We introduce Adversarial Diffusion Distillation (ADD), a novel training approach that
efficiently samples large-scale foundational image diffusion models in just 1-4 steps while …

We introduce Score identity Distillation (SiD), an innovative data-free method that distills the
generative capabilities of pretrained diffusion models into a single-step generator. SiD not …

Abstract Knowledge distillation methods have recently shown to be a promising direction to
speedup the synthesis of large-scale diffusion models by requiring only a few inference …

This is an expository article on the score-based diffusion models, with a particular focus on
the formulation via stochastic differential equations (SDE). After a gentle introduction, we …

Consistency models, which were proposed to mitigate the high computational overhead
during the sampling phase of diffusion models, facilitate single-step sampling while attaining …

The softmax activation function plays a crucial role in the success of large language models
(LLMs), particularly in the self-attention mechanism of the widely adopted Transformer …

Large generative diffusion models have revolutionized text-to-image generation and offer
immense potential for conditional generation tasks such as image enhancement restoration …

Diffusion models (DMs) have achieved remarkable generative performance particularly with
the introduction of stochastic differential equations (SDEs). Nevertheless a gap emerges in …

Recently, a series of diffusion-aware distillation algorithms have emerged to alleviate the
computational overhead associated with the multi-step inference process of Diffusion …

We propose a method to distill a complex multistep diffusion model into a single-step
conditional GAN student model, dramatically accelerating inference, while preserving image …