Saved in:
Bibliographic Details
Main Authors: Gao, Yicheng, Xu, Gonghan, Wang, Zhe, Cohan, Arman
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.04424
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915077966266368
author Gao, Yicheng
Xu, Gonghan
Wang, Zhe
Cohan, Arman
author_facet Gao, Yicheng
Xu, Gonghan
Wang, Zhe
Cohan, Arman
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2411_04424
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Bayesian Calibration of Win Rate Estimation with LLM Evaluators
Gao, Yicheng
Xu, Gonghan
Wang, Zhe
Cohan, Arman
Computation and Language
Artificial Intelligence
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.
title Bayesian Calibration of Win Rate Estimation with LLM Evaluators
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2411.04424