lazily create up_descending after state dict is already loaded, but o…

…nly do it once

flux/modules/autoencoder.py

@@ -106,16 +106,13 @@ def forward(self, x: Tensor):
         return x
 
 class DownBlock(nn.Module):
-    def __init__(self, block, downsample=None):
+    def __init__(self, block: list, downsample: nn.Module) -> None:
         super().__init__()
+        # we're doing this instead of a flat nn.Sequential to preserve the keys "block" "downsample"
         self.block = nn.Sequential(*block)
-        # attn
-        if downsample:
-            self.downsample = downsample
-        else:
-            self.downsample = nn.Identity()
+        self.downsample = downsample
 
-    def forward(self, x):
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
         return self.downsample(self.block(x))
 
 
@@ -143,8 +140,9 @@ def __init__(
         self.in_ch_mult = in_ch_mult
         down_layers = []
         block_in = self.ch
+        # ideally, this would all append to a single flat nn.Sequential
+        # we cannot do this due to the existing state dict keys
         for i_level in range(self.num_resolutions):
-            # attn = nn.ModuleList()
             block_in = ch * in_ch_mult[i_level]
             block_out = ch * ch_mult[i_level]
             block_layers = []
@@ -156,7 +154,7 @@ def __init__(
                 downsample = Downsample(block_in)
                 curr_res = curr_res // 2
             else:
-                downsample = None
+                downsample = nn.Identity()
             down_layers.append(DownBlock(block_layers, downsample))
         self.down = nn.Sequential(*down_layers)
 
@@ -186,17 +184,13 @@ def forward(self, x: Tensor) -> Tensor:
         return h
 
 class UpBlock(nn.Module):
-    def __init__(self, block, upsample=None):
+    def __init__(self, block: list, upsample: nn.Module) -> None:
         super().__init__()
         self.block = nn.Sequential(*block)
-        if upsample:
-            self.upsample = upsample
-        else:
-            self.upsample = nn.Identity()
+        self.upsample = upsample
 
-    def forward(self, x):
-        x = self.block(x)
-        return self.upsample(x)
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return self.upsample(self.block(x))
 
 
 class Decoder(nn.Module):
@@ -234,7 +228,7 @@ def __init__(
 
         # upsampling
         up_blocks = []
-        # 3, 2, 1, 0
+        # 3, 2, 1, 0, descending order
         for i_level in reversed(range(self.num_resolutions)):
             level_blocks = []
             block_out = ch * ch_mult[i_level]
@@ -245,16 +239,25 @@ def __init__(
                 upsample = Upsample(block_in)
                 curr_res = curr_res * 2
             else:
-                upsample = None
-            # ??? gross
-            # 0, 1, 2, 3
+                upsample = nn.Identity()
+            # 0, 1, 2, 3, ascending order
             up_blocks.insert(0, UpBlock(level_blocks, upsample)) # prepend to get consistent order
         self.up = nn.Sequential(*up_blocks)
 
         # end
         self.norm_out = nn.GroupNorm(num_groups=32, num_channels=block_in, eps=1e-6, affine=True)
         self.conv_out = nn.Conv2d(block_in, out_ch, kernel_size=3, stride=1, padding=1)
 
+    # this is a hack to get something like property but only evaluate it once
+    # we're doing it like this so that up_descending isn't in the state_dict keys
+    # without adding anything conditional to the main flow
+    def __getattr__(self, name):
+        if name == "up_descending":
+            self.up_descending = nn.Sequential(*reversed(self.up))
+            Decoder.__getattr__ = nn.Module.__getattr__
+            return self.up_descending
+        return super().__getattr__(name)
+
     def forward(self, z: Tensor) -> Tensor:
         # z to block_in
         h = self.conv_in(z)
@@ -265,7 +268,7 @@ def forward(self, z: Tensor) -> Tensor:
         h = self.mid.block_2(h)
 
         # upsampling
-        h = self.up(h)
+        h = self.up_descending(h)
 
         # end
         h = self.norm_out(h)