Metraipolis › archive › dc_f97dc39434cb
ReplayAudit: Auditing Outsourced LLM Fine-Tuning via Sealed Records and Selective Replay — arXiv:2611.09607
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.09607
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-09T22:40:35.157Z
- SHA-256 of the archived text
2a8dd012bc52b6a7b8d86ecf5df392516b8578dd72fe827fa7d262ac5418410c- Stable URL of this page
https://metraipolis.com/documento/dc_f97dc39434cb
How to cite
ReplayAudit: Auditing Outsourced LLM Fine-Tuning via Sealed Records and Selective Replay — arXiv:2611.09607 — arXiv (en), https://arxiv.org/abs/2611.09607 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f97dc39434cb (Metraipolis archive dc_f97dc39434cb, sha256 2a8dd012bc52b6a7b8d86ecf5df392516b8578dd72fe827fa7d262ac5418410c)
BibTeX
@misc{dcf97dc39434cb,
title = {ReplayAudit: Auditing Outsourced LLM Fine-Tuning via Sealed Records and Selective Replay — arXiv:2611.09607},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.09607},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f97dc39434cb; sha256 2a8dd012bc52b6a7b8d86ecf5df392516b8578dd72fe827fa7d262ac5418410c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.09607. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Auditing outsourced LLM fine-tuning requires checking an untrusted provider's training while controlling verification cost. Uniform sampling ignores training-time signals, while checking a suspicious record can require replaying several preceding steps. We present ReplayAudit, a telemetry-guided auditing system that seals training records before challenges and independently replays sampled intervals from sealed checkpoints. Loss changes and parameter-update norms guide stratified random sampling; replay determines whether the selected computation agrees with its records. We relate sampling probabilities to replay work and evaluate the system on real LoRA training. A 7B BF16 prototype rejects a false update-norm report during a random audit; targeted checks also reject false loss reports and skipped updates. We calibrate sampling budgets offline to at least a 90% probability of directly selecting an injected step. For two learning-rate deviation settings, executing these budgets on the same honest 1.5B trajectory reduces median complete-audit wall time by 49% and 67% relative to uniform sampling. When skipped updates evade the high-anomaly stratum, the corresponding audit instead costs 38% more. These results show how deviation capture and actual replay work jointly determine the cost of telemetry-guided auditing. Tema degli esploratori: fiducia digitale e crittografia (elenco arXiv cs.CR, nuovi lavori del giorno). Categorie arXiv: Cryptography and Security (cs.CR) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.09607 — 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 2026-11-09T22:33:17.037Z dalla pagina https://arxiv.org/list/cs.CR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).