PrivacyScalpel: Enhancing LLM Privacy via Interpretable Feature Intervention with Sparse Autoencoders

Documenti analoghi

• IncogniText: Privacy-enhancing Conditional Text Anonymization via LLM-based Private Attribute Randomization
  di: Frikha, Ahmed, et al.
  Pubblicazione: (2024)
• PII-Compass: Guiding LLM training data extraction prompts towards the target PII via grounding
  di: Nakka, Krishna Kanth, et al.
  Pubblicazione: (2024)
• PII-Scope: A Comprehensive Study on Training Data PII Extraction Attacks in LLMs
  di: Nakka, Krishna Kanth, et al.
  Pubblicazione: (2024)
• Mammo-SAE: Interpreting Breast Cancer Concept Learning with Sparse Autoencoders
  di: Nakka, Krishna Kanth
  Pubblicazione: (2025)
• ObfuscaTune: Obfuscated Offsite Fine-tuning and Inference of Proprietary LLMs on Private Datasets
  di: Frikha, Ahmed, et al.
  Pubblicazione: (2024)