Q$13$: What is the primary purpose of $L2$ regularization in neural networks?

a$)$ Encourages sparsity in weights

b$)$ Reduces large weight values

c$)$ Prevents exploding gradients

d$)$ Enhances model interpretability