ray.rllib.utils.torch_utils.softmax_cross_entropy_with_logits#

ray.rllib.utils.torch_utils.softmax_cross_entropy_with_logits(logits: numpy.array | jnp.ndarray | tf.Tensor | torch.Tensor, labels: numpy.array | jnp.ndarray | tf.Tensor | torch.Tensor) → numpy.array | jnp.ndarray | tf.Tensor | torch.Tensor[source]#

Same behavior as tf.nn.softmax_cross_entropy_with_logits.

Parameters:

  • x – The input predictions.

  • labels – The labels corresponding to x.

Returns:

The resulting softmax cross-entropy given predictions and labels.