Open Myrkiriad-coder opened 1 year ago
Yes, that is expected Every time you change the order of your equation, you must also change the parameters too. Reduce the learning rate if your error plot is noisy. You can also tweak the number of the nodes, iterations, ...etc. Just like what I did here: https://youtu.be/l-CjXFmcVzY?t=872
Hello,
Big fan of your videos, I decided to look into your code this time !
In neural_network_tutorials.ipynb, 4/ Deep Neural Networks (Numpy).
If I change
ys = xs**2
, for anything else (ys = xs**3, ys = xs**4, ys = np.sin(xs), ys = np.abs(xs)
etc....), there is no fit. Any reason for that ?