InstructEval: Instruction-Tuned Text Evaluator from Human Preference

Wenhao Wu, Wei Li, Xinyan Xiao, Jiachen Liu, Sujian Li


Abstract
This paper explores to construct a general text evaluator based on open-source Large Language Models (LLMs), a domain predominantly occupied by commercial counterparts such as GPT-4. Recognizing the limitations of open-source models like Llama in evaluative tasks, we introduce InstructEval, a general multi-aspect text evaluator developed through instruction tuning of open-source LLMs. To overcome the shortage of annotated resources for multi-aspect evaluations, InstructEval combines extensive open Human Preference Modeling (HPM) datasets with a small set of multi-aspect annotated data.This approach not only enhances effectiveness in overall evaluation tasks but also exhibits improved performance in multi-aspect evaluation tasks.As demonstrated by our extensive experiments, InstructEval achieves comparable or superior performance to commercial LLMs like ChatGPT or GPT-4 in terms of both overall and multi-aspect evaluation.
Anthology ID:
2024.findings-acl.799
Volume:
Findings of the Association for Computational Linguistics: ACL 2024
Month:
August
Year:
2024
Address:
Bangkok, Thailand
Editors:
Lun-Wei Ku, Andre Martins, Vivek Srikumar
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
13462–13474
Language:
URL:
https://aclanthology.org/2024.findings-acl.799
DOI:
10.18653/v1/2024.findings-acl.799
Bibkey:
Cite (ACL):
Wenhao Wu, Wei Li, Xinyan Xiao, Jiachen Liu, and Sujian Li. 2024. InstructEval: Instruction-Tuned Text Evaluator from Human Preference. In Findings of the Association for Computational Linguistics: ACL 2024, pages 13462–13474, Bangkok, Thailand. Association for Computational Linguistics.
Cite (Informal):
InstructEval: Instruction-Tuned Text Evaluator from Human Preference (Wu et al., Findings 2024)
Copy Citation:
PDF:
https://preview.aclanthology.org/dois-2013-emnlp/2024.findings-acl.799.pdf