Kandinsky LabKandinsky 2
Kandinsky 2.0
Kandinsky 2.0 uses latent diffusion and two multilingual text encoders for image generation.
Distinguish the generator from its encoders
The text encoders are mCLIP-XLMR and mT5-small. Their parameters are additional to the diffusion U-Net.ai-forever/Kandinsky_2.0 — model card and files
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | ai-forever/Kandinsky_2.0 |
| Diffusion U-Net | 1.2B parametersai-forever/Kandinsky_2.0 — model card and files |
| Weight license | Apache 2.0ai-forever/Kandinsky_2.0 — model card and files |