Metraipolis › archive › dc_a78dfaa9970e

Program-Verified Self-Evolution for Vision-Language Models — arXiv:2609.33855

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.33855
License (as archived)
CC0-1.0
Archived
2026-10-07T21:01:50.733Z
SHA-256 of the archived text
67df639d753a1d8daa6d15af69839caa0bed690858213c3b239f701d239dc9f6
Stable URL of this page
https://metraipolis.com/documento/dc_a78dfaa9970e

How to cite

Program-Verified Self-Evolution for Vision-Language Models — arXiv:2609.33855 — arXiv (en), https://arxiv.org/abs/2609.33855 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a78dfaa9970e (Metraipolis archive dc_a78dfaa9970e, sha256 67df639d753a1d8daa6d15af69839caa0bed690858213c3b239f701d239dc9f6)
BibTeX
@misc{dca78dfaa9970e,
  title = {Program-Verified Self-Evolution for Vision-Language Models — arXiv:2609.33855},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33855},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a78dfaa9970e; sha256 67df639d753a1d8daa6d15af69839caa0bed690858213c3b239f701d239dc9f6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33855. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Self-evolving vision-language models train on questions they generate from unlabeled images. Since these questions have no gold answers, prior methods label them by majority vote over sampled answers or by a model judge. In a human evaluation, we find that 24\% of majority-vote labels and 18\% of model-judge labels produced during self-evolution are wrong. To address this problem, we present Verifiable QA Generation for Self-Evolving Models (VQS), which changes how the model judges answers. Instead of voting on an answer, the model parses each image into a structured record, such as a scene graph, a chart table, or a diagram graph. Fixed programs then write a question from the record and compute its answer. The model still acts as a visual checker, but it only confirms the individual facts the program reads, one short claim at a time. These claim-level checks select the parser's training targets, so the parser also improves without labels. Human raters find 94\% of VQS answers correct, against 76\% for majority voting. Across ten benchmarks, VQS improves Qwen3-VL by up to 3.18 points at the 2B, 4B, and 8B scales and outperforms the strongest self-evolving baseline at each. Gains keep growing over three training rounds, reaching 3.84 points at 2B. Code is released at this https URL

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE).
Pagina del lavoro: https://arxiv.org/abs/2609.33855 — 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 2026-10-07T20:29:02.769Z 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).