ai-forever/Kandinsky-2

Kandinsky 2 — multilingual text2image latent diffusion model

47
/ 100
Emerging

Built on a two-stage diffusion architecture with a CLIP-based prior that maps text embeddings to image embeddings before latent decoding, enabling text-guided image generation, inpainting, and image fusion. Version 2.2 upgrades to CLIP-ViT-G encoding and adds ControlNet support for spatial control over generation. Available via PyTorch with pretrained checkpoints on Hugging Face, supporting variable resolutions and LoRA fine-tuning.

2,817 stars. No commits in the last 6 months.

Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 21 / 25

How are scores calculated?

Stars

2,817

Forks

315

Language

Jupyter Notebook

License

Apache-2.0

Last pushed

May 01, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/ai-forever/Kandinsky-2"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.