Towards a Holistic and Automated Evaluation Framework for Multi-Level Comprehension of LLMs in Book-Length Contexts

摘要

We introduce HAMLET, a holistic and automated framework for evaluating the long-context comprehension of large language models (LLMs). HAMLET structures key information of source texts into a three-level hierarchy at root-, branch-, and leaf-levels, and employs query-focused summarization to evaluate how well models faithfully recall the key information at each level. To validate the reliability of our fully automated pipeline, we conduct a systematic human study, demonstrating that our automatic evaluation achieves over 90% agreement with expert human judgments, while reducing the evaluation cost by up to 25\times. HAMLET reveals that LLMs struggle with fine-grained comprehension, especially at the leaf level, and are sensitive to positional effects like the lost-in-the-middle. Analytical queries pose greater challenges than narrative ones, and consistent performance gaps emerge between open-source and proprietary models, as well as across model scales. Our code and dataset are publicly available at https://github.com/DISL-Lab/HAMLET.