Are Prompt-based Models Clueless?

Pride Kavumba, Ryo Takahashi, Yusuke Oda

Finetuning large pre-trained language models with a task-specific head has advanced the state-of-the-art on many natural language understanding benchmarks. However, models with a t…