Week 1 Is Gradient Checking Useful in Practice?

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

Since Tensorflow does automatic difference, as model developers we don’t implement the backpropagation manually for the customized models.

In that sense, is gradient checking techinque still useful in practice?

2

Gradient checking is a useful if you ever have to write your own backpropagation code, and need to check that its results are reasonable.

Not all platforms you may want to use can support tools like TensorFlow.