Finetuned Language Models are Zero-Shot Learners
Jason Wei u00b7 Maarten Bosma u00b7 Vincent Zhao u00b7 Kelvin Guu u00b7 Wei Yu u00b7 Brian Lester u00b7 Nan Du u00b7 Andrew Dai u00b7 Quoc V Le
This paper explores a simple method for improving the zero-shot learning abilities of language models. We show that instruction tuningu2014finetuning language models on a collection of datasets described via instructionsu2014substantially improves zero-shot performance on unseen tasks. We take a 137B parameter pretrained language model and instruction tune it on over 60 NLP datasets verbalized via natural language instruction templates. We evaluate this instruction-tuned model, which we call FLAN, on unseen task types. FLAN substantially improves the performance of its unmodified counterpart and surpasses zero-shot 175B GPT-3 on 20 of 25 datasets that we evaluate. FLAN even outperforms few-shot GPT-3 by a large margin on ANLI, RTE, BoolQ, AI2-ARC, OpenbookQA, and StoryCloze. Ablation studies reveal that number of finetuning datasets, model scale, and natural language instructions are key to the success of instruction tuning.
