Metraipolis › archive › dc_4bd446b3e150

SelfImprove-Bench: Evaluating Generalization in Self-Improving Agents — arXiv:2611.02797

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.02797
License (as archived)
CC0-1.0
Archived
2026-11-03T10:19:46.157Z
SHA-256 of the archived text
fa842ac35bdb005b241de9a41c26f8f99dc2ea70b63e8b0ec1bf237d35200720
Stable URL of this page
https://metraipolis.com/documento/dc_4bd446b3e150

How to cite

SelfImprove-Bench: Evaluating Generalization in Self-Improving Agents — arXiv:2611.02797 — arXiv (en), https://arxiv.org/abs/2611.02797 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4bd446b3e150 (Metraipolis archive dc_4bd446b3e150, sha256 fa842ac35bdb005b241de9a41c26f8f99dc2ea70b63e8b0ec1bf237d35200720)
BibTeX
@misc{dc4bd446b3e150,
  title = {SelfImprove-Bench: Evaluating Generalization in Self-Improving Agents — arXiv:2611.02797},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.02797},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4bd446b3e150; sha256 fa842ac35bdb005b241de9a41c26f8f99dc2ea70b63e8b0ec1bf237d35200720},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.02797. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recursively self-improving (RSI) systems are increasingly viewed as a promising means of accelerating progress toward artificial general intelligence. They aim to autonomously improve their capabilities by refining their working pipelines and recursively improving the strategies that guide subsequent refinements, with gains generalizing to unseen this http URL , existing evaluations lack reliable metrics to distinguish capability gains from scenario fitting, sustainable and extensible protocols for unseen tasks and difficulties, and fine-grained attribution of improvement to evolution traces. To this end, we propose SelfImprove-Bench,an RSI evaluation framework that organizes tasks into three primary capabilities (i.e., Exploring, Planning, Reasoning) as scenarios for guiding model evolution. We instantiated the tasks on games for controllability and reproducibility, and pairs them with scalable generation and orchestration driven by task templates and difficulty coefficients. Beyond task-level scores, SelfImprove-Bench organizes evaluation into generalization ladder levels G0 to G4 that measure how far gains transfer from training conditions. Experiments reveal that the evolutionary gains of existing RSI methods are inconsistent across generalization levels.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.MA, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA).
Pagina del lavoro: https://arxiv.org/abs/2611.02797 — 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 2026-11-03T10:19:34.839Z dalla pagina https://arxiv.org/list/cs.MA/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).