def softmax(x, name=None):
"""
Squashes the input values `x` such that they add up to 1:
:math:`softmax(x) = {\exp(x_i) - \max_{x_i \in x}(\exp(x_i)) \over {\sum_{x_i \in x} \exp(x_i)- \max_{x_i \in x}(\exp(x_i)) }}`
The term :math:`\max_{x_i \in x}(\exp(x_i))` is subtracted for numerical
stability.
Example:
>>> C.eval(C.softmax([[1, 1, 2, 3]]))
[array([[[ 0.082595, 0.082595, 0.224515, 0.610296]]])]
>>> C.eval(C.softmax([1, 1]))
[array([[ 0.5, 0.5]])]
Args:
x: any :class:`cntk.graph.ComputationNode` that outputs a tensor
Returns:
:class:`cntk.graph.ComputationNode`
"""
from cntk.ops.cntk2 import Softmax
op = Softmax(x)
wrap_numpy_arrays(op)
op.rank = op._.rank
return op
def softmax(x, name=None):
"""
Squashes the input values `x` such that they add up to 1:
:math:`softmax(x) = {\exp(x_i) - \max_{x_i \in x}(\exp(x_i)) \over {\sum_{x_i \in x} \exp(x_i)- \max_{x_i \in x}(\exp(x_i)) }}`
The term :math:`\max_{x_i \in x}(\exp(x_i))` is subtracted for numerical
stability.
Example:
>>> C.eval(C.softmax([[1, 1, 2, 3]]))
[array([[[ 0.082595, 0.082595, 0.224515, 0.610296]]])]
>>> C.eval(C.softmax([1, 1]))
[array([[ 0.5, 0.5]])]
Args:
x: numpy array or any :class:`cntk.graph.ComputationNode` that outputs a tensor
name (str): the name of the node in the network
Returns:
:class:`cntk.graph.ComputationNode`
"""
from cntk.ops.cntk2 import Softmax
op = Softmax(x)
wrap_numpy_arrays(op)
op.rank = op._.rank
return op
def softmax(x, name=None):
"""
Softmax operation. Squashes the input values `x` such that they add up to 1:
:math:`softmax(x) = {\exp(x_i) - \max_{x_i \in x}(\exp(x_i)) \over {\sum_{x_i \in x} \exp(x_i)- \max_{x_i \in x}(\exp(x_i)) }}`
The term :math:`\max_{x_i \in x}(\exp(x_i))` is subtracted for numerical
stability.
Example:
>>> C.eval(C.softmax([[1, 1, 2, 3]]))
[array([[[ 0.082595, 0.082595, 0.224515, 0.610296]]])]
>>> C.eval(C.softmax([1, 1]))
[array([[ 0.5, 0.5]])]
Args:
x: any :class:`cntk.graph.ComputationNode` that outputs a tensor
Returns:
:class:`cntk.graph.ComputationNode`
"""
from cntk.ops.cntk2 import Softmax
return Softmax(x)