Metraipolis › archive › dc_707b0a45f796
PulseQuant: Propagation-Guided Subspace Correction for 4-Bit Video Diffusion Transformers — arXiv:2609.33384
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.33384
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T23:51:34.356Z
- SHA-256 of the archived text
7bdc464b1f897cab955f5aeb966502141f3cf15ee92fc7efe9b7e369978c0f69- Stable URL of this page
https://metraipolis.com/documento/dc_707b0a45f796
How to cite
PulseQuant: Propagation-Guided Subspace Correction for 4-Bit Video Diffusion Transformers — arXiv:2609.33384 — arXiv (en), https://arxiv.org/abs/2609.33384 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_707b0a45f796 (Metraipolis archive dc_707b0a45f796, sha256 7bdc464b1f897cab955f5aeb966502141f3cf15ee92fc7efe9b7e369978c0f69)
BibTeX
@misc{dc707b0a45f796,
title = {PulseQuant: Propagation-Guided Subspace Correction for 4-Bit Video Diffusion Transformers — arXiv:2609.33384},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.33384},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_707b0a45f796; sha256 7bdc464b1f897cab955f5aeb966502141f3cf15ee92fc7efe9b7e369978c0f69},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33384. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Quantization errors in video diffusion transformers can be amplified or attenuated by subsequent denoising updates, making local reconstruction error an incomplete predictor of final impact. We introduce PulseQuant, a 4-bit post-training quantization method that combines trajectory sensitivity with activation geometry to guide offline calibration. Isolated block--step interventions estimate propagation risk, which prioritizes sensitive trajectory states during row-radius selection. With these radii fixed, response-subspace correction uses neighboring-code edits to reduce residual components along dominant activation directions. Both stages preserve the original 4-bit weight representation. Controlled interventions show that short-horizon propagated error predicts final latent error more reliably than immediate block-output error, supporting calibration beyond local reconstruction objectives. Evaluations on Wan models, Self Forcing, and MiniMax-H3 demonstrate improvements in key consistency and dense-reference metrics while remaining competitive on other attributes across model scales and generation paradigms. 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/2609.33384 — 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 2026-10-07T23:29:27.197Z 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).