SubmissionNumber#=%=#269
FinalPaperTitle#=%=#HIT-MI&T Lab at SemEval-2024 Task 6: DeBERTa-based Entailment Model is a Reliable Hallucination Detector
ShortPaperTitle#=%=#
NumberOfPages#=%=#11
CopyrightSigned#=%=#Wei Liu
JobTitle#==#
Organization#==#
Abstract#==#This paper describes our submission for SemEval-2024 Task 6: SHROOM, a Shared-task on Hallucinations and Related Observable Overgeneration Mistakes. We propose four groups of methods for hallucination detection: 1) Entailment Recognition; 2) Similarity Search; 3) Factuality Verification; 4) Confidence Estimation. The four methods rely on either the semantic relationship between the hypothesis and its source (target) or on the model-aware features during decoding. We participated in both the model-agnostic and model-aware tracks. Our method's effectiveness is validated by our high rankings 3rd in the model-agnostic track and 5th in the model-aware track. We have released our code on GitHub.
Author{1}{Firstname}#=%=#Wei
Author{1}{Lastname}#=%=#Liu
Author{1}{Username}#=%=#liuweihit
Author{1}{Email}#=%=#liuweihit2023@163.com
Author{1}{Affiliation}#=%=#Harbin Institute of Technology
Author{2}{Firstname}#=%=#Wanyao
Author{2}{Lastname}#=%=#Shi
Author{2}{Email}#=%=#shiwanyao@qq.com
Author{2}{Affiliation}#=%=#Northwest Normal University
Author{3}{Firstname}#=%=#Zijian
Author{3}{Lastname}#=%=#Zhang
Author{3}{Email}#=%=#zhangzj0318@qq.com
Author{3}{Affiliation}#=%=#Harbin Institute of Technology
Author{4}{Firstname}#=%=#Hui
Author{4}{Lastname}#=%=#Huang
Author{4}{Username}#=%=#huang_hui
Author{4}{Email}#=%=#22b903058@stu.hit.edu.cn
Author{4}{Affiliation}#=%=#Harbin Institute of Technology

==========
èéáğö