How Multitask Learning Boosts Zero-Shot Generalization

Описание: What if zero-shot generalization was directly induced by explicit multitask learning? In this chat with Alex Ratner, Steve Bach discusses how the researchers from Big Science Workshop build trained models that able to attain strong zero-shot performance on several standard datasets, often outperforming models up to 16x their size.

More related videos:    • Foundation Models: The Future of AI  

#brownuniversity #machinelearning #zeroshotlearning