Forward function.
(self, x)
| 600 | param.requires_grad = False |
| 601 | |
| 602 | def forward(self, x): |
| 603 | """Forward function.""" |
| 604 | if self.deep_stem: |
| 605 | x = self.stem(x) |
| 606 | else: |
| 607 | x = self.conv1(x) |
| 608 | x = self.norm1(x) |
| 609 | x = self.relu(x) |
| 610 | x = self.maxpool(x) |
| 611 | outs = [] |
| 612 | for i, layer_name in enumerate(self.res_layers): |
| 613 | res_layer = getattr(self, layer_name) |
| 614 | x = res_layer(x) |
| 615 | if i in self.out_indices: |
| 616 | outs.append(x) |
| 617 | return tuple(outs) |
| 618 | |
| 619 | def train(self, mode=True): |
| 620 | """Convert the model into training mode while keep normalization layer |