Metraipolis › archive › dc_d6222bd3cff3

Evaluating and Improving LLM Self-Modeling — arXiv:2608.30980

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30980
License (as archived)
CC0-1.0
Archived
2026-10-17T07:55:10.125Z
SHA-256 of the archived text
a4eb056c02464245d46ec98ceefc73f8c227749833cb3b069f2705d2ed651107
Stable URL of this page
https://metraipolis.com/documento/dc_d6222bd3cff3

How to cite

Evaluating and Improving LLM Self-Modeling — arXiv:2608.30980 — arXiv (en), https://arxiv.org/abs/2608.30980 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d6222bd3cff3 (Metraipolis archive dc_d6222bd3cff3, sha256 a4eb056c02464245d46ec98ceefc73f8c227749833cb3b069f2705d2ed651107)
BibTeX
@misc{dcd6222bd3cff3,
  title = {Evaluating and Improving LLM Self-Modeling — arXiv:2608.30980},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30980},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d6222bd3cff3; sha256 a4eb056c02464245d46ec98ceefc73f8c227749833cb3b069f2705d2ed651107},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30980. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We study self-modeling: an LLM's ability to answer questions about its own behavior. We focus on verifiable behavioral questions, such as whether a prompt edit would change the model's final answer. To measure this capability, we introduce a benchmark that tests diverse types of self-modeling questions. Current models show non-trivial but limited self-modeling skill, and make systematic mistakes on simple counterfactual questions about their own behavior. To improve self-modeling skill, we develop a scalable synthetic-data pipeline that produces self-modeling training data, and show that reinforcement-learning can improve aggregate self-modeling skill across three open-source model families with some transfer to held-out tasks. These gains, however, do not seem to constitute introspection consistently: improved self-modeling may not arise from privileged access to the model's internal decision process.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.30980 — 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 2026-10-17T07:28:26.936Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).