Новое поколение нейросети от Сбера для создания фотореалистичных изображений по текстовому запросу. Благодаря обновленной архитектуре и улучшенному пониманию русского языка, Kandinsky 3.0 выводит отечественный генеративный ИИ на мировой уровень.
We present Kandinsky 3.0, a large-scale text-to-image generation model based on latent diffusion, continuing the series of text-to-image Kandinsky models and reflecting our progress to achieve higher quality and realism of image generation. Compared to previous versions of Kandinsky 2.x, Kandinsky 3.0 leverages a two times larger U-Net[1] backbone, a ten times larger text encoder and removes diffusion mapping. We describe the architecture of the model, the data collection procedure, the training technique, and the production system of user interaction. We focus on the key components that, as we have identified as a result of a large number of experiments, had the most significant impact on improving the quality of our model compared to the others. By our side-by-side comparisons, Kandinsky becomes better in text understanding and works better on specific domains. The project is available at https://ai-forever.github.io/Kandinsky-3