The ruDALL-E neural network generates pictures based on descriptions in foreign languages
Within a week after the release of ruDALL-E, users around the world generated more than 3 million images, using various machine translation systems to generate Russian-language queries. When entering text, the model independently detects the input language and generates the corresponding image. The prototype for the creation of ruDALL-E was a neural network DALL-E for the English language, which was first introduced by OpenAI in 2021. Researchers from the American company did not put the model in the public domain, and limited to a general description of the architecture and an impressive set of examples of the model, hand-picked. The model exists in two versions: ruDALL-E XL, with 1.3 billion parameters, and ruDALL-E XXL with 12 billion parameters.