Help me write a poem: Instruction Tuning as a Vehicle for Collaborative Poetry Writing

TL;DR

This paper introduces CoPoet, a system that uses instruction-tuned large language models to facilitate collaborative poetry writing, enabling users to craft poems on diverse topics with improved quality and user satisfaction.

Contribution

We develop and evaluate CoPoet, a novel instruction-tuned LLM system that enhances collaborative poetry creation and satisfies unseen instructions, outperforming existing models in user preference.

Findings

Users successfully wrote diverse poems with CoPoet.

Collaboratively written poems were preferred over non-assisted ones.

The system performs competitively with InstructGPT on unseen instructions.

Abstract

Recent work in training large language models (LLMs) to follow natural language instructions has opened up exciting opportunities for natural language interface design. Building on the prior success of LLMs in the realm of computer-assisted creativity, we aim to study if LLMs can improve the quality of user-generated content through collaboration. We present CoPoet, a collaborative poetry writing system. In contrast to auto-completing a user's text, CoPoet is controlled by user instructions that specify the attributes of the desired text, such as Write a sentence about `love' or Write a sentence ending in `fly'. The core component of our system is a language model fine-tuned on a diverse collection of instructions for poetry writing. Our model is not only competitive with publicly available LLMs trained on instructions (InstructGPT), but is also capable of satisfying unseen compositional instructions. A study with 15 qualified crowdworkers shows that users successfully write poems with CoPoet on diverse topics ranging from Monarchy to Climate change. Further, the collaboratively written poems are preferred by third-party evaluators over those written without the system.