Multitask Learning Architectures

Daniel and Hattie discuss the concept of multitask learning and its architectural implications. They explore the idea of using task-specific masks instead of separate heads in a network, highlighting the potential benefits and applications of this approach.