Diffusion Model

GitHub - huggingface/diffusers: 🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch and FLAX.

使用Diffusers来实现Stable Diffusion 🧨

“Snippets: Importing libraries”的副本 - Colab

“diffusers_training_example.ipynb”的副本 - Colab

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda.sh

graph TD;
    subgraph "Encoder (Downsampling)"
        A["Input Image"] -->|"DownBlock2D (128)"| B
        B -->|"DownBlock2D (128)"| C
        C -->|"DownBlock2D (256)"| D
        D -->|"DownBlock2D (256)"| E
        E -->|"AttnDownBlock2D (512)"| F
        F -->|"DownBlock2D (512)"| Bottleneck["Bottleneck Layer"]
    end
    
    subgraph "Decoder (Upsampling)"
        Bottleneck -->|"UpBlock2D (512)"| G
        G -->|"AttnUpBlock2D (512)"| H
        H -->|"UpBlock2D (256)"| I
        I -->|"UpBlock2D (256)"| J
        J -->|"UpBlock2D (128)"| K
        K -->|"UpBlock2D (128)"| L
        L -->|"Output Image"| M
    end