Towards Better Evaluation of Instruction-Following: A Case-Study in Summarization
This work addresses the open problem of instruction-following evaluation for LLMs, which is crucial for improving model reliability and usability, but it is incremental as it builds on existing prompt-based approaches.
The paper tackles the problem of evaluating how well large language models follow user instructions by conducting a meta-evaluation of various metrics on a new dataset, riSum, containing 300 document-instruction pairs with 900 answers rated by humans, and proposes new LLM-based reference-free evaluation methods that perform on par with costly reference-based metrics.
Despite recent advances, evaluating how well large language models (LLMs) follow user instructions remains an open problem. While evaluation methods of language models have seen a rise in prompt-based approaches, limited work on the correctness of these methods has been conducted. In this work, we perform a meta-evaluation of a variety of metrics to quantify how accurately they measure the instruction-following abilities of LLMs. Our investigation is performed on grounded query-based summarization by collecting a new short-form, real-world dataset riSum, containing 300 document-instruction pairs with 3 answers each. All 900 answers are rated by 3 human annotators. Using riSum, we analyze the agreement between evaluation methods and human judgment. Finally, we propose new LLM-based reference-free evaluation methods that improve upon established baselines and perform on par with costly reference-based metrics that require high-quality summaries.