Fix: Remove momentum from SGD to show standard optimizer behavior

[ShindeShivam]

Chapter : 11
Cell : Faster Optimizers [47]

Changed:
optimizer = tf.keras.optimizers.SGD(learning_rate=0.001, momentum=0.9)

To:
optimizer = tf.keras.optimizers.SGD(learning_rate=0.001)

Why change was needed:

While working through Chapter 11, I noticed that the SGD optimizer was being initialized with a momentum value.
Since this section appears to be comparing various optimizers — including plain SGD — I felt that including momentum here might unintentionally misrepresent how standard SGD behaves. Momentum definitely helps improve performance, but it has been added separately after this cell to SGD.
I removed the momentum parameter so the optimizer now reflects vanilla SGD.

**I’ve attached the optimizer loss plots from before and after the change (see images below). **
In the original version, the “SGD” curve seemed to perform better than expected — and after inspecting the code, I realized it was actually using momentum. After the fix, SGD’s curve is now visible and shows the slower convergence we typically expect from it.

Before->

After->

[ShindeShivam]

I’ve also opened a couple of other PRs recently (#196 and #202)

[ageron]

Great catch, thanks! Merged. 👍