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}
}
Simple helper library to convert a collection of numpy data to tfrecord, and build a tensorflow dataset from the tfrecord.

numpy2tfrecord Simple helper library to convert a collection of numpy data to tfrecord, and build a tensorflow dataset from the tfrecord. Installation

Ryo Yonetani 2 Jan 16, 2022
[ICCV'21] UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction

UNISURF: Unifying Neural Implicit Surfaces and Radiance Fields for Multi-View Reconstruction Project Page | Paper | Supplementary | Video This reposit

331 Dec 28, 2022
Official implementation of the PICASO: Permutation-Invariant Cascaded Attentional Set Operator

PICASO Official PyTorch implemetation for the paper PICASO:Permutation-Invariant Cascaded Attentive Set Operator. Requirements Python 3 torch = 1.0 n

Samira Zare 0 Dec 23, 2021
Official PyTorch implementation of SyntaSpeech (IJCAI 2022)

SyntaSpeech: Syntax-Aware Generative Adversarial Text-to-Speech | | | | 中文文档 This repository is the official PyTorch implementation of our IJCAI-2022

Zhenhui YE 116 Nov 24, 2022
Rule Extraction Methods for Interactive eXplainability

REMIX: Rule Extraction Methods for Interactive eXplainability This repository contains a variety of tools and methods for extracting interpretable rul

Mateo Espinosa Zarlenga 21 Jan 03, 2023
Object detection and instance segmentation toolkit based on PaddlePaddle.

Object detection and instance segmentation toolkit based on PaddlePaddle.

9.3k Jan 02, 2023
Sparse-dense operators implementation for Paddle

Sparse-dense operators implementation for Paddle This module implements coo, csc and csr matrix formats and their inter-ops with dense matrices. Feel

北海若 3 Dec 17, 2022
Code for our CVPR 2022 Paper "GEN-VLKT: Simplify Association and Enhance Interaction Understanding for HOI Detection"

GEN-VLKT Code for our CVPR 2022 paper "GEN-VLKT: Simplify Association and Enhance Interaction Understanding for HOI Detection". Contributed by Yue Lia

Yue Liao 47 Dec 04, 2022
A novel pipeline framework for multi-hop complex KGQA task. About the paper title: Improving Multi-hop Embedded Knowledge Graph Question Answering by Introducing Relational Chain Reasoning

Rce-KGQA A novel pipeline framework for multi-hop complex KGQA task. This framework mainly contains two modules, answering_filtering_module and relati

金伟强 -上海大学人工智能小渣渣~ 16 Nov 18, 2022
Colar: Effective and Efficient Online Action Detection by Consulting Exemplars, CVPR 2022.

Colar: Effective and Efficient Online Action Detection by Consulting Exemplars This repository is the official implementation of Colar. In this work,

LeYang 246 Dec 13, 2022
Synthesizing Long-Term 3D Human Motion and Interaction in 3D in CVPR2021

Long-term-Motion-in-3D-Scenes This is an implementation of the CVPR'21 paper "Synthesizing Long-Term 3D Human Motion and Interaction in 3D". Please ch

Jiashun Wang 76 Dec 13, 2022
In this project I played with mlflow, streamlit and fastapi to create a training and prediction app on digits

Fastapi + MLflow + streamlit Setup env. I hope I covered all. pip install -r requirements.txt Start app Go in the root dir and run these Streamlit str

76 Nov 23, 2022
An implementation of Equivariant e2 convolutional kernals into a convolutional self attention network, applied to radio astronomy data.

EquivariantSelfAttention An implementation of Equivariant e2 convolutional kernals into a convolutional self attention network, applied to radio astro

2 Nov 09, 2021
The code of “Similarity Reasoning and Filtration for Image-Text Matching” [AAAI2021]

SGRAF PyTorch implementation for AAAI2021 paper of “Similarity Reasoning and Filtration for Image-Text Matching”. It is built on top of the SCAN and C

Ronnie_IIAU 149 Dec 22, 2022
An atmospheric growth and evolution model based on the EVo degassing model and FastChem 2.0

EVolve Linking planetary mantles to atmospheric chemistry through volcanism using EVo and FastChem. Overview EVolve is a linked mantle degassing and a

Pip Liggins 2 Jan 17, 2022
Official implementation of the NeurIPS 2021 paper Online Learning Of Neural Computations From Sparse Temporal Feedback

Online Learning Of Neural Computations From Sparse Temporal Feedback This repository is the official implementation of the NeurIPS 2021 paper Online L

Lukas Braun 3 Dec 15, 2021
A simple editor for captions in .SRT file extension

WaySRT A simple editor for captions in .SRT file extension The program doesn't use any external dependecies, just run: python way_srt.py {file_name.sr

Gustavo Lopes 3 Nov 16, 2022
"Projelerle Yapay Zeka Ve Bilgisayarlı Görü" Kitabımın projeleri

"Projelerle Yapay Zeka Ve Bilgisayarlı Görü" Kitabımın projeleri Bu Github Reposundaki tüm projeler; kaleme almış olduğum "Projelerle Yapay Zekâ ve Bi

Ümit Aksoylu 4 Aug 03, 2022
PSML: A Multi-scale Time-series Dataset for Machine Learning in Decarbonized Energy Grids

PSML: A Multi-scale Time-series Dataset for Machine Learning in Decarbonized Energy Grids The electric grid is a key enabling infrastructure for the a

Texas A&M Engineering Research 19 Jan 07, 2023
Additional environments compatible with OpenAI gym

Decentralized Control of Quadrotor Swarms with End-to-end Deep Reinforcement Learning A codebase for training reinforcement learning policies for quad

Zhehui Huang 40 Dec 06, 2022