Extends WildChat
WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback
ACL 2026 · First arXiv submission
Converts naturally occurring user feedback into preferred and dispreferred responses. Its pipeline detects satisfaction and dissatisfaction, constructs preference examples, and evaluates responses against checklists derived from the feedback.
Data used
148,715 multi-turn conversations from the original WildChat corpus, used to identify feedback and construct 20,281 preference examples.
Findings
WildFeedback contains 20,281 preference examples. Fine-tuned models improve on the paper’s alignment and checklist-based evaluations.
Sections 2–3: feedback mining and evaluation. Appendix E: WildChat source. Read the source.
Dates and versions
First arXiv version. ArXiv version history.
ACL 2026 proceedings paper.
Cite this paper
Taiwei Shi, Zhuoer Wang, Longqi Yang, Ying-Chun Lin, Zexue He, Mengting Wan, Pei Zhou, Sujay Kumar Jauhar, Sihao Chen, Shan Xia, Hongfei Zhang, Jieyu Zhao, Xiaofeng Xu, Xia Song, Jennifer Neville. 2026. WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).
Show BibTeX
@inproceedings{wildfeedback2024,
title = {{WildFeedback: Aligning LLMs With In-situ User Interactions And Feedback}},
author = {Shi, Taiwei and Wang, Zhuoer and Yang, Longqi and Lin, Ying-Chun and He, Zexue and Wan, Mengting and Zhou, Pei and Jauhar, Sujay Kumar and Chen, Sihao and Xia, Shan and Zhang, Hongfei and Zhao, Jieyu and Xu, Xiaofeng and Song, Xia and Neville, Jennifer},
year = {2026},
booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
publisher = {Association for Computational Linguistics},
pages = {36701--36725},
doi = {10.18653/v1/2026.acl-long.1701},
url = {https://aclanthology.org/2026.acl-long.1701/}
}