Improving Attributed Long-form Question Answering with Intent Awareness
For researchers and practitioners using LLMs for knowledge-intensive report generation, this work offers a method to improve output quality by incorporating authorial intents.
The authors propose enhancing LLMs' intent awareness to improve long-form question answering, achieving +2.9 and +12.3 absolute point improvements over baselines for large and small models, respectively, in scientific report generation.
Large language models (LLMs) are increasingly being used to generate comprehensive, knowledge-intensive reports. However, while these models are trained on diverse academic papers and reports, they are not exposed to the reasoning processes and intents that guide authors in crafting these documents. We hypothesize that enhancing a model's intent awareness can significantly improve the quality of generated long-form reports. We develop and employ structured, tag-based schemes to better elicit underlying implicit intents to write or cite. We demonstrate that these extracted intents enhance both zero-shot generation capabilities in LLMs and enable the creation of high-quality synthetic data for fine-tuning smaller models. Our experiments reveal improved performance across various challenging scientific report generation tasks, with an average improvement of +2.9 and +12.3 absolute points for large and small models over baselines, respectively. Furthermore, our analysis illuminates how intent awareness enhances model citation usage and substantially improves report readability.