{
  "id": 1383,
  "url": "https://arxiv.org/abs/2606.06946v1",
  "title": "Auditing Training Data in Domain-adapted LLMs: LoRA-MINT",
  "summary": "We present LoRA-MINT, a new methodology for Membership Inference Test (MINT) applied to recent Large Language Models (LLMs) fine-tuned for specific Natural Language Processing (NLP) tasks through Low-Rank Adaptation (LoRA). The primary goal is to assess whether individual samples were part of the training data of these adapted models, providing a useful auditing tool for the management of intellectual property and sensitive data. Our analysis explores the relationship between model perplexity an",
  "authors": "Gonzalo Mancera, Daniel DeAlcala, Aythami Morales, Julian Fierrez, Ruben Tolosana, Francisco Jurado",
  "category": "research",
  "topics": "copyright-ip,transparency",
  "orgs": null,
  "regions": null,
  "published_at": "2026-06-05T06:19:03.000Z",
  "fetched_at": "2026-07-14T14:15:12.459Z",
  "source_slug": "arxiv-ethics",
  "source_name": "arXiv",
  "source_homepage": "https://arxiv.org",
  "ethics_ai_record_url": "https://ethics.ai/record/1383",
  "original_url": "https://arxiv.org/abs/2606.06946v1",
  "evidence_status": "source-only",
  "attribution": "via ethics.ai"
}