Gonzalo Mancera
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
inLoRA-MINT: Internal Representations for Auditing Training Data in Domain-adapted LLMs
According to the description, a preprint proposes testing whether specific samples were used to fine-tune a model with LoRA, by reading internal representations rather than needing full access to parameters. It reports scores of 95% to 98% on four models and three datasets, and the authors say it supports auditing for privacy and AI governance compliance.