Metraipolis › archive › dc_ca22a2b7523c
Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction — arXiv:2610.19883
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.19883
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T14:19:01.928Z
- SHA-256 of the archived text
d57e9663c7995c429ce8ef8c29c35566c1d6e2e93ba4bcce440252e16ad6e75f- Stable URL of this page
https://metraipolis.com/documento/dc_ca22a2b7523c
How to cite
Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction — arXiv:2610.19883 — arXiv (en), https://arxiv.org/abs/2610.19883 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ca22a2b7523c (Metraipolis archive dc_ca22a2b7523c, sha256 d57e9663c7995c429ce8ef8c29c35566c1d6e2e93ba4bcce440252e16ad6e75f)
BibTeX
@misc{dcca22a2b7523c,
title = {Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction — arXiv:2610.19883},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.19883},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ca22a2b7523c; sha256 d57e9663c7995c429ce8ef8c29c35566c1d6e2e93ba4bcce440252e16ad6e75f},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.19883. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
The growing interest in behavior foundation models, combined with recent advancements in generative modeling, has led to an appealing recipe for reinforcement learning from unlabeled data: pre-train a generative model to model occupancies, i.e., discounted state visitations, on reward-free trajectories, such that a policy can be easily extracted once a small reward-labeled dataset becomes available. We decompose this recipe along three axes: (i) the degree to which the multi-modal structure of occupancies must be captured, (ii) the class of generative model, and (iii) the method used for policy extraction. Along each axis, we consider existing choices, propose simpler alternatives, and validate each decision experimentally. This analysis culminates in a novel, streamlined method that leverages drifting, a modern one-step generative sampler, to model the occupancy. We show that this enables policy extraction via direct differentiation through the occupancy model, and bypasses the need for critic distillation. Overall, we observe that this nimble setup leads to improvements in performance on standard benchmarks. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.19883 — 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 2026-10-15T13:53:35.665Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).