mindspore.ops.SoftmaxCrossEntropyWithLogits

class mindspore.ops.SoftmaxCrossEntropyWithLogits(*args, **kwargs)[source]

Gets the softmax cross-entropy value between logits and labels with one-hot encoding.

The updating formulas of SoftmaxCrossEntropyWithLogits algorithm are as follows,

\[\begin{split}\begin{array}{ll} \\ p_{ij} = softmax(X_{ij}) = \frac{\exp(x_i)}{\sum_{j = 0}^{N-1}\exp(x_j)} \\ loss_{ij} = -\sum_j{Y_{ij} * ln(p_{ij})} \end{array}\end{split}\]

where \(X\) represents logits. \(Y\) represents label. \(loss\) represents output.

Inputs:
  • logits (Tensor) - Input logits, with shape \((N, C)\). Data type must be float16 or float32.

  • labels (Tensor) - Ground truth labels, with shape \((N, C)\), has the same data type with logits.

Outputs:

Tuple of 2 tensors, the loss shape is (N,), and the dlogits with the same shape as logits.

Raises
  • TypeError – If dtype of logits or labels is neither float16 nor float32.

  • TypeError – If logits or labels is not a Tensor.

  • ValueError – If shape of logits is not the same as labels.

Supported Platforms:

Ascend GPU CPU

Examples

>>> logits = Tensor([[2, 4, 1, 4, 5], [2, 1, 2, 4, 3]], mindspore.float32)
>>> labels = Tensor([[0, 0, 0, 0, 1], [0, 0, 0, 1, 0]], mindspore.float32)
>>> softmax_cross = ops.SoftmaxCrossEntropyWithLogits()
>>> loss, dlogits = softmax_cross(logits, labels)
>>> print(loss)
[0.5899297  0.52374405]
>>> print(dlogits)
[[ 0.02760027  0.20393994  0.01015357  0.20393994 -0.44563377]
 [ 0.08015892  0.02948882  0.08015892 -0.4077012   0.21789455]]