Metraipolis › archive › dc_f1ac4ab866ec

World Models Need Not Be Right to Rank Policies Right: Relative Bounds and Audit Limits — arXiv:2611.01139

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.01139
License (as archived)
CC0-1.0
Archived
2026-11-03T10:35:42.607Z
SHA-256 of the archived text
01090994f40de06957cc295dd9a9460240a5f1a7c94a59f8fbc2cfc6cb3d41cf
Stable URL of this page
https://metraipolis.com/documento/dc_f1ac4ab866ec

How to cite

World Models Need Not Be Right to Rank Policies Right: Relative Bounds and Audit Limits — arXiv:2611.01139 — arXiv (en), https://arxiv.org/abs/2611.01139 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f1ac4ab866ec (Metraipolis archive dc_f1ac4ab866ec, sha256 01090994f40de06957cc295dd9a9460240a5f1a7c94a59f8fbc2cfc6cb3d41cf)
BibTeX
@misc{dcf1ac4ab866ec,
  title = {World Models Need Not Be Right to Rank Policies Right: Relative Bounds and Audit Limits — arXiv:2611.01139},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.01139},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f1ac4ab866ec; sha256 01090994f40de06957cc295dd9a9460240a5f1a7c94a59f8fbc2cfc6cb3d41cf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01139. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Selecting a reinforcement-learning (RL) policy usually requires additional environment interaction, while learned world models offer cheap but potentially biased evaluations. This raises two distinct questions: can shared model errors preserve pairwise rankings despite poor absolute values, and do they make a small number of real audits more informative? We derive an exact pairwise simulation identity and a relative bound separating common-mode Bellman residuals from policy-specific error. The analysis motivates RAVEL (Relative Auditing of Value Errors in Learned models), a selective policy tournament whose eliminations use held-out real outcomes. We test 45 SAC, TD3, and PPO policies per task in DMC cheetah-run and walker-walk. At rollout horizon 500, model rankings remain strong (Kendall tau = 0.962 and 0.800), and same-run differential-error ratios are 0.674 and 0.106. Yet mean paired audit-error correlations are only 0.007 and 0.090, and pairing changes standard deviation by -2.5 percent and +1.9 percent. Simultaneous intervals attain 99.98 percent and 99.75 percent coverage, but certify only 13.2 percent and 0 percent of decisions at eight audits. RAVEL improves low-budget walker-walk regret, whereas real-only successive halving is better at the largest budget, rejecting the precommitted twofold efficiency target. Frozen-model optimization lowers real return without increasing predicted return on average. Thus ranking fidelity, useful audit covariance, and upward model exploitation are distinct properties, separated in part by closed-loop trajectory divergence.

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/2611.01139 — 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 2026-11-03T10:29:48.746Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).