ResNet_32.py 文件源码

def h(self, x, train, finetune):
        # First convolution layer.
        h = self[0](x)

        h = F.dropout(h, ratio=self.dropout, train=train)

        # Residual blocks.
        for i in range(1, len(self) - 2):
            h = self[i](h, train, finetune)

        # Batch normalization.
        h = self[-2](h, test=not train, finetune=finetune)
        h = F.relu(h)

        # Average pooling.
        h = F.max_pooling_2d(h, ksize=2, pad=0)

        # Prediction layer 5.
        h = self[-1](h)
        h = F.reshape(h, (h.data.shape[0], 5))

        return h