Metraipolis › archive › dc_fe2a3016874c

"World Knowledge" in the Weights: Reading Concept Circuits of Vision Transformers — arXiv:2609.09055

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.09055
License (as archived)
CC0-1.0
Archived
2026-10-16T15:12:32.581Z
SHA-256 of the archived text
6d75e57799304af6df787c27ecd52ae9a6c15aa3780c2729f2a712b215506838
Stable URL of this page
https://metraipolis.com/documento/dc_fe2a3016874c

How to cite

"World Knowledge" in the Weights: Reading Concept Circuits of Vision Transformers — arXiv:2609.09055 — arXiv (en), https://arxiv.org/abs/2609.09055 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fe2a3016874c (Metraipolis archive dc_fe2a3016874c, sha256 6d75e57799304af6df787c27ecd52ae9a6c15aa3780c2729f2a712b215506838)
BibTeX
@misc{dcfe2a3016874c,
  title = {"World Knowledge" in the Weights: Reading Concept Circuits of Vision Transformers — arXiv:2609.09055},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.09055},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fe2a3016874c; sha256 6d75e57799304af6df787c27ecd52ae9a6c15aa3780c2729f2a712b215506838},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.09055. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision transformers (ViTs) have achieved remarkable generalization across visual domains, yet little is known about how they internally represent the structure of the world. To address this gap, we use Cross-Layer Transcoders (CLTs) to read concept circuits from ViTs: directed graphs whose nodes correspond to sparse, interpretable concepts and edges capture concept interactions across layers. Our method yields two complementary views of model behavior. The global concept circuit is input-invariant and can be recovered directly from learned cross-layer weights, exposing the reusable "world knowledge" encoded in the model. The instance concept circuit is input-dependent and identifies the concepts and pathways actually used for a specific prediction, enabling faithful example-level explanations. We demonstrate the utility of concept circuits in three ways: (1) Automatic spurious correlation discovery: leveraging the statistics of our global concept circuits to identify shortcut dependencies within the model. (2) Spurious correlation removal: intervening on the instance concept circuit to steer the model towards correct predictions. Empirical results show that our method outperforms existing counterparts by 11.0% on the Waterbird dataset. (3) Model comparison: contrasting the global concept circuits of different foundation models (e.g., CLIP vs. DINO) to reveal how supervision paradigms shape representational structure. Our code is available at this https URL

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.09055 — 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 2026-10-16T14:49:57.389Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).