Move all weight initializations from private methods to constructors (#5331)

* Move weight initialization in constructors.

* Fixing mypy for ViT.

* remove unnecessary import

torchvision/models/googlenet.py

@@ -90,9 +90,6 @@ class GoogLeNet(nn.Module):
         self.fc = nn.Linear(1024, num_classes)
 
         if init_weights:
-            self._initialize_weights()
-
-    def _initialize_weights(self) -> None:
             for m in self.modules():
                 if isinstance(m, nn.Conv2d) or isinstance(m, nn.Linear):
                     torch.nn.init.trunc_normal_(m.weight, mean=0.0, std=0.01, a=-2, b=2)

torchvision/models/mnasnet.py

@@ -128,15 +128,7 @@ class MNASNet(torch.nn.Module):
         ]
         self.layers = nn.Sequential(*layers)
         self.classifier = nn.Sequential(nn.Dropout(p=dropout, inplace=True), nn.Linear(1280, num_classes))
-        self._initialize_weights()
 
-    def forward(self, x: Tensor) -> Tensor:
-        x = self.layers(x)
-        # Equivalent to global avgpool and removing H and W dimensions.
-        x = x.mean([2, 3])
-        return self.classifier(x)
-
-    def _initialize_weights(self) -> None:
         for m in self.modules():
             if isinstance(m, nn.Conv2d):
                 nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
@@ -149,6 +141,12 @@ class MNASNet(torch.nn.Module):
                 nn.init.kaiming_uniform_(m.weight, mode="fan_out", nonlinearity="sigmoid")
                 nn.init.zeros_(m.bias)
 
+    def forward(self, x: Tensor) -> Tensor:
+        x = self.layers(x)
+        # Equivalent to global avgpool and removing H and W dimensions.
+        x = x.mean([2, 3])
+        return self.classifier(x)
+
     def _load_from_state_dict(
         self,
         state_dict: Dict,

torchvision/models/optical_flow/raft.py

@@ -132,9 +132,6 @@ class FeatureEncoder(nn.Module):
 
         self.conv = nn.Conv2d(layers[3], layers[4], kernel_size=1)
 
-        self._init_weights()
-
-    def _init_weights(self):
         for m in self.modules():
             if isinstance(m, nn.Conv2d):
                 nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")

torchvision/models/regnet.py

@@ -366,20 +366,6 @@ class RegNet(nn.Module):
         self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
         self.fc = nn.Linear(in_features=current_width, out_features=num_classes)
 
-        # Init weights and good to go
-        self._reset_parameters()
-
-    def forward(self, x: Tensor) -> Tensor:
-        x = self.stem(x)
-        x = self.trunk_output(x)
-
-        x = self.avgpool(x)
-        x = x.flatten(start_dim=1)
-        x = self.fc(x)
-
-        return x
-
-    def _reset_parameters(self) -> None:
         # Performs ResNet-style weight initialization
         for m in self.modules():
             if isinstance(m, nn.Conv2d):
@@ -393,6 +379,16 @@ class RegNet(nn.Module):
                 nn.init.normal_(m.weight, mean=0.0, std=0.01)
                 nn.init.zeros_(m.bias)
 
+    def forward(self, x: Tensor) -> Tensor:
+        x = self.stem(x)
+        x = self.trunk_output(x)
+
+        x = self.avgpool(x)
+        x = x.flatten(start_dim=1)
+        x = self.fc(x)
+
+        return x
+
 
 def _regnet(arch: str, block_params: BlockParams, pretrained: bool, progress: bool, **kwargs: Any) -> RegNet:
     norm_layer = kwargs.pop("norm_layer", partial(nn.BatchNorm2d, eps=1e-05, momentum=0.1))

torchvision/models/vgg.py

@@ -50,16 +50,6 @@ class VGG(nn.Module):
             nn.Linear(4096, num_classes),
         )
         if init_weights:
-            self._initialize_weights()
-
-    def forward(self, x: torch.Tensor) -> torch.Tensor:
-        x = self.features(x)
-        x = self.avgpool(x)
-        x = torch.flatten(x, 1)
-        x = self.classifier(x)
-        return x
-
-    def _initialize_weights(self) -> None:
             for m in self.modules():
                 if isinstance(m, nn.Conv2d):
                     nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
@@ -72,6 +62,13 @@ class VGG(nn.Module):
                     nn.init.normal_(m.weight, 0, 0.01)
                     nn.init.constant_(m.bias, 0)
 
