Method __init__

diffusers/src/diffusers/models/autoencoders/vae.py:841–875

(
        self,
        in_channels: int,
        out_channels: int,
        num_blocks: Tuple[int, ...],
        block_out_channels: Tuple[int, ...],
        act_fn: str,
    )

    """

    def __init__(
        self,
        in_channels: int,
        out_channels: int,
        num_blocks: Tuple[int, ...],
        block_out_channels: Tuple[int, ...],
        act_fn: str,
    ):
        super().__init__()

        layers = []
        for i, num_block in enumerate(num_blocks):
            num_channels = block_out_channels[i]

            if i == 0:
                layers.append(nn.Conv2d(in_channels, num_channels, kernel_size=3, padding=1))
            else:
                layers.append(
                    nn.Conv2d(
                        num_channels,
                        num_channels,
                        kernel_size=3,
                        padding=1,
                        stride=2,
                        bias=False,
                    )
                )

            for _ in range(num_block):
                layers.append(AutoencoderTinyBlock(num_channels, num_channels, act_fn))

        layers.append(nn.Conv2d(block_out_channels[-1], out_channels, kernel_size=3, padding=1))

        self.layers = nn.Sequential(*layers)
        self.gradient_checkpointing = False

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        r"""The forward method of the `EncoderTiny` class."""