ajbrock/BigGAN-PyTorch
A PyTorch implementation of BigGAN for training large-scale generative adversarial networks on high-resolution image synthesis.
Not currently ranked — collecting fresh signals.
star history
This repository provides an unofficial PyTorch implementation of BigGAN based on the paper Large Scale GAN Training for High Fidelity Natural Image Synthesis. It enables training of BigGAN models on ImageNet with support for 4-8 GPU configurations using gradient accumulation to achieve effective mega-batch sizes. The code handles dataset preparation, inception score calculation, and FID evaluation, with pretrained models available for download.
Frequently asked
- What is ajbrock/BigGAN-PyTorch?
- A PyTorch implementation of BigGAN for training large-scale generative adversarial networks on high-resolution image synthesis.
- Is BigGAN-PyTorch open source?
- Yes — ajbrock/BigGAN-PyTorch is open source, released under the MIT license.
- What language is BigGAN-PyTorch written in?
- ajbrock/BigGAN-PyTorch is primarily written in Python.
- How popular is BigGAN-PyTorch?
- ajbrock/BigGAN-PyTorch has 2.9k stars on GitHub.
- Where can I find BigGAN-PyTorch?
- ajbrock/BigGAN-PyTorch is on GitHub at https://github.com/ajbrock/BigGAN-PyTorch.