chanmuzi
<LLM, Zero-shot> [T0] Multitask Prompted Training Enables Zero-Shot Task Generalization (2022.03)