Week 1 Assignment 1 Define output layers of the model

Course Q&A TensorFlow: Advanced Techniques Specialization Custom Models, Layers and Loss Functions with TF
1

The instruction asks us to use default activation. But it keeps showing errors. The problem was solved when I use the sigmoid function as activation for wine quality. Why is that?

layer_3_activation_check: wine_quality layer has an incorrect activation. Please check implementation.
Expected: <function sigmoid at 0x7ff5ee33b9e0>
Result: <function linear at 0x7ff5ee33bb90>
Please open utils.py if you want to see the unit test here.

1 Tests passed
1 Tests failed

Exception Traceback (most recent call last)
in
----> 1 utils.test_final_model(final_model)

~/work/release/W1_Assignment/utils.py in test_final_model(final_model)
279 ]
280
→ 281 test_loop(test_cases)
282
283 def test_model_compile(model):

~/work/release/W1_Assignment/utils.py in test_loop(test_cases)
25 print(’\033[92m’, success," Tests passed")
26 print(’\033[91m’, fails, " Tests failed")
—> 27 raise Exception(“Please check the error messages above.”)
28
29 def test_white_df(white_df):

Exception: Please check the error messages above.

2

It says use default for wine quality i.e. no activation at all, and sigmoid for wine type output. Note there are 2 output branches in this neural network.