Natural Instructions: Benchmarking Generalization to New Tasks from Natural Language Instructions

Open in new window