2 papers
cs.LG2026
Activation-Conditioned Self-Distillation
Zhexi Lu, Subhajit Chaudhury, Tejaswini Pedapati +2
On-policy self-distillation uses a model as its own teacher to provide dense supervision for reasoning, often through reference-solution conditioning. Providing privileged informat…
cs.CR2025
In-Context Probing for Membership Inference in Fine-Tuned Language Models
Zhexi Lu, Hongliang Chi, Nathalie Baracaldo +3
Membership inference attacks (MIAs) pose a critical privacy threat to fine-tuned large language models (LLMs), especially when models are adapted to domain-specific tasks using sen…