__init__() takes from 1 to 2 positional arguments but 3 were given

调用EntropyMaximization类计算loss时报错，class定义如下

class EntropyMaximization(torch.nn.Module):
    """Entropy Maximization loss

    Arguments:
        t : temperature
    """

    def __init__(self, t=1.):
        super(EntropyMaximization, self).__init__()
        self.t = t

    def forward(self, lbl, pred):
        """Compute loss.

        Arguments:
            lbl (torch.tensor:float): predictions, not confidence, not label.
            pred (torch.tensor:float): predictions.

        Returns:
            loss (torch.tensor:float): entropy maximization loss

        """
        loss = torch.mean(torch.sum(F.softmax(lbl/self.t, dim=-1) * F.log_softmax(pred/self.t, dim=-1), dim=-1))
        return loss

错误代码：

Lf = EntropyMaximization(logits_leaf, logits_leaf)

正确代码：

loss_enm = EntropyMaximization()
Lf = loss_enm(logits_leaf, logits_leaf)

通常在使用 PyTorch 中的自定义模型或损失函数时，你需要先实例化类，然后才能调用其中的方法，比如 forward 方法