lucidrains/imagen-pytorch
A PyTorch implementation of Google's Imagen, a cascaded diffusion model for text-to-image synthesis.

Velocity · 7d
+5.7
★ / day
Trend
→steady
star history
This repository provides a PyTorch implementation of Google’s Imagen, a state-of-the-art text-to-image neural network. It uses cascaded DDPM conditioned on text embeddings from a pretrained T5 model, with dynamic clipping for classifier-free guidance and memory-efficient UNet design. The model enables generating high-quality images from text prompts using diffusion-based generative modeling.