mse_loss(input, target, reduction='mean') → Tensor¶
This operator computes the mean squared error (squared L2 norm) loss between each element in input and target.
input (Tensor) – The input Tensor.
target (Tensor) – The target Tensor.
reduction (string, optional) – Specifies the reduction to apply to the output:
'none': no reduction will be applied,
'mean': the sum of the output will be divided by the number of elements in the output,
'sum': the output will be summed. Default:
>>> import oneflow as flow >>> import oneflow.nn.functional as F >>> input = flow.randn(3, 4, requires_grad=True) >>> target = flow.rand(3, 4, requires_grad=False) >>> loss = F.mse_loss(input, target) >>> loss.backward()