[Bug] ”weight=class_weight“ should be modified for ”pos_weight=class_weight“

loss = F.binary_cross_entropy_with_logits( pred, label, weight=class_weight, reduction=‘none’)

it should be modified for “loss = F.binary_cross_entropy_with_logits( pred, label, pos_weight=class_weight, reduction=‘none’)

def binary_cross_entropy(pred,
                         label,
                         weight=None,
                         reduction='mean',
                         avg_factor=None,
                         class_weight=None):
    r"""Calculate the binary CrossEntropy loss with logits.

    Args:
        pred (torch.Tensor): The prediction with shape (N, \*).
        label (torch.Tensor): The gt label with shape (N, \*).
        weight (torch.Tensor, optional): Element-wise weight of loss with shape
            (N, ). Defaults to None.
        reduction (str): The method used to reduce the loss.
            Options are "none", "mean" and "sum". If reduction is 'none' , loss
            is same shape as pred and label. Defaults to 'mean'.
        avg_factor (int, optional): Average factor that is used to average
            the loss. Defaults to None.
        class_weight (torch.Tensor, optional): The weight for each class with
            shape (C), C is the number of classes. Default None.

    Returns:
        torch.Tensor: The calculated loss
    """
    assert pred.dim() == label.dim()
    # Ensure that the size of class_weight is consistent with pred and label to
    # avoid automatic boracast,
    if class_weight is not None:
        N = pred.size()[0]
        class_weight = class_weight.repeat(N, 1)
    loss = F.binary_cross_entropy_with_logits(
        pred, label, weight=class_weight, reduction='none')

    # apply weights and do the reduction
    if weight is not None:
        assert weight.dim() == 1
        weight = weight.float()
        if pred.dim() > 1:
            weight = weight.reshape(-1, 1)
    loss = weight_reduce_loss(
        loss, weight=weight, reduction=reduction, avg_factor=avg_factor)
    return loss

[mzr1996]

Hello, in binary cross-entropy, pos_weight means “a weight of positive examples”, instead of “a weight given to each class”. In regular cross-entropy, PyTorch provides weight, and describes it as “a manual rescaling weight given to each class.”. Therefore, to keep the same behavior, we define class_weight as “class-wise weight”, differ from pos_weight.

	class-wise
	hat	glasses
positive	0	1
negative	1	0

As for your need, we will add the pos_weight parameter in our BCE loss recently.

[Ezra-Yu]

we have added pos_weight in BCE refer to #515