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = self.features(x)
+        x = self.avgpool(x)
+        x = torch.flatten(x, 1)
+        x = self.classifier(x)
+        return x
+
 
 def make_layers(cfg: List[Union[str, int]], batch_norm: bool = False) -> nn.Sequential:
     layers: List[nn.Module] = []

torchvision/models/video/resnet.py

@@ -223,7 +223,17 @@ class VideoResNet(nn.Module):
         self.fc = nn.Linear(512 * block.expansion, num_classes)
 
         # init weights
-        self._initialize_weights()
+        for m in self.modules():
+            if isinstance(m, nn.Conv3d):
+                nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
+                if m.bias is not None:
+                    nn.init.constant_(m.bias, 0)
+            elif isinstance(m, nn.BatchNorm3d):
+                nn.init.constant_(m.weight, 1)
+                nn.init.constant_(m.bias, 0)
+            elif isinstance(m, nn.Linear):
+                nn.init.normal_(m.weight, 0, 0.01)
+                nn.init.constant_(m.bias, 0)
 
         if zero_init_residual:
             for m in self.modules():
@@ -270,19 +280,6 @@ class VideoResNet(nn.Module):
 
         return nn.Sequential(*layers)
 
-    def _initialize_weights(self) -> None:
-        for m in self.modules():
-            if isinstance(m, nn.Conv3d):
-                nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
-                if m.bias is not None:
-                    nn.init.constant_(m.bias, 0)
-            elif isinstance(m, nn.BatchNorm3d):
-                nn.init.constant_(m.weight, 1)
-                nn.init.constant_(m.bias, 0)
-            elif isinstance(m, nn.Linear):
-                nn.init.normal_(m.weight, 0, 0.01)
-                nn.init.constant_(m.bias, 0)
-
 
 def _video_resnet(arch: str, pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VideoResNet:
     model = VideoResNet(**kwargs)

torchvision/models/vision_transformer.py

@@ -44,9 +44,7 @@ class MLPBlock(nn.Sequential):
         self.dropout_1 = nn.Dropout(dropout)
         self.linear_2 = nn.Linear(mlp_dim, in_dim)
         self.dropout_2 = nn.Dropout(dropout)
-        self._init_weights()
 
-    def _init_weights(self):
         nn.init.xavier_uniform_(self.linear_1.weight)
         nn.init.xavier_uniform_(self.linear_2.weight)
         nn.init.normal_(self.linear_1.bias, std=1e-6)
@@ -211,26 +209,27 @@ class VisionTransformer(nn.Module):
             heads_layers["head"] = nn.Linear(representation_size, num_classes)
 
         self.heads = nn.Sequential(heads_layers)
-        self._init_weights()
 
-    def _init_weights(self):
         if isinstance(self.conv_proj, nn.Conv2d):
             # Init the patchify stem
             fan_in = self.conv_proj.in_channels * self.conv_proj.kernel_size[0] * self.conv_proj.kernel_size[1]
             nn.init.trunc_normal_(self.conv_proj.weight, std=math.sqrt(1 / fan_in))
+            if self.conv_proj.bias is not None:
                 nn.init.zeros_(self.conv_proj.bias)
-        else:
+        elif self.conv_proj.conv_last is not None and isinstance(self.conv_proj.conv_last, nn.Conv2d):
             # Init the last 1x1 conv of the conv stem
             nn.init.normal_(
                 self.conv_proj.conv_last.weight, mean=0.0, std=math.sqrt(2.0 / self.conv_proj.conv_last.out_channels)
             )
+            if self.conv_proj.conv_last.bias is not None:
                 nn.init.zeros_(self.conv_proj.conv_last.bias)
 
-        if hasattr(self.heads, "pre_logits"):
+        if hasattr(self.heads, "pre_logits") and isinstance(self.heads.pre_logits, nn.Linear):
             fan_in = self.heads.pre_logits.in_features
             nn.init.trunc_normal_(self.heads.pre_logits.weight, std=math.sqrt(1 / fan_in))
             nn.init.zeros_(self.heads.pre_logits.bias)
 
+        if isinstance(self.heads.head, nn.Linear):
             nn.init.zeros_(self.heads.head.weight)
             nn.init.zeros_(self.heads.head.bias)