def forward()

in models/modules.py [0:0]

• 51 lines of code
• 12 McCabe index (conditional complexity)

    def forward(self, input):
        weight, bias = self.get_weight()
        c = input.shape[1]
        if (
            self.mode == "BatchNorm"
            and self.width_factors_list is not None
            and self.width_factor in self.width_factors_list
        ):
            # Normally, we expect width_factors_list to be empty, because we
            # only want to use it if we are running sanity checks (e.g.
            # recreating the original performance or something).
            idx = self.width_factors_list.index(self.width_factor)
            kwargs = {
                "input": input,
                "running_mean": self.bn[idx].running_mean[:c],
                "running_var": self.bn[idx].running_var[:c],
                "weight": weight[:c],
                "bias": bias[:c],
                "training": self.training,
                "momentum": self.momentum,
                "eps": self.eps,
            }
        elif self.mode in ("InstanceNorm", "BatchNorm"):
            # Sanity check, since we're not tracking running stats.
            running_mean = self.running_mean
            if self.running_mean is not None:
                running_mean = running_mean[:c]

            running_var = self.running_var
            if self.running_var is not None:
                running_var = running_var[:c]

            kwargs = {
                "input": input,
                "running_mean": running_mean,
                "running_var": running_var,
                "weight": weight[:c],
                "bias": bias[:c],
                "momentum": self.momentum,
                "eps": self.eps,
            }

            if self.mode == "BatchNorm":
                kwargs["training"] = self.training

        elif self.mode == "GroupNorm":
            num_groups = self.num_groups
            if num_groups == "full":
                num_groups = c
            kwargs = {
                "input": input,
                "num_groups": num_groups,
                "weight": weight[:c],
                "bias": bias[:c],
                "eps": self.eps,
            }
        else:
            raise NotImplementedError(f"Invalid mode {self.mode}.")

        return self.bn_func(**kwargs)