Metraipolis › archive › dc_be3f89abac25
PhysFieldBench: Can Multimodal Models Understand Physical Fields? — arXiv:2609.34072
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.34072
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:57:20.438Z
- SHA-256 of the archived text
4e138162cd0f5c6ad074b9fde71f5d9145a5bea4f2fbac5fe1224d83ade0c446- Stable URL of this page
https://metraipolis.com/documento/dc_be3f89abac25
How to cite
PhysFieldBench: Can Multimodal Models Understand Physical Fields? — arXiv:2609.34072 — arXiv (en), https://arxiv.org/abs/2609.34072 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_be3f89abac25 (Metraipolis archive dc_be3f89abac25, sha256 4e138162cd0f5c6ad074b9fde71f5d9145a5bea4f2fbac5fe1224d83ade0c446)
BibTeX
@misc{dcbe3f89abac25,
title = {PhysFieldBench: Can Multimodal Models Understand Physical Fields? — arXiv:2609.34072},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.34072},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_be3f89abac25; sha256 4e138162cd0f5c6ad074b9fde71f5d9145a5bea4f2fbac5fe1224d83ade0c446},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34072. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Multimodal large language models (MLLMs) are increasingly envisioned as core components of scientific and engineering agents, yet their ability to interpret physical fields remains poorly understood. Existing physics benchmarks largely emphasize textbook problem solving or intuitive physical reasoning, leaving open whether MLLMs can infer physically meaningful information from continuous field observations. We introduce PhysFieldBench, a benchmark comprising 24 tasks and 1,160 evaluation examples across controlled equation fields, simulated physical fields, and observed physical fields. The tasks assess three forms of inference: identifying physical mechanisms, comparing latent control variables, and predicting outcome properties. Across representative open-source and proprietary MLLMs, zero-shot performance is low: the best model achieves a chance-normalized score of 29.3, while several open-source models remain near chance. In contrast, a task-specific supervised vision transformer performs substantially better, demonstrating that the inputs contain learnable physical information. To diagnose these failures, a structured self-explanation analysis attributes most errors to missed visual patterns and incorrect visual-to-physical mappings. Further, to explore whether post-training can improve physical inference and generalize to unseen tasks, we compare supervised fine-tuning with final answers or chain-of-thought supervision and reinforcement learning. Final-answer supervision performs best overall but transfers less effectively, whereas reinforcement learning after chain-of-thought supervision achieves the best generalization. Together, these findings highlight the need to improve visual-to-physical grounding and cross-task generalization for MLLMs to reliably interpret physical fields in scientific and engineering workflows. 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/2609.34072 — 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 2026-10-07T20:29:01.415Z 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).