Pytorch Sigmoid Function what is e

Question

I have a question on setting up the sigmoid function in pytroch.

So I define it as this

# Sigmoid function
def sigmoid(x):
    return 1/(1 + torch.exp(-x))

But then looking at the sigmoid function from here http://mathworld.wolfram.com/SigmoidFunction.html

The sigmoid should be defined as

y = 1/(1 + e^-x)

I see the 1/(1+ part but I don't get the e^-x part. Can someone explain why

torch.exp(-x) == e^-x

What is e here? Is that the tensor. But i thought that x was the tensor

e is a mathematical constant see [here](https://en.wikipedia.org/wiki/E_(mathematical_constant)). e^x is basically mathematical syntactic sugar for the exp(x) function but they mean the same thing. — jodag, Jan 22 '20 at 05:14
so your are saying the exp() is e here? Is that what .exp(-x) does — MNM, Jan 22 '20 at 05:20
x is just a variable so exp() = e^ its just a different representation of the same thing. — jodag, Jan 22 '20 at 05:20
Seeing you were first could you make your explanation the answer and I will accept it. Thank you @jodag — MNM, Jan 22 '20 at 05:23

score 4 · Accepted Answer · edited Jan 22 '20 at 05:28

4

Here e is the exponential constant e and torch.exp(-x) == e^-x

edited Jan 22 '20 at 05:28

jodag

answered Jan 22 '20 at 05:21

Dishin H Goyani

1

so the parameter that you plug into will be x correct. So e does not have to be defined. – MNM Jan 22 '20 at 05:22

1 Answers1