Metraipolis › archive › dc_cecc0186dea6
MA-FPPO: Multi-Agent Flow-Pretrained Policy Optimization — arXiv:2609.32594
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32594
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:55:45.826Z
- SHA-256 of the archived text
b0c25b7ab3035f28739c6097ead63e080bab1f5201add053dade99c3b87dca8f- Stable URL of this page
https://metraipolis.com/documento/dc_cecc0186dea6
How to cite
MA-FPPO: Multi-Agent Flow-Pretrained Policy Optimization — arXiv:2609.32594 — arXiv (en), https://arxiv.org/abs/2609.32594 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cecc0186dea6 (Metraipolis archive dc_cecc0186dea6, sha256 b0c25b7ab3035f28739c6097ead63e080bab1f5201add053dade99c3b87dca8f)
BibTeX
@misc{dccecc0186dea6,
title = {MA-FPPO: Multi-Agent Flow-Pretrained Policy Optimization — arXiv:2609.32594},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32594},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cecc0186dea6; sha256 b0c25b7ab3035f28739c6097ead63e080bab1f5201add053dade99c3b87dca8f},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32594. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multi-agent flow policies learn cooperative behavior from fixed offline datasets, but often struggle to complete tasks in situations not covered by the offline data. In these situations, agents must both adapt to changes in the environment and coordinate with one another, yet action patterns learned offline are often insufficient for effective adaptation and coordination. To address this problem, we propose Multi-Agent Flow-Pretrained Policy Optimization (MA-FPPO), which uses online fine-tuning to improve the cooperative behavior of models pretrained with flow matching through new interactions with the environment. Building on the behavior learned during pretraining, we construct policies with explicit action likelihoods for discrete and continuous action spaces. We then update the pretrained model using shared team advantages to further improve coordination based on team performance. Our method achieves, on average, relative gains of 52.8% over the strongest listed offline baselines across 30 settings and 29.8% over purely online learning across 38 comparisons with matched online budgets and evaluation protocols. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.32594 — 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 2026-10-07T20:29:00.882Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).