GPV-I

Researchers have proposed task-agnostic architectures for image classification tasks and language tasks. New work proposes a single architecture for vision-language tasks.