The implementation of "Prismer: A Vision-Language Model with Multi-Task Experts". - View it on GitHub