### Swish activation function $$x \cdot \sigma(x)$$

### Residual block A residual block has two convolution layers with group normalization. Each resolution is processed with two residua

## U-Net

### Up block This combines `ResidualBlock` and `AttentionBlock`. These are used in the second half of U-Net at each resolution.

### Down block This combines `ResidualBlock` and `AttentionBlock`. These are used in the first half of U-Net at each resolution.

### Scale down the feature map by $\frac{1}{2} \times$

### Middle block It combines a `ResidualBlock`, `AttentionBlock`, followed by another `ResidualBlock`. This block is applied at the lo

### Embeddings for $t$

### Scale up the feature map by $2 \times$