Instruction Induction: From Few Examples to Natural Language Task Descriptions

TL;DR

This paper demonstrates that large language models can infer natural language task descriptions from few examples, enabling explicit instruction induction which improves task understanding and performance.

Contribution

The paper introduces the instruction induction challenge, a new dataset, and an evaluation metric, showing that instruction generation ability emerges in sufficiently large and aligned models like InstructGPT.

Findings

InstructGPT achieves 65.7% of human performance in instruction execution.

GPT-3 achieves only 9.8% of human performance.

Instruction induction emerges in large, instruction-following models.

Abstract

Large language models are able to perform a task by conditioning on a few input-output demonstrations - a paradigm known as in-context learning. We show that language models can explicitly infer an underlying task from a few demonstrations by prompting them to generate a natural language instruction that fits the examples. To explore this ability, we introduce the instruction induction challenge, compile a dataset consisting of 24 tasks, and define a novel evaluation metric based on executing the generated instruction. We discover that, to a large extent, the ability to generate instructions does indeed emerge when using a model that is both large enough and aligned to follow instructions; InstructGPT achieves 65.7% of human performance in our execution-based metric, while the original GPT-3 model reaches only 9.8% of human performance. This surprising result suggests that instruction induction might be a viable learning paradigm in and of itself, where instead of fitting a set of latent continuous parameters to the data, one searches for the best description in the natural language hypothesis space.