Leave No Document Behind: Benchmarking Long-Context LLMs with Extended Multi-Doc QA
This addresses the need for realistic benchmarks in long-context AI applications, though it is incremental as it builds on existing evaluation frameworks.
The authors tackled the problem of evaluating long-context LLMs by proposing a new benchmark called Loong, which uses extended multi-document QA with relevant documents to simulate real-world scenarios, and found that existing models and RAG perform poorly, indicating significant room for improvement.
Long-context modeling capabilities have garnered widespread attention, leading to the emergence of Large Language Models (LLMs) with ultra-context windows. Meanwhile, benchmarks for evaluating long-context LLMs are gradually catching up. However, existing benchmarks employ irrelevant noise texts to artificially extend the length of test cases, diverging from the real-world scenarios of long-context applications. To bridge this gap, we propose a novel long-context benchmark, Loong, aligning with realistic scenarios through extended multi-document question answering (QA). Unlike typical document QA, in Loong's test cases, each document is relevant to the final answer, ignoring any document will lead to the failure of the answer. Furthermore, Loong introduces four types of tasks with a range of context lengths: Spotlight Locating, Comparison, Clustering, and Chain of Reasoning, to facilitate a more realistic and comprehensive evaluation of long-context understanding. Extensive experiments indicate that existing long-context language models still exhibit considerable potential for enhancement. Retrieval augmented generation (RAG) achieves poor performance, demonstrating that Loong can reliably assess the model's long-context modeling capabilities.