huggingface/pytorch-pretrained-BigGAN

Health v2 · maintenance only

Adoption not part of the score

1039 stars · 178 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A PyTorch reimplementation of DeepMind's BigGAN generator with pretrained weights at 128, 256, and 512 pixel resolutions, plus scripts to convert the original TensorFlow Hub models. It lets users generate high-fidelity images from class labels and noise samples with a simple pip-installable API.

Use cases

• generate images from imagenet class names with a pretrained GAN
• run BigGAN in PyTorch instead of TensorFlow
• convert DeepMind BigGAN weights from TF Hub to PyTorch
• sample images with truncation trick noise
• experiment with GAN latent space interpolation

When to choose

• you want pretrained BigGAN image generation in PyTorch
• you need an op-for-op port of the TensorFlow BigGAN generator
• you want a simple pip-installable GAN for class-conditional image synthesis

When to avoid

• you need the BigGAN discriminator
• you want to train a GAN from scratch
• you need a maintained library - the repo has not been updated since 2021
• you want modern diffusion-based image generation

Facets

library · maturity maintenance

machine-learning deep-learning image-processing machine-learning deep-learning computer-vision image-processing python gan biggan pytorch pretrained-models image-generation generative-models

Member repositories

For agents

markdown · JSON · MCP: product_card(name="huggingface/pytorch-pretrained-BigGAN")