PEFT methods (LoRA, Adapter) achieve lower membership-inference AUC than full fine-tuning, indicating reduced memorisation, but DP's protective effect is weaker for PEFT models.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Efficient and Private: Memorisation under differentially private parameter-efficient fine-tuning in language models
PEFT methods (LoRA, Adapter) achieve lower membership-inference AUC than full fine-tuning, indicating reduced memorisation, but DP's protective effect is weaker for PEFT models.