Awesome

Diffusion 扩散模型

本notebook系列将介绍一种新的生成模型: diffusion 扩散概率模型。

我们将探讨一系列最新 diffusion 的模型以及相关酷炫的应用,将包括如下内容（其中 [x] 表示已经完成）：

• 1 Score matching
• 2 Langevin dynamics
• 3 DPM(2015):Deep unsupervised learning using nonequilibrium thermodynamics
• 4 NCSN(2020): Noise conditional score networks
• 5 DDPM(2020): Denoising Diffusion Probabilistic Models
• 6 WAVEGRAD(2020): ESTIMATING GRADIENTS FOR WAVEFORM GENERATION
• 7 DDIM(2021): DENOISING DIFFUSION IMPLICIT MODELS
• 8 IDDPM(2021): Improved Denoising Diffusion Probabilistic Models
• 9 SDE(2021): SCORE-BASED GENERATIVE MODELING THROUGH STOCHASTIC DIFFERENTIAL EQUATIONS
• 10 Guided Diffusion(2021): Diffusion Models Beat GANs on Image Synthesis
• 11 Classifier Free Diffusion(2021): Classifier-Free Diffusion Guidance
• 12 Latent Diffusion (2022): High-Resolution Image Synthesis with Latent Diffusion Models
• 13 CLIP guided diffusion
• 14 Augmented CLIP Guided Diffusion
• 15 Disco Diffusion
• 16 High Resolution Image Synthesis with Latent Diffusion