Fix gradient_descent returning after first epoch due to indentation bug

[Copilot]

gradient_descent was silently returning after a single epoch because return [W1, w2] was indented inside the for loop body. This meant weights were never actually trained regardless of epochs, making all Task 2.9 plots useless.

Change

• gradient_descent in DL-Assignment01.ipynb: dedented return [W1, w2] from 8 spaces to 4 spaces so it executes after the loop completes.

# Before — returns after epoch 1
    for _ in range(epochs):
        ...
        return [W1, w2]  # inside loop

# After — returns after all epochs
    for _ in range(epochs):
        ...
    return [W1, w2]  # outside loop

Original prompt

can’t see your notebook from this workspace (the project folder is empty on my side), so I can’t directly judge your exact plots yet.

For Task 2.9, a “good” result is usually:

The predicted curve follows the data trend across the full required range (X0: [-3,3], X1: [-4,4], X2: [-6,5]).
It is smooth and plausible (not wildly oscillating between points).
Train and validation/test loss are both low and not far apart.
Different network types give consistent fits, with complexity matching dataset difficulty.
Quick way to decide “good or not”:

Compare MSE (or your loss) on validation/test for all 3 networks per dataset.
Pick the one with lowest validation/test loss, then confirm the plot is not clearly overfit (wiggly curve) or underfit (too flat/simple).
Common reasons plots look “not promising”:

Wrong R passed to plotting.
Model left in training mode during prediction (model.eval() missing, if PyTorch).
Input/output scaling mismatch during plotting.
Too few epochs or poor learning rate.
Using non-optimized hyperparameters by mistake.
If you paste your Task 2.9 code cell + one plot screenshot, I can pinpoint what’s wrong.

create one large figure

figure = pyplot.figure(figsize=(15,9))
trained_thetas = [
[Theta0_X0, Theta0_X1, Theta0_X2], # model_type 0
[Theta1_X0, Theta1_X1, Theta1_X2], # model_type 1
[Theta2_X0, Theta2_X1, Theta2_X2] # model_type 2
]

ranges = [
[-3, 3], # for X0
[-4, 4], # for X1
[-6, 5] # for X2
]

iterate over the model types

for model_type in range(3):
# iterate over the datasets
for i, X in enumerate([X0, X1, X2]):
# activate correct plot
pyplot.subplot(3, 3, 3*model_type+i+1)
# call the plot function with the correct dataset with their ranges, and model type with their trained Theta parameters
plot(X, model_type, trained_thetas[model_type][i], ranges[i])
pyplot.title(f"N{model_type} on X{i}")

pyplot.tight_layout()
pyplot.show()

Created from VS Code.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.