Community
Open-weights 278M PII detection model and PII-SCOPE benchmark released by Protegrity
A RoBERTa fine-tune trained on 1.2B synthetic text samples detects PII from surrounding context and can return positions and types or a fully redacted input. A technical report and Hugging Face model card accompany it.
Every outlet that ran it 1 article
r/MachineLearningCommunityLead
A new, open-weights 278M PII detection model and reference benchmark [R]