This repository contains the official PyTorch implementation for the ECCV 2022 paper "Patch Similarity Aware Data-Free Quantization for Vision Transformers". To the best of our knowledge, this is the first work on data-free quantization for vision transformers. Below are instructions for reproducing the results.
- To install PSAQ-ViT and develop locally:
git clone https://github.com/zkkli/PSAQ-ViT.git
cd PSAQ-ViT- You can quantize and evaluate a single model using the following command:
python test_quant.py [--model] [--dataset] [--w_bit] [--a_bit] [--mode]
optional arguments:
--model: Model architecture, the choises can be:
deit_tiny, deit_small, deit_base, swin_tiny, and swin_small.
--dataset: Path to ImageNet dataset.
--w_bit: Bit-precision of weights, default=8.
--a_bit: Bit-precision of activation, default=8.
--mode: Mode of calibration data,
0: Generated fake data (PSAQ-ViT)
1: Gaussian noise
2: Real data- Example: Quantize DeiT-B with generated fake data (PSAQ-ViT).
python test_quant.py --model deit_base --dataset <YOUR_DATA_DIR> --mode 0- Example: Quantize DeiT-B with Gaussian noise.
python test_quant.py --model deit_base --dataset <YOUR_DATA_DIR> --mode 1- Example: Quantize DeiT-B with Real data.
python test_quant.py --model deit_base --dataset <YOUR_DATA_DIR> --mode 2Below are the experimental results of our proposed PSAQ-ViT that you should get on ImageNet dataset using an RTX 3090 GPU.
| Model | Prec. | Top-1(%) | Prec. | Top-1(%) |
|---|---|---|---|---|
| DeiT-T (72.21) | W4/A8 | 65.57 | W8/A8 | 71.56 |
| DeiT-S (79.85) | W4/A8 | 73.23 | W8/A8 | 76.92 |
| DeiT-B (81.85) | W4/A8 | 77.05 | W8/A8 | 79.10 |
| Swin-T (81.35) | W4/A8 | 71.79 | W8/A8 | 75.35 |
| Swin-S (83.20) | W4/A8 | 75.14 | W8/A8 | 76.64 |
We appreciate it if you would please cite the following paper if you found the implementation useful for your work:
@inproceedings{li2022psaqvit,
title={Patch Similarity Aware Data-Free Quantization for Vision Transformers},
author={Li, Zhikai and Ma, Liping and Chen, Mengjuan and Xiao, Junrui and Gu, Qingyi},
booktitle={European Conference on Computer Vision},
pages={154--170},
year={2022}
}