Train model with dataset in PyTorch:
model:- VGG16
- ResNet18
- DLA
- WRN-28-10
dataset:- CIFAR10
- CIFAR100
- SVHN
- MNIST
- FashionMNIST
Clone this repo, then install all dependencies:
pip install -r requirements.txt
The code was tested with Python 3.6.9.
| Model | CIFAR10 | CIFAR100 | SVHN | MNIST | FashionMNIST |
|---|---|---|---|---|---|
| VGG16 | 94.32% | 74.87% | 96.17% | 99.64% | 93.88% |
| ResNet18 | 95.69% | 78.24% | 96.64% | 99.70% | 94.10% |
| DLA | 95.87% | 78.62% | 96.83% | 99.63% | 94.06% |
| WRN-28-10 | 95.94% | 81.33% | 97.02% | 99.71% | 94.36% |
Training configurations, checkpoint models, and output logs can be found here.