Adapting Language Models for Zero-shot Learning by Meta-tuning on Dataset and Prompt Collections | Read Paper on Bytez