Skip to content

mszulc913/dvae-pytorch

Repository files navigation

Discrete Variational Autoencoder in PyTorch

Experiments with Discrete Variational Autoencoder in PyTorch.

Installation

To install the pinned versions of the dependencies (versions I used to conduct the experiments) run the following:

pip install --upgrade pip
pip install -r requirements.txt -r dev-requirements.txt

If you want to install newer versions of the dependencies you can just run

pip install ".[dev]"

Note: If you want to use a GPU/TPU acceleration, make sure you have a compatible version of CUDA installed on your system.

Run

You can run an example experiment with CIFAR10 using the train.py script:

python -m dvae_pytorch.training.train --config configs/dvae_cifar10.yaml

You can track your experiment with tensorboard:

tensorboard --logdir tb_logs

Tests

To run the tests execute the following command:

python -m pytest

Releases

No releases published

Packages

No packages published

Languages