Kandinsky LabruDALL-E
ruDALL-E Emojich
Emojich fine-tunes the 1.3B Malevich model for emoji-style generation. Its task-specific training uses 2,749 text-emoji pairs.
Keep the language model's generalization while changing image style
The documented fine-tuning freezes feed-forward and self-attention layers and changes the weighting of the image-codebook loss. This is a trained derivative, not a prompt preset.ai-forever/rudalle-Emojich — model card and files
Emoji generation and background removal are different stages
Practical guidance. The generated image may still need selection and preparation for a sticker slot. A transparent sticker conversion is separate from generating its visual content.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | ai-forever/rudalle-Emojich |
| Base model | ruDALL-E Malevich XLai-forever/rudalle-Emojich — model card and files |
| Code license | Apache 2.0; the model card has no separate license label.ruDALL-E: original release repository |
Related models
- ruDALL-E Malevich XL
Kandinsky Lab · ruDALL-E
A 1.3B Russian text-to-image model trained by Sber AI and SberDevices using the DALL-E approach.
Launch date not confirmedDownloadable