| 108 | |
| 109 | class Encoder(nn.Module): |
| 110 | def __init__( |
| 111 | self, |
| 112 | resolution: int, |
| 113 | in_channels: int, |
| 114 | ch: int, |
| 115 | ch_mult: list[int], |
| 116 | num_res_blocks: int, |
| 117 | z_channels: int, |
| 118 | ): |
| 119 | super().__init__() |
| 120 | self.ch = ch |
| 121 | self.num_resolutions = len(ch_mult) |
| 122 | self.num_res_blocks = num_res_blocks |
| 123 | self.resolution = resolution |
| 124 | self.in_channels = in_channels |
| 125 | # downsampling |
| 126 | self.conv_in = nn.Conv2d(in_channels, self.ch, kernel_size=3, stride=1, padding=1) |
| 127 | |
| 128 | curr_res = resolution |
| 129 | in_ch_mult = (1,) + tuple(ch_mult) |
| 130 | self.in_ch_mult = in_ch_mult |
| 131 | self.down = nn.ModuleList() |
| 132 | block_in = self.ch |
| 133 | for i_level in range(self.num_resolutions): |
| 134 | block = nn.ModuleList() |
| 135 | attn = nn.ModuleList() |
| 136 | block_in = ch * in_ch_mult[i_level] |
| 137 | block_out = ch * ch_mult[i_level] |
| 138 | for _ in range(self.num_res_blocks): |
| 139 | block.append(ResnetBlock(in_channels=block_in, out_channels=block_out)) |
| 140 | block_in = block_out |
| 141 | down = nn.Module() |
| 142 | down.block = block |
| 143 | down.attn = attn |
| 144 | if i_level != self.num_resolutions - 1: |
| 145 | down.downsample = Downsample(block_in) |
| 146 | curr_res = curr_res // 2 |
| 147 | self.down.append(down) |
| 148 | |
| 149 | # middle |
| 150 | self.mid = nn.Module() |
| 151 | self.mid.block_1 = ResnetBlock(in_channels=block_in, out_channels=block_in) |
| 152 | self.mid.attn_1 = AttnBlock(block_in) |
| 153 | self.mid.block_2 = ResnetBlock(in_channels=block_in, out_channels=block_in) |
| 154 | |
| 155 | # end |
| 156 | self.norm_out = nn.GroupNorm(num_groups=32, num_channels=block_in, eps=1e-6, affine=True) |
| 157 | self.conv_out = nn.Conv2d(block_in, 2 * z_channels, kernel_size=3, stride=1, padding=1) |
| 158 | |
| 159 | def forward(self, x: Tensor) -> Tensor: |
| 160 | # downsampling |