Instruction-Guided Poetry Generation in Arabic and Its Dialects

2026-04-30

AI Summary

This work addresses the limited capability of existing large language models in generating controllable Arabic poetry, as they predominantly focus on analytical tasks and lack support for stylistic and metrical control. We propose the first instruction-guided poetry generation framework tailored for Modern Standard Arabic and multiple dialects, accompanied by the release of the first high-quality, multitask instruction dataset supporting composition, continuation, revision, and analysis. Leveraging supervised fine-tuning with carefully designed control-condition templates and evaluating through a combination of automatic metrics and native-speaker human assessments, our approach significantly outperforms baseline models, effectively enabling high-quality, instruction-driven controllable poetry generation.

Abstract

Poetry has long been a central art form for Arabic speakers, serving as a powerful medium of expression and cultural identity. While modern Arabic speakers continue to value poetry, existing research on Arabic poetry within Large Language Models (LLMs) has primarily focused on analysis tasks such as interpretation or metadata prediction, e.g., rhyme schemes and titles. In contrast, our work addresses the practical aspect of poetry creation in Arabic by introducing controllable generation capabilities to assist users in writing poetry. Specifically, we present a large-scale, carefully curated instruction-based dataset in Modern Standard Arabic (MSA) and various Arabic dialects. This dataset enables tasks such as writing, revising, and continuing poems based on predefined criteria, including style and rhyme, as well as performing poetry analysis. Our experiments show that fine-tuning LLMs on this dataset yields models that can effectively generate poetry that is aligned with user requirements, based on both automated metrics and human evaluation with native Arabic speakers. The data and the code are available at https://github.com/mbzuai-nlp/instructpoet-ar

Problem

• Arabic poetry
• instruction-guided generation
• controllable text generation
• Arabic dialects
• poetry creation

Innovation

• instruction-based dataset
• controllable poetry generation
• Arabic dialects
• large language models
• fine-tuning