Changing lambda does not change results too much

Course Q&A Deep Learning Specialization Improving Deep Neural Networks: Hyperparameter tun
1

Hello,

I am playing around with the code in Chapter 5 in the Week 1 Assignment 2 – L2 Regularization. The code works well, and the default value of Lambda was set to 0.7. However, when I change it from 0.95 to as low as even 0.3, the test set accuracy does not change by much at all. It stays above 92% for sure. Any thoughts?

Maybe this is not a good dataset to observe any significant effects of regularization??

Thanks,

2

Hi there,
What’s your result if setting lambda to 0.0?