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