Metraipolis › archive › dc_bcc1bbac4c95

SCOPE: Semantic Bounded Credit Optimization for Prompt Engineering in Text-to-image Generation — arXiv:2610.22264

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.22264
License (as archived)
CC0-1.0
Archived
2026-10-16T10:36:48.883Z
SHA-256 of the archived text
01c255008ff493fc3ed930b5278428d4df891ea7c68a3657432b27950af7df73
Stable URL of this page
https://metraipolis.com/documento/dc_bcc1bbac4c95

How to cite

SCOPE: Semantic Bounded Credit Optimization for Prompt Engineering in Text-to-image Generation — arXiv:2610.22264 — arXiv (en), https://arxiv.org/abs/2610.22264 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bcc1bbac4c95 (Metraipolis archive dc_bcc1bbac4c95, sha256 01c255008ff493fc3ed930b5278428d4df891ea7c68a3657432b27950af7df73)
BibTeX
@misc{dcbcc1bbac4c95,
  title = {SCOPE: Semantic Bounded Credit Optimization for Prompt Engineering in Text-to-image Generation — arXiv:2610.22264},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.22264},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bcc1bbac4c95; sha256 01c255008ff493fc3ed930b5278428d4df891ea7c68a3657432b27950af7df73},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.22264. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reinforcement learning (RL) enables language models to improve text-to-image (T2I) generation by rewriting prompts for a frozen image generator. Existing methods typically apply a single signal derived from the image to individual subword tokens, although objects, counts, attributes, and relations form coherent semantic spans. We propose SCOPE, an objective based on Group Relative Policy Optimization (GRPO) that aligns update weighting with this structure. The rewriter jointly generates an enhanced prompt and its semantic partition. SCOPE pools shifts in policy log probabilities within each span to derive a shared coefficient for its token losses while retaining a single advantage for each completion. It requires neither separate span rewards nor additional image renders. Across Qwen-Image, SD3, and FLUX.1-dev, SCOPE achieves average relative gains over vanilla GRPO of 9.48% on GeneVal and 7.62% on T2I-CompBench. Ablation studies show that both semantic boundaries and pooling within spans contribute to these gains, supporting semantic structure as a practical inductive bias for prompt engineering with RL.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.22264 — 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 2026-10-16T08:52:02.728Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).