Metraipolis › archive › dc_44e630fe702c

Learning to Act with World Models by Valuing What Comes Next — arXiv:2611.12200

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.12200
License (as archived)
CC0-1.0
Archived
2026-11-10T18:41:47.101Z
SHA-256 of the archived text
89d7abf7834e12999fabee97137010ed7ff12eb77df3ee6e6c2abeb7701d85b2
Stable URL of this page
https://metraipolis.com/documento/dc_44e630fe702c

How to cite

Learning to Act with World Models by Valuing What Comes Next — arXiv:2611.12200 — arXiv (en), https://arxiv.org/abs/2611.12200 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_44e630fe702c (Metraipolis archive dc_44e630fe702c, sha256 89d7abf7834e12999fabee97137010ed7ff12eb77df3ee6e6c2abeb7701d85b2)
BibTeX
@misc{dc44e630fe702c,
  title = {Learning to Act with World Models by Valuing What Comes Next — arXiv:2611.12200},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.12200},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_44e630fe702c; sha256 89d7abf7834e12999fabee97137010ed7ff12eb77df3ee6e6c2abeb7701d85b2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.12200. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

A sparse terminal reward tells an agent little about its steps. Much of the missing information is implicit in the environment's dynamics, because each state transition can change the conditions for success long before the reward arrives. A world model trained on real transitions can learn to predict what each action changes, but a predicted observation does not by itself show whether success has become more likely. With a binary terminal reward, PPO's critic makes this judgment by estimating the current policy's probability of success, but only for states the agent reaches. Our method, WM-PPO, elegantly brings the two together. For each action the policy samples, executed or not, the world model predicts the next observation and the critic values the predicted successor. In the policy channel, comparisons among these values give the policy auxiliary advantages; in the value channel, their mean gives the critic an auxiliary target. A shared gate weights both channels according to how reliably comparisons with similar margins predicted real outcomes earlier in training. The policy, critic and world model all learn from the agent's own interaction, with no teacher or privileged information, and only the policy is used at inference. With a comparable number of real environment steps, WM-PPO raises success from PPO's 63.5% to 87.5% on ALFWorld and from 39.3% to 52.9% on ScienceWorld. It matches the strongest PPO refinements we tested on ALFWorld and exceeds the best of them by 6.5 points on ScienceWorld. The trained policy more often completes prerequisite steps before placing an object or committing to an answer.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.12200 — 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 2026-11-10T18:37:37.542Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).