Bayesian Calibration of Win Rate Estimation with LLM Evaluators

Document detail

ID

oai:arXiv.org:2411.04424

Topic

Computer Science - Computation and... Computer Science - Artificial Inte...

Author

Gao, Yicheng Xu, Gonghan Wang, Zhe Cohan, Arman

Year

2024

listing date

1/1/2025

Keywords

estimation methods llm evaluators

Metrics

Abstract

Recent advances in large language models (LLMs) show the potential of using LLMs as evaluators for assessing the quality of text generations from LLMs.

However, applying LLM evaluators naively to compare or judge between different systems can lead to unreliable results due to the intrinsic win rate estimation bias of LLM evaluators.

In order to mitigate this problem, we propose two calibration methods, Bayesian Win Rate Sampling (BWRS) and Bayesian Dawid-Skene, both of which leverage Bayesian inference to more accurately infer the true win rate of generative language models.

We empirically validate our methods on six datasets covering story generation, summarization, and instruction following tasks.

We show that both our methods are effective in improving the accuracy of win rate estimation using LLMs as evaluators, offering a promising direction for reliable automatic text quality evaluation.

;Comment: Accepted by EMNLP 2024

Gao, Yicheng,Xu, Gonghan,Wang, Zhe,Cohan, Arman, 2024, Bayesian Calibration of Win Rate Estimation with LLM Evaluators