16 70.87 78.42 80.90 80.41 82.57 82.45

Citation

If you find this repo useful in your research, please consider citing the following paper:

@misc{
    lin2021fqvit,
    title={FQ-ViT: Fully Quantized Vision Transformer without Retraining}, 
    author={Yang Lin and Tianyu Zhang and Peiqin Sun and Zheng Li and Shuchang Zhou},
    year={2021},
    eprint={2111.13824},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}
Code for CVPR2019 Towards Natural and Accurate Future Motion Prediction of Humans and Animals

Motion prediction with Hierarchical Motion Recurrent Network Introduction This work concerns motion prediction of articulate objects such as human, fi

Shuang Wu 85 Dec 11, 2022
《Geo Word Clouds》paper implementation

《Geo Word Clouds》paper implementation

Russellwzr 2 Jan 28, 2022
Face Library is an open source package for accurate and real-time face detection and recognition

Face Library Face Library is an open source package for accurate and real-time face detection and recognition. The package is built over OpenCV and us

52 Nov 09, 2022
Algo-burn - Script to configure an Algorand address as a "burn" address for one or more ASA tokens

Algorand Burn Address This is a simple script to illustrate how a "burn address"

GSD 5 May 10, 2022
Code and data form the paper BERT Got a Date: Introducing Transformers to Temporal Tagging

BERT Got a Date: Introducing Transformers to Temporal Tagging Satya Almasian*, Dennis Aumiller*, and Michael Gertz Heidelberg University Contact us vi

54 Dec 04, 2022
Google Landmark Recogntion and Retrieval 2021 Solutions

Google Landmark Recogntion and Retrieval 2021 Solutions In this repository you can find solution and code for Google Landmark Recognition 2021 and Goo

Vadim Timakin 5 Nov 25, 2022
OoD Minimum Anomaly Score GAN - Code for the Paper 'OMASGAN: Out-of-Distribution Minimum Anomaly Score GAN for Sample Generation on the Boundary'

OMASGAN: Out-of-Distribution Minimum Anomaly Score GAN for Sample Generation on the Boundary Out-of-Distribution Minimum Anomaly Score GAN (OMASGAN) C

- 8 Sep 27, 2022
Official PyTorch Implementation for InfoSwap: Information Bottleneck Disentanglement for Identity Swapping

InfoSwap: Information Bottleneck Disentanglement for Identity Swapping Code usage Please check out the user manual page. Paper Gege Gao, Huaibo Huang,

Grace Hešeri 56 Dec 20, 2022
Aesara is a Python library that allows one to define, optimize, and efficiently evaluate mathematical expressions involving multi-dimensional arrays.

Aesara is a Python library that allows one to define, optimize, and efficiently evaluate mathematical expressions involving multi-dimensional arrays.

Aesara 898 Jan 07, 2023
Rename Images with Auto Generated Neural Image Captions

Recaption Images with Generated Neural Image Caption Example Usage: Commandline: Recaption all images from folder /home/feng/Downloads/images to folde

feng wang 3 May 01, 2022
Code for "Diversity can be Transferred: Output Diversification for White- and Black-box Attacks"

Output Diversified Sampling (ODS) This is the github repository for the NeurIPS 2020 paper "Diversity can be Transferred: Output Diversification for W

50 Dec 11, 2022
Multiband spectro-radiometric satellite image analysis with K-means cluster algorithm

Multi-band Spectro Radiomertric Image Analysis with K-means Cluster Algorithm Overview Multi-band Spectro Radiomertric images are images comprising of

Chibueze Henry 6 Mar 16, 2022
This codebase is the official implementation of Test-Time Classifier Adjustment Module for Model-Agnostic Domain Generalization (NeurIPS2021, Spotlight)

Test-Time Classifier Adjustment Module for Model-Agnostic Domain Generalization This codebase is the official implementation of Test-Time Classifier A

47 Dec 28, 2022
code for paper -- "Seamless Satellite-image Synthesis"

Seamless Satellite-image Synthesis by Jialin Zhu and Tom Kelly. Project site. The code of our models borrows heavily from the BicycleGAN repository an

Light 14 Apr 05, 2022
Source codes of CenterTrack++ in 2021 ICME Workshop on Big Surveillance Data Processing and Analysis

MOT Tracked object bounding box association (CenterTrack++) New association method based on CenterTrack. Two new branches (Tracked Size and IOU) are a

36 Oct 04, 2022
TransZero++: Cross Attribute-guided Transformer for Zero-Shot Learning

TransZero++ This repository contains the testing code for the paper "TransZero++: Cross Attribute-guided Transformer for Zero-Shot Learning" submitted

Shiming Chen 6 Aug 16, 2022
Ground truth data for the Optical Character Recognition of Historical Classical Commentaries.

OCR Ground Truth for Historical Commentaries The dataset OCR ground truth for historical commentaries (GT4HistComment) was created from the public dom

Ajax Multi-Commentary 3 Sep 08, 2022
A TensorFlow Implementation of "Deep Multi-Scale Video Prediction Beyond Mean Square Error" by Mathieu, Couprie & LeCun.

Adversarial Video Generation This project implements a generative adversarial network to predict future frames of video, as detailed in "Deep Multi-Sc

Matt Cooper 704 Nov 26, 2022
A bunch of random PyTorch models using PyTorch's C++ frontend

PyTorch Deep Learning Models using the C++ frontend Gettting started Clone the repo 1. https://github.com/mrdvince/pytorchcpp 2. cd fashionmnist or

Vince 0 Jul 13, 2021
Ranger - a synergistic optimizer using RAdam (Rectified Adam), Gradient Centralization and LookAhead in one codebase

Ranger-Deep-Learning-Optimizer Ranger - a synergistic optimizer combining RAdam (Rectified Adam) and LookAhead, and now GC (gradient centralization) i

Less Wright 1.1k Dec 21, 2022