torch.nn.functional.binary_cross_entropy_with_logits
function binary_cross_entropy_with_logits(input: Tensor, target: Tensor, options?: BinaryCrossEntropyWithLogitsFunctionalOptions): AnyTensorBinary cross entropy with logits loss. More numerically stable than sigmoid + binary_cross_entropy.
Parameters
inputTensor- Tensor of logits (unbounded)
targetTensor- Tensor of binary targets (0 or 1)
optionsBinaryCrossEntropyWithLogitsFunctionalOptionsoptional