C4W1 the first program work

in exercise2:

How do we check the value of the convolution kernel? Is it randomly generated?Because in the tutorial, I did not find any line prompting you to manually enter the values of these convolution kernels.


If you look at the test code, the values are randomly generated before invoking conv_single_step.

In tensorflow, a Conv layer is initialized based on the initializer provided at construction time.
One built, you can invoke weights, biases = conv_layer.get_weights() to get the state of conv layer.