Self-Instruct: Aligning Language Models with Self-Generated Instructions

TL;DR

Self-Instruct is a method that improves language models' ability to follow instructions by generating and filtering its own training data, reducing reliance on human annotations and achieving performance comparable to models trained with private data.

Contribution

The paper introduces Self-Instruct, a novel self-supervised approach for instruction tuning that leverages model-generated data to enhance instruction-following capabilities.

Findings

33% improvement on Super-NaturalInstructions

Performs comparably to InstructGPT-001

Outperforms existing public instruction datasets

Abstract

Large "instruction-tuned" language models (i.e., finetuned to respond to instructions) have demonstrated a remarkable ability to generalize zero-shot to new tasks. Nevertheless, they depend heavily on human-written instruction data that is often limited in quantity, diversity, and creativity, therefore hindering the generality of the tuned model. We introduce Self-Instruct, a framework for improving the instruction-following capabilities of pretrained language models by bootstrapping off their own generations. Our pipeline generates instructions, input, and output samples from a language model, then filters invalid or similar ones before using them to finetune the original model. Applying our method to the vanilla GPT3, we demonstrate a 33% absolute improvement over the original model on Super-NaturalInstructions, on par with the performance of InstructGPT-001, which was trained with private user data and human annotations. For further evaluation, we curate a set of expert-written instructions for novel tasks, and show through human evaluation that tuning GPT3 with Self-Instruct outperforms using existing public instruction datasets by a large margin, leaving only a 5% absolute gap behind InstructGPT-001. Self-Instruct provides an almost annotation-free method for aligning pre-trained language models with instructions, and we release our large synthetic dataset to facilitate future studies on instruction tuning. Our code and data are available at https://github.com/yizhongw/self-instruct.