FanOutQA: A Multi-Hop, Multi-Document Question Answering Benchmark for Large Language Models

TL;DR

FanOutQA introduces a new dataset and benchmark for evaluating large language models on complex multi-hop, multi-document reasoning questions, revealing current models' limitations in inter-document reasoning.

Contribution

The paper presents FanOutQA, a high-quality dataset and benchmark for multi-hop, multi-document reasoning, along with human-annotated decompositions, to evaluate and improve LLMs.

Findings

Contemporary LLMs still struggle with inter-document reasoning.

Benchmark results show room for improvement in multi-hop question answering.

The dataset and tools are publicly available for further research.

Abstract

One type of question that is commonly found in day-to-day scenarios is ``fan-out'' questions, complex multi-hop, multi-document reasoning questions that require finding information about a large number of entities. However, there exist few resources to evaluate this type of question-answering capability among large language models. To evaluate complex reasoning in LLMs more fully, we present FanOutQA, a high-quality dataset of fan-out question-answer pairs and human-annotated decompositions with English Wikipedia as the knowledge base. We formulate three benchmark settings across our dataset and benchmark 7 LLMs, including GPT-4, LLaMA 2, Claude-2.1, and Mixtral-8x7B, finding that contemporary models still have room to improve reasoning over inter-document dependencies in a long context. We provide our dataset and open-source tools to run models to encourage evaluation at https://fanoutqa.com