Metraipolis › archive › dc_f3c613ac2330
RuleForge: Inferring and Instantiating Latent Visual Rules as Executable Programs — arXiv:2610.38737
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.38737
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-01T02:41:45.258Z
- SHA-256 of the archived text
4a35bb55a16c9ccc5111e7c4e1b87cc45e471dcafc8d0a2d6619ce690f045356- Stable URL of this page
https://metraipolis.com/documento/dc_f3c613ac2330
How to cite
RuleForge: Inferring and Instantiating Latent Visual Rules as Executable Programs — arXiv:2610.38737 — arXiv (en), https://arxiv.org/abs/2610.38737 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f3c613ac2330 (Metraipolis archive dc_f3c613ac2330, sha256 4a35bb55a16c9ccc5111e7c4e1b87cc45e471dcafc8d0a2d6619ce690f045356)
BibTeX
@misc{dcf3c613ac2330,
title = {RuleForge: Inferring and Instantiating Latent Visual Rules as Executable Programs — arXiv:2610.38737},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.38737},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f3c613ac2330; sha256 4a35bb55a16c9ccc5111e7c4e1b87cc45e471dcafc8d0a2d6619ce690f045356},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.38737. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Visual rule reasoning asks whether multimodal models can recover an abstract concept from a few examples. Existing studies either infer rules from visual examples but evaluate only classification or verbalization, or supervise executable generation with rules provided as input. Consequently, the relationship between visual rule understanding and reliable rule instantiation remains insufficiently explored. We present Bongard-Forge100k, a controlled benchmark, together with RuleForge, a backbone-agnostic framework that jointly learns visual rule understanding and rule-conditioned executable generation. The benchmark contains 108,797 problems: 86,000 for training, 19,297 for testing familiar rules, and 3,500 for testing held-out rule compositions. The framework progresses through rule-understanding training, joint understanding and scene-program supervision, and refinement via execution-grounded reinforcement learning with Understanding Replay. Executing generated programs under multiple random seeds provides verifiable feedback on executability, rule satisfaction, diversity, and robustness. With UniLIP as the primary backbone, RuleForge improves multiple-choice accuracy over the base model by 20.93 percentage points on familiar rules and 28.42 percentage points on held-out rule compositions, while increasing scene-level rule satisfaction on familiar rules from 2.91% to 70.99%. Additional experiments on InternVL3.5, Qwen3-VL, and Show-o2 demonstrate the effectiveness of the proposed training data and framework across diverse multimodal architectures. Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2610.38737 — 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 2026-11-01T02:16:43.899Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).