Metraipolis › archive › dc_5f732745bb28
VaLoP: Value Gradient Lookahead and Pullback for Reward-Guided Diffusion — arXiv:2610.35607
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.35607
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-27T07:59:12.122Z
- SHA-256 of the archived text
e7f2e99b9396d3c5f0e944021e0a60c4969204df3125f4bedcf1e40478806d7c- Stable URL of this page
https://metraipolis.com/documento/dc_5f732745bb28
How to cite
VaLoP: Value Gradient Lookahead and Pullback for Reward-Guided Diffusion — arXiv:2610.35607 — arXiv (en), https://arxiv.org/abs/2610.35607 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5f732745bb28 (Metraipolis archive dc_5f732745bb28, sha256 e7f2e99b9396d3c5f0e944021e0a60c4969204df3125f4bedcf1e40478806d7c)
BibTeX
@misc{dc5f732745bb28,
title = {VaLoP: Value Gradient Lookahead and Pullback for Reward-Guided Diffusion — arXiv:2610.35607},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.35607},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5f732745bb28; sha256 e7f2e99b9396d3c5f0e944021e0a60c4969204df3125f4bedcf1e40478806d7c},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.35607. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reward guidance steers a pretrained diffusion or flow model toward images or actions that a reward model prefers, at sampling time and without retraining the model for each reward. It follows the gradient of a value function, the expected reward given the noisy state. The two common estimates of this gradient, gradients through a one-step denoised estimate and learned value models, are least accurate at high noise, where the coarse structure of the sample is decided. Interestingly, we find that both errors are posterior covariances, which grow with the uncertainty about the clean sample, and hence at higher noise. To reduce them, we evaluate the value at a less noisy state, one sampler step ahead, where this uncertainty is smaller, and pull its gradient back to the current state. This lookahead places both estimators at the two ends of one family and reveals that an intermediate lookahead requires one additional term, a correction through the generator. We therefore propose Value Gradient Lookahead and Pullback (VaLoP), which computes this correction with one vector-Jacobian product, and VaLoP-Fast, which amortizes it with a small head that, for text-to-image models, is trained with a single reward and reused for the others. Across various models, rewards, and benchmark datasets, both robustly outperform existing value-gradient guidance on the target reward, VaLoP-Fast at about the cost of learned-value guidance. Both extend to critic-guided robot policies, where VaLoP-Fast attains the highest success rates on LIBERO and RoboCasa. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.35607 — 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 2026-10-27T07:55:30.766Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).