Elucidating the Design Space of Diffusion-Based Generative Models 论文精读
关注B站可以观看更多实战教学视频:hallo128的个人空间
【更新中】EDM论文精读
论文链接
(1)论文:Elucidating the Design Space of Diffusion-Based Generative Models
(2)引用: Karras T, Aittala M, Aila T, et al. Elucidating the design space of diffusion-based generative models[J]. Advances in Neural Information Processing Systems, 2022, 35: 26565-26577.
(3)论文链接: https://doi.org/10.48550/arXiv.2206.00364
(4)代码链接: https://github.com/NVlabs/edm
附录B
【更新中】EDM论文精读
附录B.1
宋飏Score-Based的SDE形式推导出:
- 扰动核
- 边缘分布
- ODE
附录B.2
(1)随着时间t,不同噪声大小sigma(t)对应的ODE【公式1】
(2)随着时间t,不同缩放大小s(t)对应的ODE【公式4】
【更新中】EDM论文精读