For some reason my optimization runs really slow. For example when I ran the learning rate test (trying 3 different learning rates), which was expected to take a minute, it took about 3.
I am vectorizing everything… no loops.
Any idea?
I wouldn’t worry about that. It’s probably just that you got unlucky and the AWS VM you were running on at that point was busy. If your code is not vectorized, the performance degradation is usually much worse than 3x.
I’ve also been noticing that the notebooks that run lots of iterations of training in general have been slow lately.
Thank you Paul. That makes sense.