Learning Rate (η):
0.04
Convergence Tolerance (ε):
0.01
Start Gradient Descent