# How to Choose a Threshold for an Evaluation Metric for Large Language Models

arXiv: https://arxiv.org/abs/2412.12148 · HTML: https://mlsanity.com/paper/2412.12148/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 7 of its 23 references that are in the map.
- First posted on arXiv 10 Dec 2024.

## Builds on (most cited first, 7 of 7)

- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [A Survey of Large Language Models](https://mlsanity.com/paper/2303.18223/) (2023)
- [Ragas: Automated Evaluation of Retrieval Augmented Generation](https://mlsanity.com/paper/2309.15217/) (2023)
- [Evaluating Large Language Models: A Comprehensive Survey](https://mlsanity.com/paper/2310.19736/) (2023)
- [Lynx: An Open Source Hallucination Evaluation Model](https://mlsanity.com/paper/2407.08488/) (2024)
- [Schroedinger's Threshold: When the AUC doesn't predict Accuracy](https://mlsanity.com/paper/2404.03344/) (2024)
- [Model Validation Practice in Banking: A Structured Approach for Predictive Models](https://mlsanity.com/paper/2410.13877/) (2024)

## Abstract (arXiv)

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.