Denoising VAE

This project was written for our bachelor's degree project in computer science at KTH in Stockholm, Sweden.

Getting started

Details will be published later.

Set up remote Docker environment for GPU-accelerated training

This installation guide is based on this article from Pendragon AI.

The only modification we made was to replace these commands:

docker pull tensorflow/tensorflow:2.14.0-gpu

docker run --name my_tensorflow -it --gpus all tensorflow/tensorflow:2.14.0-gpu bash

With these instead:

docker pull tensorflow/tensorflow:2.16.1-gpu

docker run --name my_tensorflow -it --gpus all tensorflow/tensorflow:2.16.1-gpu bash

To get the address of the container, run:

docker-compose logs image-ldm | grep -o 'http://127.0.0.1:8888/?token=[^ ]*' | tail -n1

Licensing

Portions of this project are modifications of the Keras Team's example of a Variational Autoencoder by fchollet and used according to terms described in the Apache 2.0 License. A copy of the Apache License is included in this repository in the LICENSE file.