What are the Essential Factors in Crafting Effective Long Context Multi-Hop Instruction Datasets? Insights and Best Practices

TL;DR

This paper introduces the MIMG framework to generate high-quality, multi-hop instruction data for long context tasks, significantly improving model performance over existing synthetic data methods.

Contribution

The paper presents the Multi-agent Interactive Multi-hop Generation (MIMG) framework, enhancing synthetic data quality for long context multi-hop tasks and systematically analyzing data generation strategies.

Findings

High-quality, multi-hop data exceeds 85% in the proposed framework.

Synthetic data can outperform models trained on larger human-annotated datasets.

The MIMG framework improves long context understanding in language models.

Abstract

Recent advancements in large language models (LLMs) with extended context windows have significantly improved tasks such as information extraction, question answering, and complex planning scenarios. In order to achieve success in long context tasks, a large amount of work has been done to enhance the long context capabilities of the model through synthetic data. Existing methods typically utilize the Self-Instruct framework to generate instruction tuning data for better long context capability improvement. However, our preliminary experiments indicate that less than 35% of generated samples are multi-hop, and more than 40% exhibit poor quality, limiting comprehensive understanding and further research. To improve the quality of synthetic data, we propose the Multi-agent Interactive Multi-hop Generation (MIMG) framework, incorporating a Quality Verification Agent, a Single-hop Question Generation Agent, a Multiple Question Sampling Strategy, and a Multi-hop Question Merger Agent. This framework improves the data quality, with the proportion of high-quality, multi-hop, and diverse data exceeding 85%. Furthermore, we systematically investigate strategies for document selection, question merging, and validation techniques through extensive experiments across various models. Our findings show that our synthetic high-quality long-context instruction data significantly enhances model performance, even surpassing models trained on larger amounts of human-annotated data. Our code is available at: https://github.com/WowCZ/LongMIT.