Hey there, my formula for scheduling decay rate is:
leaning_rate = 1 / (1 + decay_rate * np.floor(epoch_num / time_interval)) * learning_rate0
The expected result is after 100 epochs the learning rate will drop from 0.5 to .384.
When I run the calculations on a calculator or in excel it works fine, but in my python notebook, the learning rate stays at 0.5
Any ideas?
You are updating a different variable name on the left hand side of that assignment statement.
If it makes you feel any better, I missed it the first time I looked at the code, too. 
I was assuming it was probably an “order of operations” problem on the RHS or the fact that you were using integer values for the constants. It’s actually a much easier problem than that.
Oh goodness yeah that’ll do it… thanks mate. I think my eyes were a little worn down by the time I got to this section.