Forward function.
(self, x)
| 657 | param.requires_grad = False |
| 658 | |
| 659 | def forward(self, x): |
| 660 | """Forward function.""" |
| 661 | if self.deep_stem: |
| 662 | x = self.stem(x) |
| 663 | else: |
| 664 | x = self.conv1(x) |
| 665 | x = self.norm1(x) |
| 666 | x = self.relu(x) |
| 667 | x = self.maxpool(x) |
| 668 | outs = [] |
| 669 | for i, layer_name in enumerate(self.res_layers): |
| 670 | res_layer = getattr(self, layer_name) |
| 671 | x = res_layer(x) |
| 672 | if i in self.out_indices: |
| 673 | outs.append(x) |
| 674 | return tuple(outs) |
| 675 | |
| 676 | def train(self, mode=True): |
| 677 | """Convert the model into training mode while keep normalization layer |