"""
This script demonstrates the implementation of the Sigmoid Linear Unit (SiLU)
or swish function.
* https://en.wikipedia.org/wiki/Rectifier_(neural_networks)
* https://en.wikipedia.org/wiki/Swish_function
The function takes a vector x of K real numbers as input and returns x * sigmoid(x).
Swish is a smooth, non-monotonic function defined as f(x) = x * sigmoid(x).
Extensive experiments shows that Swish consistently matches or outperforms ReLU
on deep networks applied to a variety of challenging domains such as
image classification and machine translation.
This script is inspired by a corresponding research paper.
* https://arxiv.org/abs/1710.05941
"""
import numpy as np
def sigmoid(vector: np.array) -> np.array:
"""
Mathematical function sigmoid takes a vector x of K real numbers as input and
returns 1/ (1 + e^-x).
https://en.wikipedia.org/wiki/Sigmoid_function
>>> sigmoid(np.array([-1.0, 1.0, 2.0]))
array([0.26894142, 0.73105858, 0.88079708])
"""
return 1 / (1 + np.exp(-vector))
def sigmoid_linear_unit(vector: np.array) -> np.array:
"""
Implements the Sigmoid Linear Unit (SiLU) or swish function
Parameters:
vector (np.array): A numpy array consisting of real
values.
Returns:
swish_vec (np.array): The input numpy array, after applying
swish.
Examples:
>>> sigmoid_linear_unit(np.array([-1.0, 1.0, 2.0]))
array([-0.26894142, 0.73105858, 1.76159416])
>>> sigmoid_linear_unit(np.array([-2]))
array([-0.23840584])
"""
return vector * sigmoid(vector)
if __name__ == "__main__":
import doctest
doctest.testmod()