Metraipolis › archive › dc_e08050941172

VGAS: Variance-Reduced Guidance and Adaptive Selection for Training-Free Reward Alignment in Discrete Diffusion — arXiv:2608.26585

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.26585
License (as archived)
CC0-1.0
Archived
2026-10-17T16:09:45.393Z
SHA-256 of the archived text
0a03d9a14c72824036c1ec12ca90e90cad6251940e157c257b68816ef5f63b34
Stable URL of this page
https://metraipolis.com/documento/dc_e08050941172

How to cite

VGAS: Variance-Reduced Guidance and Adaptive Selection for Training-Free Reward Alignment in Discrete Diffusion — arXiv:2608.26585 — arXiv (en), https://arxiv.org/abs/2608.26585 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e08050941172 (Metraipolis archive dc_e08050941172, sha256 0a03d9a14c72824036c1ec12ca90e90cad6251940e157c257b68816ef5f63b34)
BibTeX
@misc{dce08050941172,
  title = {VGAS: Variance-Reduced Guidance and Adaptive Selection for Training-Free Reward Alignment in Discrete Diffusion — arXiv:2608.26585},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.26585},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e08050941172; sha256 0a03d9a14c72824036c1ec12ca90e90cad6251940e157c257b68816ef5f63b34},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.26585. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Masked discrete diffusion models perform strongly on text, code, and biological sequences, but their training objective rewards only naturalness, and retraining the generator for every new reward is expensive. Inference-time steering of a frozen model either guides the sampler by the reward gradient or searches over several trajectories, and recent samplers combine the two. Such combinations are assembled as pipelines that leave three choices at their defaults: a guidance estimate resting on one Gumbel draw per sample, a reward tilting placed without reference to the distribution the combination then targets, and a selection temperature held fixed although the spread of per-step rewards drifts. We identify that distribution and settle the three choices against it. We therefore propose Variance-reduced Guidance and Adaptive Selection (VGAS), a simple yet effective inference-time framework that reduces the variance of the guidance estimate for both reward types, applies the reward tilting in the clean-token logits, where the pretrained schedule is preserved, and sets the selection temperature per step. Across regulatory DNA, protein and small-molecule benchmarks, VGAS attains the best training-free reward and matches or surpasses a reward-fine-tuned generator.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-28 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Computational Engineering, Finance, and Science (cs.CE); Quantitative Methods (q-bio.QM); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2608.26585 — 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 2026-10-17T15:34:49.913Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).