@inproceedings{chiang-lee-2023-closer,
title = "A Closer Look into Using Large Language Models for Automatic Evaluation",
author = "Chiang, Cheng-Han and
Lee, Hung-yi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2023",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.findings-emnlp.599",
doi = "10.18653/v1/2023.findings-emnlp.599",
pages = "8928--8942",
abstract = "Using large language models (LLMs) to evaluate text quality has recently gained popularity. Some existing prior works explore the idea of using LLMs for evaluation, while they differ in some details of the evaluation process. In this paper, we analyze *LLM evaluation* and *G-Eval*, and we discuss how those details in the evaluation process change how well the ratings given by LLMs correlate with human ratings. We find that the auto Chain-of-Thought (CoT) used in G-Eval does not always make G-Eval more aligned with human ratings. We also show that forcing the LLM to output only a numeric rating, as in G-Eval, is suboptimal. Last, we reveal that asking the LLM to explain its own ratings consistently improves the correlation between the ChatGPT and human ratings and pushes state-of-the-art (SoTA) correlations on two meta-evaluation datasets.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="chiang-lee-2023-closer">
<titleInfo>
<title>A Closer Look into Using Large Language Models for Automatic Evaluation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Cheng-Han</namePart>
<namePart type="family">Chiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hung-yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2023</title>
</titleInfo>
<name type="personal">
<namePart type="given">Houda</namePart>
<namePart type="family">Bouamor</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Juan</namePart>
<namePart type="family">Pino</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kalika</namePart>
<namePart type="family">Bali</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Singapore</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Using large language models (LLMs) to evaluate text quality has recently gained popularity. Some existing prior works explore the idea of using LLMs for evaluation, while they differ in some details of the evaluation process. In this paper, we analyze *LLM evaluation* and *G-Eval*, and we discuss how those details in the evaluation process change how well the ratings given by LLMs correlate with human ratings. We find that the auto Chain-of-Thought (CoT) used in G-Eval does not always make G-Eval more aligned with human ratings. We also show that forcing the LLM to output only a numeric rating, as in G-Eval, is suboptimal. Last, we reveal that asking the LLM to explain its own ratings consistently improves the correlation between the ChatGPT and human ratings and pushes state-of-the-art (SoTA) correlations on two meta-evaluation datasets.</abstract>
<identifier type="citekey">chiang-lee-2023-closer</identifier>
<identifier type="doi">10.18653/v1/2023.findings-emnlp.599</identifier>
<location>
<url>https://aclanthology.org/2023.findings-emnlp.599</url>
</location>
<part>
<date>2023-12</date>
<extent unit="page">
<start>8928</start>
<end>8942</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T A Closer Look into Using Large Language Models for Automatic Evaluation
%A Chiang, Cheng-Han
%A Lee, Hung-yi
%Y Bouamor, Houda
%Y Pino, Juan
%Y Bali, Kalika
%S Findings of the Association for Computational Linguistics: EMNLP 2023
%D 2023
%8 December
%I Association for Computational Linguistics
%C Singapore
%F chiang-lee-2023-closer
%X Using large language models (LLMs) to evaluate text quality has recently gained popularity. Some existing prior works explore the idea of using LLMs for evaluation, while they differ in some details of the evaluation process. In this paper, we analyze *LLM evaluation* and *G-Eval*, and we discuss how those details in the evaluation process change how well the ratings given by LLMs correlate with human ratings. We find that the auto Chain-of-Thought (CoT) used in G-Eval does not always make G-Eval more aligned with human ratings. We also show that forcing the LLM to output only a numeric rating, as in G-Eval, is suboptimal. Last, we reveal that asking the LLM to explain its own ratings consistently improves the correlation between the ChatGPT and human ratings and pushes state-of-the-art (SoTA) correlations on two meta-evaluation datasets.
%R 10.18653/v1/2023.findings-emnlp.599
%U https://aclanthology.org/2023.findings-emnlp.599
%U https://doi.org/10.18653/v1/2023.findings-emnlp.599
%P 8928-8942
Markdown (Informal)
[A Closer Look into Using Large Language Models for Automatic Evaluation](https://aclanthology.org/2023.findings-emnlp.599) (Chiang & Lee, Findings 2023)
ACL