SubmissionNumber#=%=#221
FinalPaperTitle#=%=#Team MLab at SemEval-2024 Task 8: Analyzing Encoder Embeddings for Detecting LLM-generated Text
ShortPaperTitle#=%=#
NumberOfPages#=%=#5
CopyrightSigned#=%=#Kevin Li
JobTitle#==#
Organization#==#Stanford University - 450 Jane Stanford Way, Stanford, CA, 94305
Abstract#==#This paper explores solutions to the challenges posed by the widespread use of LLMs, particularly in the context of identifying human-written versus machine-generated text. Focusing on Subtask B of SemEval 2024 Task 8, we compare the performance of RoBERTa and DeBERTa models. Subtask B involved identifying not only human or machine text but also the specific LLM responsible for generating text, where our DeBERTa model outperformed the RoBERTa baseline by over 10% in leaderboard accuracy. The results highlight the rapidly growing capabilities of LLMs and importance of keeping up with the latest advancements. Additionally, our paper presents visualizations using PCA and t-SNE that showcase the DeBERTa model's ability to cluster different LLM outputs effectively. These findings contribute to understanding and improving AI methods for detecting machine-generated text, allowing us to build more robust and traceable AI systems in the language ecosystem.
Author{1}{Firstname}#=%=#Kevin
Author{1}{Lastname}#=%=#Li
Author{1}{Email}#=%=#kevinli7@stanford.edu
Author{1}{Affiliation}#=%=#Stanford University
Author{2}{Firstname}#=%=#Kenan
Author{2}{Lastname}#=%=#Hasanaliyev
Author{2}{Username}#=%=#claserken
Author{2}{Email}#=%=#kenanhas@stanford.edu
Author{2}{Affiliation}#=%=#Stanford MLab
Author{3}{Firstname}#=%=#Sally
Author{3}{Lastname}#=%=#Zhu
Author{3}{Email}#=%=#salzhu@stanford.edu
Author{3}{Affiliation}#=%=#Stanford University
Author{4}{Firstname}#=%=#George
Author{4}{Lastname}#=%=#Altshuler
Author{4}{Email}#=%=#gwa@stanford.edu
Author{4}{Affiliation}#=%=#Stanford University
Author{5}{Firstname}#=%=#Alden
Author{5}{Lastname}#=%=#Eberts
Author{5}{Email}#=%=#ajeberts@stanford.edu
Author{5}{Affiliation}#=%=#Stanford University
Author{6}{Firstname}#=%=#Eric
Author{6}{Lastname}#=%=#Chen
Author{6}{Email}#=%=#ericc27@stanford.edu
Author{6}{Affiliation}#=%=#Stanford University
Author{7}{Firstname}#=%=#Kate
Author{7}{Lastname}#=%=#Wang
Author{7}{Email}#=%=#kyw1923@stanford.edu
Author{7}{Affiliation}#=%=#Stanford University
Author{8}{Firstname}#=%=#Emily
Author{8}{Lastname}#=%=#Xia
Author{8}{Email}#=%=#emxia18@stanford.edu
Author{8}{Affiliation}#=%=#Stanford University
Author{9}{Firstname}#=%=#Eli
Author{9}{Lastname}#=%=#Browne
Author{9}{Email}#=%=#ebrowne@stanford.edu
Author{9}{Affiliation}#=%=#Stanford University
Author{10}{Firstname}#=%=#Ian
Author{10}{Lastname}#=%=#Chen
Author{10}{Email}#=%=#ianychen@stanford.edu
Author{10}{Affiliation}#=%=#Stanford University
Author{11}{Firstname}#=%=#Umut
Author{11}{Lastname}#=%=#Eren
Author{11}{Email}#=%=#umuteren@stanford.edu
Author{11}{Affiliation}#=%=#Stanford University

==========
èéáğö