CM3leon is a generative AI tool for creating text from images and vice versa that is characterized by high-quality results despite being trained and developed with five times less computational power than other AI-based transformation methods.
CM3leon’s technology can generate text and images from arbitrary sequences of other content consisting of text and images. It is also the first multimodal language model trained with a recipe adapted from text-only language models.
The system can also edit a pre-existing image from a text indication, and in this case Meta uses as an example modifications made to the famous Vermeer painting “The Girl with the Pearl Earring” from text suggestions such as “What would he look like if he were a man with a beard?” or “Put some sunglasses on him”.
https://www.reasonwhy.es/actualidad/cm3leon-nueva-inteligencia-artificial-meta-genera-textos-imagenes-menos-recursos
