Metraipolis › archive › dc_bb05e459bc3a

ReSight-SMC: Two-Stage Power Sampling via Island SMC with Visual Scouts — arXiv:2609.34905

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34905
License (as archived)
CC0-1.0
Archived
2026-10-07T23:52:33.363Z
SHA-256 of the archived text
c8e9c4fc20b9a0fcf80c5e8cc328f04ba12bd50839b805bc65232b7e6555bb36
Stable URL of this page
https://metraipolis.com/documento/dc_bb05e459bc3a

How to cite

ReSight-SMC: Two-Stage Power Sampling via Island SMC with Visual Scouts — arXiv:2609.34905 — arXiv (en), https://arxiv.org/abs/2609.34905 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bb05e459bc3a (Metraipolis archive dc_bb05e459bc3a, sha256 c8e9c4fc20b9a0fcf80c5e8cc328f04ba12bd50839b805bc65232b7e6555bb36)
BibTeX
@misc{dcbb05e459bc3a,
  title = {ReSight-SMC: Two-Stage Power Sampling via Island SMC with Visual Scouts — arXiv:2609.34905},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34905},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bb05e459bc3a; sha256 c8e9c4fc20b9a0fcf80c5e8cc328f04ba12bd50839b805bc65232b7e6555bb36},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34905. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Power sampling has emerged as a training-free approach to LLM reasoning, eliciting capabilities comparable to reinforcement learning by sharpening the model distribution over complete responses. Despite this success, power sampling remains underexplored in large vision-language models (LVLMs). We transfer Power-SMC to LVLM decoding by defining a sequence-power target conditioned on both the image and the prompt. This direct transfer provides a strong training-free baseline, but leaves two aspects of finite-particle multimodal inference unaddressed. At the particle level, global resampling can collapse genealogies, while particle-based power sampling does not diversify trajectories through distinct visual cues in multimodal decoding, limiting exploration under a finite particle budget. At the answer level, sequence-level sharpening makes distinct reasoning trajectories compete even when they support the same answer. We introduce ReSight-SMC, a verifier-free two-stage power sampler for LVLM inference. Its first stage uses ancestry-isolated SMC islands to preserve independent trajectory families and routes a bounded set of prefix-conditioned visual scouts to prefix-relevant image regions while discouraging redundant overlap. Each scout temporarily increases attention to the image tokens and emphasizes its routed region. Exact importance correction preserves the base LVLM sequence-power target. The second stage aggregates terminal importance mass by canonical answer, powers the answer marginal, and samples an answer together with a supporting trajectory. Across four LVLM backbones and five benchmarks, ReSight-SMC achieves stronger aggregate performance than Power-SMC over both the reasoning and perception benchmark groups. Without post-training, it remains competitive in aggregate with backbone-matched models trained using reinforcement learning.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.34905 — 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 2026-10-07T23:29:27.498Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).