Metraipolis › archive › dc_d7aab05b59eb
Readout Blindness: VLM Scores Miss the Spatial Direction Their Frozen Encoders Retain — arXiv:2610.06324
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.06324
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:32:29.262Z
- SHA-256 of the archived text
a27ccf9a276fa8eed28486c2c6963343839d730230f6bfacc6b043c6ae6fbc80- Stable URL of this page
https://metraipolis.com/documento/dc_d7aab05b59eb
How to cite
Readout Blindness: VLM Scores Miss the Spatial Direction Their Frozen Encoders Retain — arXiv:2610.06324 — arXiv, https://arxiv.org/abs/2610.06324 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d7aab05b59eb (Metraipolis archive dc_d7aab05b59eb, sha256 a27ccf9a276fa8eed28486c2c6963343839d730230f6bfacc6b043c6ae6fbc80)
BibTeX
@misc{dcd7aab05b59eb,
title = {Readout Blindness: VLM Scores Miss the Spatial Direction Their Frozen Encoders Retain — arXiv:2610.06324},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.06324},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d7aab05b59eb; sha256 a27ccf9a276fa8eed28486c2c6963343839d730230f6bfacc6b043c6ae6fbc80},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.06324. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
CLIP-like vision-language models remain a cornerstone of multimodal systems, yet their scores stay near chance on directed spatial relations, such as whether one object is left of another. We call this failure readout blindness and analyze, theoretically and empirically, why deployed scores miss the direction: when scoring rules treat the subject and object symmetrically, direction cancels regardless of encoder training. Guided by this analysis, we introduce Antisymmetric Displacement Readout (ADR), which aligns caption words with image patches in the frozen features and scores each relation by the signed displacement between matched object centroids. Notably, ADR succeeds without additional training or learned parameters, thereby demonstrating that directional information remains in the frozen encoder. However, text and world priors can inflate accuracy, so we further introduce prior deflation, which measures the benefit of the image-text pairing as the grounded gain over a null that pairs each item with an unrelated image. Extensive experiments across encoder families show that ADR substantially improves over deployed scores, which remain near chance on most direction-balanced sets even for fine-tuned encoders. Compared with more complex readouts, ADR outperforms the evaluated MLLM likelihood readouts and is competitive with their chat inference at a small fraction of the computation. These results support our claim that directional information can be recovered from frozen features by an appropriate readout. Our implementation and evaluation kit will be publicly available. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.06324 — 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 2026-10-06T05:27:06.033Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).