arXiv cs.AI· Zihe Huang, Sirui Zhou, Haozhi Han, Ruge Zhang, Teng Ma, Tao Ma, Zheng Liu, Ce Zhang, Ting Cao, Yunxin Liu, Yunquan Zhang, Yifeng Chen, Kun Li·· 1 小时前AI 评分48
F³:什么才是世界动作模型的好经验?
F$^3$: What Makes Good Experience in World-Action Models?
AI 导读
论文提出 F³ 框架,研究世界动作模型(WAM)如何利用自身经验,指出经验的价值取决于存什么、何时用、如何注入,并用 Form、Fit、Fusion 三个机制完成这三项决策。
正文 · 原文
Authors:Zihe Huang, Sirui Zhou, Haozhi Han, Ruge Zhang, Teng Ma, Tao Ma, Zheng Liu, Ce Zhang, Ting Cao, Yunxin Liu, Yunquan Zhang, Yifeng Chen, Kun Li
Abstract:Deployed world-action models (WAMs) can keep improving only if they make good use of their own experience. Existing methods supply retrieved rollouts or demonstrations to the policy as context, yet such experience often turns previously solved tasks into failures. What, then, makes good experience for a WAM? We show that experience is valuable only in use: its value depends not on whether the source rollout succeeded but on what is stored, when it is used, and where it enters the policy. We therefore propose F$^3$, which makes these three decisions with Form, Fit, and Fusion. Form compresses successful and failed rollouts alike into short, verified operation records; Fit teaches the policy to read a record and to predict when it helps; Fusion injects the record's images, text and actions into the policy as a prefix and an action prior. Trained once on only 200 examples and deployed with a growing experience library, the model achieves state-of-the-art results on MolmoSpaces and RoboLab without any training on RoboLab: over native Cosmos 3 Nano, it raises success by 12.0 pp and 10.8 pp, reduces steps to success by 12.2% and 11.3%, and cuts regressions by 57%.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.24708 [cs.AI] |
| (or arXiv:2611.24708v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.24708 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zihe Huang [view email]
[v1]
Wed, 18 Nov 2026 16:33:15 UTC (8,286 KB)
来源:arXiv cs.AI · arxiv.org