Metraipolis › archive › dc_c2aedd558d21
Auditable Unlearning for LLMs — arXiv:2611.05866
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.05866
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-05T18:16:49.281Z
- SHA-256 of the archived text
05c344b8d231f6d277ec5906c9bd9c575257cbaeb2f1697bf314e76b19a53872- Stable URL of this page
https://metraipolis.com/documento/dc_c2aedd558d21
How to cite
Auditable Unlearning for LLMs — arXiv:2611.05866 — arXiv (en), https://arxiv.org/abs/2611.05866 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c2aedd558d21 (Metraipolis archive dc_c2aedd558d21, sha256 05c344b8d231f6d277ec5906c9bd9c575257cbaeb2f1697bf314e76b19a53872)
BibTeX
@misc{dcc2aedd558d21,
title = {Auditable Unlearning for LLMs — arXiv:2611.05866},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.05866},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c2aedd558d21; sha256 05c344b8d231f6d277ec5906c9bd9c575257cbaeb2f1697bf314e76b19a53872},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.05866. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Machine unlearning aims to remove a user's data from a trained model after a deletion request. Existing methods are majorly evaluated on three axes - how well they forget, how much capability they retain, and how robust they are to membership inference attacks, but not on whether the provider can offer the user any evidence that the deletion occurred. To fill this gap, we introduce auditability as a fourth axis, defined as the separation between requested (forget set) and retained (retain set) data under a statistic the provider computes. We show that across eleven unlearning methods on TOFU, MUSE and RWKU no method satisfies all four axes, with different methods falling short on different axes. To address this, we propose a two stage method that can be used on top of any unlearning method, in the the first stage we propose ${GapWiden}$ which uses a hinge loss to increase the difference between audit statistics of forget set and retain set. In the second stage we propose ${GapVeil}$, an inference time rejection-sampler which uses the base model as the drafter and re-samples based on the audit statistics of the drafted response. We observe that using our method, auditability and leakage reach oracle parity for every unlearning method, and forget performance converges to the oracle; retain performance improves for nine of eleven methods but remains short of parity. Our results show that evidence of deletion or auditability can be constructed by design, and that treating it as an explicit requirement changes the viability of existing unlearning methods.
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.05866 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-05T18:12:49.001Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).