Wrong values when training=True

This is the correct signature of the function:

def convolutional_block(X, f, filters, s = 2, initializer=glorot_uniform):

The training phase is set inside public_tests.py via:

tf.keras.backend.set_learning_phase(True)

Please follow these steps to refresh your workspace

Posting solution code in a public topic is discouraged and can get your account suspended. It’s okay to share stacktrace on a public post and send code to a mentor via direct message. Do clean up the post.

Do not modify the function definition. It is outside of any “START CODE HERE” section.

It looks like you have copied a solution from an old version of this notebook. The APIs were redefined about a year ago (I think it was), so it does not work to copy old solutions.

Please get a clean copy of the notebook using this procedure, which is linked from the DLS FAQ Thread. Then implement the solution following the instructions in the notebook, rather than trying to copy a solution from somewhere else.

2 Likes