oneflow.nn.MarginRankingLoss

class oneflow.nn.MarginRankingLoss(margin: float = 0.0, reduction: str = 'mean')

Creates a criterion that measures the loss given inputs \(x1\), \(x2\), two 1D mini-batch Tensors, and a label 1D mini-batch tensor \(y\) (containing 1 or -1).

If \(y = 1\) then it assumed the first input should be ranked higher (have a larger value) than the second input, and vice-versa for \(y = -1\).

The loss function for each sample in the mini-batch is:

\[\text{loss}(x1, x2, y) = \max(0, -y * (x1 - x2) + \text{margin})\]
Parameters

  • margin (float, optional) – Has a default value of \(0\).

  • reduction (string, optional) – Specifies the reduction to apply to the output: 'none' | 'mean' | 'sum'. '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: 'mean'

Shape:

  • x1 : \((N, D)\) where N is the batch size and D is the size of a sample.

  • x2 : \((N, D)\) where N is the batch size and D is the size of a sample.

  • Target: \((N)\)

  • Output: scalar. If reduction is 'none', then \((N)\).

For example:

>>> import oneflow as flow
>>> import numpy as np
>>> x1 = flow.tensor(np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]), dtype=flow.float32)
>>> x2 = flow.tensor(np.array([[2, 2, 2], [2, 2, 2], [2, 2, 2]]), dtype=flow.float32)
>>> target = flow.tensor(np.array([[1, -1, 1],[-1, 1, -1], [1, 1, 1]]), dtype=flow.float32)
>>> m = flow.nn.MarginRankingLoss(margin =1.0, reduction="none")
>>> out = m(x1, x2, target)
>>> out
tensor([[2., 1., 0.],
        [3., 0., 5.],
        [0., 0., 0.]], dtype=oneflow.float32)

>>> m = flow.nn.MarginRankingLoss(margin = 0.3, reduction="sum")
>>> out = m(x1, x2, target)
>>> out
tensor(8.2000, dtype=oneflow.float32)

>>> m = flow.nn.MarginRankingLoss(margin = 10, reduction="mean")
>>> out = m(x1, x2, target)
>>> out
tensor(8.3333, dtype=oneflow.float32)