JackXu2333 / VGG11_MNIST

VGG-11 Model Tested on MNIST

VGG-11 Architecture Tested on MNIST

Introduction

This repository contains PyTorch bottom up implementation of VGG-11 model on pyTorch MNIST dataset with the following architecture.

- Conv(001, 064, 3, 1, 1) - BatchNorm(064) - ReLU - MaxPool(2, 2) 
- Conv(064, 128, 3, 1, 1) - BatchNorm(128) - ReLU - MaxPool(2, 2) 
- Conv(128, 256, 3, 1, 1) - BatchNorm(256) - ReLU 
- Conv(256, 256, 3, 1, 1) - BatchNorm(256) - ReLU - MaxPool(2, 2) 
- Conv(256, 512, 3, 1, 1) - BatchNorm(512) - ReLU 
- Conv(512, 512, 3, 1, 1) - BatchNorm(512) - ReLU - MaxPool(2, 2) 
- Conv(512, 512, 3, 1, 1) - BatchNorm(512) - ReLU 
- Conv(512, 512, 3, 1, 1) - BatchNorm(512) - ReLU - MaxPool(2, 2) 
- FC(0512, 4096) - ReLU - Dropout(0.5) 
- FC(4096, 4096) - ReLU - Dropout(0.5) 
- FC(4096, 10)

Prerequisites

The models are implemented with config:

Python==3.8.3
torch==1.9.1
torchvision==0.10.1
matplotlib==3.4.3
numpy==1.21.2

Note that the test file automatically enable CUDA when available.

MNIST dataset

The MNIST dataset are used in training and testing of VGG-11.

Training

Vanilla model

You can find the detail of training and testing the VGG-11 model on default MNIST dataset via model_training.ipynb

Data Augmented model

You can find the detail of training and testing the VGG-11 model on default MNIST dataset via model_training_augmented.ipynb

Acknowledgment

If you find the model from this repository helpful, please cite the original paper:

@article{simonyan2014very,
  title={Very deep convolutional networks for large-scale image recognition},
  author={Simonyan, Karen and Zisserman, Andrew},
  journal={arXiv preprint arXiv:1409.1556},
  year={2014}
}