
PK
Peter Kirby
· 1 min read
ResearcharXiv cs.CL
Contrastive Learning for Authorship Verification
arXiv:2609.28471v1 Announce Type: new
Abstract: Our results show that contrastive learning outperforms a classification-based approach to authorship verification under the tested settings. We identify loss function, batch size, training duration, pre-trained model, input context length, and random text span data augmentation as important factors of model performance. Based on these considerations, we develop a ModernBERT Bi-Encoder model that achieves 98.4% accuracy on the PAN21 authorship verification task.
Original source
This story was published by arXiv cs.CL and written by Peter Kirby. SyncAI.news shows a preview; the complete article is on the publisher's site.
Read the full story on arxiv.org


