I am just asking how can I use ReLU activation function in neural net library of SciSnap?
I want to get outputs greater than 1!
Is it possible at all?
If someone knows how to do it, please help me.
This is not what an activation function should do.
It should present a positive slope everywhere and be between -1 and 1 and retain the slope long enough.
Sigmoid is probably better.
Thank you for you response!
But I meant that even if I did put wanted input to 4 it converted to 1 when predicting.
It doesn't work
Show your project