python - 我在哪里可以看到 pytorch 的 MSELoss 的源代码？

Question

我使用 U-NET 网络来训练我的数据。但是我需要修改它的损失函数来减少1以下像素的损失，以减少负例对网络权重的影响。但是我在pycharm MSELOSS中打开了源代码，看到这个：

class MSELoss(_Loss):
    r"""Creates a criterion that measures the mean squared error between
    `n` elements in the input `x` and target `y`:

    :math:`{loss}(x, y)  = 1/n \sum |x_i - y_i|^2`

    `x` and `y` arbitrary shapes with a total of `n` elements each.

    The sum operation still operates over all the elements, and divides by `n`.

    The division by `n` can be avoided if one sets the internal variable
    `size_average` to `False`.

    """
    pass

我得不到任何有用的东西。

Answer 1

你去：https ://github.com/pytorch/pytorch/blob/master/torch/nn/functional.py#L1423但是，它调用C api

def mse_loss(input, target, size_average=True, reduce=True):
    """
    mse_loss(input, target, size_average=True, reduce=True) -> Variable
    Measures the element-wise mean squared error.
    See :class:`~torch.nn.MSELoss` for details.
    """
    return _pointwise_loss(lambda a, b: (a - b) ** 2, torch._C._nn.mse_loss,
input, target, size_average, reduce)

def own_mse_loss(input, target, size_average=True):
    L = (input - target) ** 2
    return torch.mean(L) if size_average else torch.sum(L)