My entry for the AWS DL1 Challenge by Devpost

Author: Remco van Akker

Project name

Using a Deep Convolutional GAN Network to Generate Synthetic Data for Medical Purposes based on Real-Life Data

Usage instructions

Train with HPU

If you want to make use of HPU, run the main script with the following command:
python3 dcgan.py --use_hpu

Distributed Training with Horovod

If you want to make use of Distributed Training via Horovod (up to 8 Gaudi cores), run the main script with the following command:
mpirun -np 8 python3 dcgan.py --use_hpu

Restore Checkpoint

You can resume from the latest checkpoint by running the script with the following command: python3 dcgan.py --use_hpu --restore

Download Dataset

You don't have to download anything manual, just replace the kaggle.json file with your own and the application will use your Kaggle API Key to download the dataset.

Dataset source: https://www.kaggle.com/andrewmvd/ocular-disease-recognition-odir5k