This repository contains a Jupyter Notebook that demonstrates the reimplementation of the paper "Top-k Training of GANs: Improving GAN Performance by Throwing Away Bad Samples" in PyTorch. The paper introduces a novel training method for Generative Adversarial Networks (GANs) that focuses on discarding poor-quality samples during the training process to enhance the overall performance.
If you use this code or find it helpful, please consider citing the original paper
To get started with this code, follow these steps:
- Clone this repository:
git clone https://github.com/LucasDedieu/Topk-Training.git
- Install the required dependencies. We recommend using a virtual environment:
pip install -r requirements.txt
-
Launch the Jupyter Notebook:
-
Open the
TopkTraining.ipynbnotebook and execute the code cells to observe the reimplementation and experiment with different settings.
The Jupyter Notebook provides a step-by-step explanation of the top-k training method for GANs described in the paper. It includes code for training and evaluating the GAN model using the CIFAR dataset. The results, including quantitative metrics and generated samples, will be displayed during the execution of the notebook.
Contributions to this project are welcome. If you find any issues or want to propose enhancements, please submit a pull request or open an issue in this repository.