Metraipolis › archive › dc_779b80a8880c

Policy-Invariant Reward Shaping from LLM Feedback: A Framework for Hybrid RL Agents — arXiv:2608.18008

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.18008
License (as archived)
CC0-1.0
Archived
2026-10-18T14:33:50.626Z
SHA-256 of the archived text
548e1e67fc0df814256aaf2f6a645f994ba257a62c55761a8a0a0bf96459ccd5
Stable URL of this page
https://metraipolis.com/documento/dc_779b80a8880c

How to cite

Policy-Invariant Reward Shaping from LLM Feedback: A Framework for Hybrid RL Agents — arXiv:2608.18008 — arXiv (en), https://arxiv.org/abs/2608.18008 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_779b80a8880c (Metraipolis archive dc_779b80a8880c, sha256 548e1e67fc0df814256aaf2f6a645f994ba257a62c55761a8a0a0bf96459ccd5)
BibTeX
@misc{dc779b80a8880c,
  title = {Policy-Invariant Reward Shaping from LLM Feedback: A Framework for Hybrid RL Agents — arXiv:2608.18008},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.18008},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_779b80a8880c; sha256 548e1e67fc0df814256aaf2f6a645f994ba257a62c55761a8a0a0bf96459ccd5},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.18008. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Combining large language models with reinforcement learning is increasingly explored, yet the theoretical status of LLM-derived reward signals is often left implicit. We formalize the hybrid LLM-planner and RL-controller architecture as a Goal-Augmented Markov Decision Process and show that when the LLM per-state progress score is used as a bounded potential function, the resulting shaping term preserves the optimal policy set even when the LLM scores are inaccurate. This guarantee is stronger than what general LLM-as-reward approaches provide. We verify the result numerically on a small MDP under four potential configurations, including an adversarial one scaled to twenty times the base reward magnitude.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-19 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.18008 — 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 2026-10-18T14:27:54.479Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-19?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).