How to Choose a Threshold for an Evaluation Metric for Large Language Models
This work tackles the problem of reliable threshold selection for LLM deployment, which is crucial for stakeholders in regulated industries, though it is incremental as it adapts existing risk management guidelines.
The paper addresses the lack of methodology for selecting robust thresholds for LLM evaluation metrics, proposing a step-by-step recipe based on risk management principles and demonstrating it on the Faithfulness metric using the HaluBench dataset.
To ensure and monitor large language models (LLMs) reliably, various evaluation metrics have been proposed in the literature. However, there is little research on prescribing a methodology to identify a robust threshold on these metrics even though there are many serious implications of an incorrect choice of the thresholds during deployment of the LLMs. Translating the traditional model risk management (MRM) guidelines within regulated industries such as the financial industry, we propose a step-by-step recipe for picking a threshold for a given LLM evaluation metric. We emphasize that such a methodology should start with identifying the risks of the LLM application under consideration and risk tolerance of the stakeholders. We then propose concrete and statistically rigorous procedures to determine a threshold for the given LLM evaluation metric using available ground-truth data. As a concrete example to demonstrate the proposed methodology at work, we employ it on the Faithfulness metric, as implemented in various publicly available libraries, using the publicly available HaluBench dataset. We also lay a foundation for creating systematic approaches to select thresholds, not only for LLMs but for any GenAI applications.