Metraipolis › archive › dc_73a784533cdb

From Representation to Readout: Mechanistic Analysis of Membership Signals in Language Models — arXiv:2610.41445

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.41445
License (as archived)
CC0-1.0
Archived
2026-11-02T04:45:20.629Z
SHA-256 of the archived text
7e494dc229ed672ecbd6dbb942b190ed08eb8e6978e4425848656231596296a1
Stable URL of this page
https://metraipolis.com/documento/dc_73a784533cdb

How to cite

From Representation to Readout: Mechanistic Analysis of Membership Signals in Language Models — arXiv:2610.41445 — arXiv (en), https://arxiv.org/abs/2610.41445 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_73a784533cdb (Metraipolis archive dc_73a784533cdb, sha256 7e494dc229ed672ecbd6dbb942b190ed08eb8e6978e4425848656231596296a1)
BibTeX
@misc{dc73a784533cdb,
  title = {From Representation to Readout: Mechanistic Analysis of Membership Signals in Language Models — arXiv:2610.41445},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41445},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_73a784533cdb; sha256 7e494dc229ed672ecbd6dbb942b190ed08eb8e6978e4425848656231596296a1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41445. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Membership inference aims to determine whether a particular example was included in a model's training data. Such membership signals are often broadly described as memorization, while the underlying mechanisms can involve both how information is represented internally and how it is accessed by the model to produce behaviors. We therefore present a mechanistic study of membership signals in large language models (LLMs). We introduce Decodable Information Score, Used Information Score, and Readout Accessibility Gap (DUG), a set of complementary measures that separately quantify membership information decodable from internal representations and the extent to which it is exposed through the model's readout. Across supervised fine-tuning (SFT) and reinforcement learning (RL)-based membership inference methods, membership-specific effects appear in both the amount of decodable information and readout under SFT, while under RL-based membership inference methods they appear primarily in readout. Activation patching further localizes this membership-specific effect to intermediate layers. To examine whether these findings extend to a more general RL setting, we further study reinforcement learning with verifiable rewards (RLVR) contamination, where the training reward is independent of the reference solution. In this setting, we find that behavioral changes can emerge without a corresponding increase in likelihood-based membership inference signals.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.41445 — 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 2026-11-02T04:42:12.468Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).