Towards Accurate Post-training Quantization for Diffusion Models

TL;DR

This paper introduces a novel data-free post-training quantization method for diffusion models that improves image generation quality by using group-wise quantization and optimal timestep selection, reducing errors significantly.

Contribution

We propose a group-wise quantization framework with optimal timestep sampling for diffusion models, enhancing accuracy without increasing computational cost.

Findings

Outperforms state-of-the-art quantization methods in image quality.

Reduces quantization errors significantly.

Maintains low computational overhead.

Abstract

In this paper, we propose an accurate data-free post-training quantization framework of diffusion models (ADP-DM) for efficient image generation. Conventional data-free quantization methods learn shared quantization functions for tensor discretization regardless of the generation timesteps, while the activation distribution differs significantly across various timesteps. The calibration images are acquired in random timesteps which fail to provide sufficient information for generalizable quantization function learning. Both issues cause sizable quantization errors with obvious image generation performance degradation. On the contrary, we design group-wise quantization functions for activation discretization in different timesteps and sample the optimal timestep for informative calibration image generation, so that our quantized diffusion model can reduce the discretization errors with negligible computational overhead. Specifically, we partition the timesteps according to the importance weights of quantization functions in different groups, which are optimized by differentiable search algorithms. We also select the optimal timestep for calibration image generation by structural risk minimizing principle in order to enhance the generalization ability in the deployment of quantized diffusion model. Extensive experimental results show that our method outperforms the state-of-the-art post-training quantization of diffusion model by a sizable margin with similar computational cost.