lucidrains/lightweight-gan
PyTorch implementation of a lightweight GAN model for training high-resolution image generators on a single GPU within days.
Not currently ranked — collecting fresh signals.
star history
This repository implements the ’lightweight’ GAN architecture proposed in ICLR 2021. The model uses skip-layer excitation in the generator combined with autoencoding self-supervised learning in the discriminator to achieve high-quality 1024x1024 image generation with reduced computational requirements. It trains efficiently on a single GPU and includes configurable augmentation strategies for low-data settings.
Frequently asked
- What is lucidrains/lightweight-gan?
- PyTorch implementation of a lightweight GAN model for training high-resolution image generators on a single GPU within days.
- Is lightweight-gan open source?
- Yes — lucidrains/lightweight-gan is open source, released under the MIT license.
- What language is lightweight-gan written in?
- lucidrains/lightweight-gan is primarily written in Python.
- How popular is lightweight-gan?
- lucidrains/lightweight-gan has 1.7k stars on GitHub.
- Where can I find lightweight-gan?
- lucidrains/lightweight-gan is on GitHub at https://github.com/lucidrains/lightweight-gan.