Metraipolis › archive › dc_49a877ddd78b

VectraYX-Vision-1B: A Sub-2B Spanish/LATAM Cybersecurity Vision-Language Model, and What Limits Its Visual Grounding — arXiv:2608.08477

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.08477
License (as archived)
CC0-1.0
Archived
2026-10-19T08:49:59.672Z
SHA-256 of the archived text
78da651a233f8907587a1bd565449f1568f3e3f74788e57c405f6a433d92d2f7
Stable URL of this page
https://metraipolis.com/documento/dc_49a877ddd78b

How to cite

VectraYX-Vision-1B: A Sub-2B Spanish/LATAM Cybersecurity Vision-Language Model, and What Limits Its Visual Grounding — arXiv:2608.08477 — arXiv (en), https://arxiv.org/abs/2608.08477 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_49a877ddd78b (Metraipolis archive dc_49a877ddd78b, sha256 78da651a233f8907587a1bd565449f1568f3e3f74788e57c405f6a433d92d2f7)
BibTeX
@misc{dc49a877ddd78b,
  title = {VectraYX-Vision-1B: A Sub-2B Spanish/LATAM Cybersecurity Vision-Language Model, and What Limits Its Visual Grounding — arXiv:2608.08477},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.08477},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_49a877ddd78b; sha256 78da651a233f8907587a1bd565449f1568f3e3f74788e57c405f6a433d92d2f7},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.08477. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We build VectraYX-Vision-1B, a sub-2B Spanish/LATAM cybersecurity vision-language model for offline use, and measure what limits its visual grounding. A frozen 1.04B-parameter decoder is coupled to a frozen vision encoder through a trainable projector: first SigLIP, then the Qwen2-VL-2B tower with a projector shaped for this http URL 's mmproj export. With SigLIP, after five fine-tuning defects were repaired, a nine-field extraction gate with a shuffled-image control passes the same 2/9 fields under every configuration that keeps the encoder and adds no text hint. A frozen-feature probe explains why. Transplanting the Qwen2-VL tower with its own merger reads an 8-nibble address SigLIP never read (0.00 to 0.81 exact match). On B8 (2,040 items, 34 fields, 16 templates, shuffled-image and best-constant controls), 9 fields pass on a Qwen-tower checkpoint, all within trained template-field pairs. Screenshot tool identification (B6) sat at exactly 0.000. That was not a perception ceiling. Stock Qwen2-VL-2B transcribes the same images at word recall 0.93 at our pixel budget (0.52 with our letterboxing). Training only the projector on B6's task format and render family lifts B6 tool identification to 0.96 and recall to 0.48-0.51. Perception of rendered text is therefore present and trainable in this frozen-backbone design. The evidence is narrow. B6 is nearly in-distribution for that projector, which also forgets part of B8 (mean field accuracy 0.53 to 0.29 on a retention check); runs are single-seed, and no checkpoint yet has both. We document four harness defects, retract an earlier B6 score, and release code, benchmarks and checkpoints.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-11 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.08477 — 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 2026-10-19T08:49:41.860Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).