Metraipolis › archive › dc_6536089e64d2
When Does a Diffusion Model Decide What to Draw ? — arXiv:2610.05645
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.05645
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:31:11.733Z
- SHA-256 of the archived text
d180befbbcb94d9ce29da7bee1c8531631ef3d28ef1afbaea6662f1a4903684f- Stable URL of this page
https://metraipolis.com/documento/dc_6536089e64d2
How to cite
When Does a Diffusion Model Decide What to Draw ? — arXiv:2610.05645 — arXiv, https://arxiv.org/abs/2610.05645 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6536089e64d2 (Metraipolis archive dc_6536089e64d2, sha256 d180befbbcb94d9ce29da7bee1c8531631ef3d28ef1afbaea6662f1a4903684f)
BibTeX
@misc{dc6536089e64d2,
title = {When Does a Diffusion Model Decide What to Draw ? — arXiv:2610.05645},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.05645},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6536089e64d2; sha256 d180befbbcb94d9ce29da7bee1c8531631ef3d28ef1afbaea6662f1a4903684f},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05645. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
A diffusion model starts from pure noise and removes it step by step. Somewhere along the way it stops being able to become "anything" and becomes committed to, say, a horse rather than a truck. We measure when this happens, and what a trained model gets wrong about it, on CIFAR-10. The most direct measurement is to freeze a half-finished image, restart the generation from that point many times, and count how often each class comes out. We call this probability the committor. Measured this way, the model settles coarse questions (vehicle or animal?) at roughly twice the noise level of fine ones (which animal?). A much cheaper measurement, the noise level at which a classifier's opinion about two classes splits into two distinct groups, gets the order of these decisions right (rank correlation 0.73-0.88) but not their exact timing. We then compare pretrained models with their Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.05645 — 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 2026-10-06T05:27:05.813Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).