the diffusion model has an u-net architecture. the architecture's both input and output are the same image. The image is transformed into a noise image and then restored into the original image. Additionally, the diffusion model connects the image text description layer into the u-net architecture's image restoration process. Is it correct?
'Papers > Machine learning' 카테고리의 다른 글
| diffusion model final explanation (0) | 2025.05.12 |
|---|---|
| openai o1 (0) | 2024.09.24 |
| end-to-end model architecture for scene analysis (0) | 2024.05.19 |
| q-learning vs RLHF (0) | 2024.01.06 |
| Q-Star in openai (0) | 2024.01.06 |