MPQ-Diff: Mixed Precision Quantization for Diffusion Models

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Maruzzelli, Rocco Manz, Lewandowski, Basile, Chen, Lydia Y.
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915041994866688
author Maruzzelli, Rocco Manz
Lewandowski, Basile
Chen, Lydia Y.
author_facet Maruzzelli, Rocco Manz
Lewandowski, Basile
Chen, Lydia Y.
contents Diffusion models (DMs) generate remarkable high quality images via the stochastic denoising process, which unfortunately incurs high sampling time. Post-quantizing the trained diffusion models in fixed bit-widths, e.g., 4 bits on weights and 8 bits on activation, is shown effective in accelerating sampling time while maintaining the image quality. Motivated by the observation that the cross-layer dependency of DMs vary across layers and sampling steps, we propose a mixed precision quantization scheme, MPQ-Diff, which allocates different bit-width to the weights and activation of the layers. We advocate to use the cross-layer correlation of a given layer, termed network orthogonality metric, as a proxy to measure the relative importance of a layer per sampling step. We further adopt a uniform sampling scheme to avoid the excessive profiling overhead of estimating orthogonality across all time steps. We evaluate the proposed mixed-precision on LSUN and ImageNet, showing a significant improvement in FID from 65.73 to 15.39, and 52.66 to 14.93, compared to their fixed precision quantization, respectively.
format Preprint
id arxiv_https___arxiv_org_abs_2412_00144
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MPQ-Diff: Mixed Precision Quantization for Diffusion Models
Maruzzelli, Rocco Manz
Lewandowski, Basile
Chen, Lydia Y.
Computer Vision and Pattern Recognition
Machine Learning
Diffusion models (DMs) generate remarkable high quality images via the stochastic denoising process, which unfortunately incurs high sampling time. Post-quantizing the trained diffusion models in fixed bit-widths, e.g., 4 bits on weights and 8 bits on activation, is shown effective in accelerating sampling time while maintaining the image quality. Motivated by the observation that the cross-layer dependency of DMs vary across layers and sampling steps, we propose a mixed precision quantization scheme, MPQ-Diff, which allocates different bit-width to the weights and activation of the layers. We advocate to use the cross-layer correlation of a given layer, termed network orthogonality metric, as a proxy to measure the relative importance of a layer per sampling step. We further adopt a uniform sampling scheme to avoid the excessive profiling overhead of estimating orthogonality across all time steps. We evaluate the proposed mixed-precision on LSUN and ImageNet, showing a significant improvement in FID from 65.73 to 15.39, and 52.66 to 14.93, compared to their fixed precision quantization, respectively.
title MPQ-Diff: Mixed Precision Quantization for Diffusion Models
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2412.00144