Topic

ShengShu Technology's Exploration and Practice in Large Multimodal Models

ShengShu Technology has been committed to the research and application of multimodal large models, and we have some of our own cutting-edge judgments and deep thoughts on the industry's chokepoints and ultimate forms. In the field of multimodal large model, how to unify the input and output of multimodal is a key issue that has been pending in the industry. For the unification of multimodal representation, through technical research and exploration and practice, we have observed some potential solutions that can unify video, audio, text and other modalities. In addition, we have very much thought about the ultimate form of multimodal big models. If the current video generation plays more of a “rendering” role, then the ultimate form of multimodal large model will play the role of intelligent brain in the future, which is similar to the o1 in the field of video generation, with strong reasoning ability, can understand the user's intention and intelligently guide their work and life. Outline: 1. ShengShu Technology's practice and technology exploration in the field of multimodal big model. 2、Some thoughts on the unified representation of multimodal big model field. 3、The ultimate form of multimodal big model