Missed TensorFlow Dev Summit? Check out the video playlist. Watch recordings


View source on GitHub

Computes softmax activations. (deprecated arguments)

    logits, axis=None, name=None, dim=None

This function performs the equivalent of

softmax = tf.exp(logits) / tf.reduce_sum(tf.exp(logits), axis)

See: https://en.wikipedia.org/wiki/Softmax_function

Example usage:

tf.nn.softmax([-1, 0., 1.])


  • logits: A non-empty Tensor, or an object whose type has a registered Tensor conversion function. Must be one of the following types: half,float32, float64. See also convert_to_tensor
  • axis: The dimension softmax would be performed on. The default is -1 which indicates the last dimension.
  • name: A name for the operation (optional).
  • dim: Deprecated alias for axis.


A Tensor. Has the same type and shape as logits.


  • InvalidArgumentError: if logits is empty or axis is beyond the last dimension of logits.
  • TypeError: If no conversion function is registered for logits to Tensor.
  • RuntimeError: If a registered conversion function returns an invalid value.