Explore the power of Conditional Variational Autoencoders (CVAEs) through this implementation trained on the MNIST dataset to generate handwritten digit images based on class labels. Utilizing the robust and versatile PyTorch library, this project showcases a straightforward yet effective approach to conditional generative modeling.
- Thorough implementation of CVAE as delineated in the
seminal paper(https://arxiv.org/abs/1312.6114). - Streamlined training and evaluation workflows for the MNIST dataset.
- Real-time visualization of generated digit images during training, alongside epoch-wise saved image outputs.
- Modular and well-commented code for ease of customization, extension, and detailed understanding.
Python 3.x(https://www.python.org/downloads/)PyTorch(https://pytorch.org/get-started/locally/)Torchvision(https://pytorch.org/vision/stable/index.html)
-
Clone the repository:
git clone https://github.com/unnir/cVAE.git cd cVAE -
(Optional) It's advisable to create a virtual environment to manage dependencies:
python -m venv env source env/bin/activate # On Windows, use +env\Scripts\activate+
-
Install the necessary libraries:
pip install torch torchvision
- Make sure the MNIST dataset is downloaded in the +../data+ directory, or modify the path in the script to your dataset location.
- Run the main script to initiate training and evaluation of the CVAE model:
python main.py
- Real-time console output of training and testing loss values.
- Generated digit images during training saved as +reconstruction_.png+ and sampled images from the latent space saved as +sample_.png+ in the current directory.
Feel free to fork the project, open issues, and submit Pull Requests. For major changes, please open an issue first to discuss what you would like to change.
MIT(https://choosealicense.com/licenses/mit/)
Auto-Encoding Variational Bayes(https://arxiv.org/abs/1312.6114) - Kingma and Welling's paper on Variational Autoencoders.PyTorch Documentation(https://pytorch.org/docs/stable/index.html) - Official documentation and tutorials.MNIST dataset(http://yann.lecun.com/exdb/mnist/) - Source of the MNIST dataset used in this project.
Vadim Borisov
