Papernotes: SELF-INSTRUCT: Aligning Language Model with Self Generated Instructions

Wang, Yizhong, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A. Smith, Daniel Khashabi, and Hannaneh Hajishirzi. “Self-Instruct: Aligning Language Model with Self Generated Instructions.” arXiv preprint arXiv:2212.10560 (2022). https://arxiv.org/abs/2212.10560

• A framework for improving the instruction following capabilities of pretrained language models by bootstrapping off its own generations.
• Generates instruction, input and output samples from a language model, then prunes them before using them to finetune the original model.
• On GPT3, 33% improvement over original model on SUPERNATURALINSTRUCTIONS dataset (Wang et al, 2022)
• InstructGPT001 is trained with private user data and human annotations.
• Data and code available at https://github.com/yizhongw/self-instruct
• A distinct motivation for SELF-INSTRUCT is to bootstrap new task definitions that may not have been defined before by any NLP practitioner.