C4_W2_A1 Assignment error

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

When I test the identity_block assignment in W2A1, I get below error Wrong Values with Training - true

I checked the filter passed and other places, doesn’t seem to find the issue.

Please help

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
c:\CNN\Assignments\W2A1\Residual_Networks.ipynb Cell 12 line 3
     26 A4 = identity_block(X, f=2, filters=[3, 3, 3],
     27                    initializer=lambda seed=0:constant(value=1))
     28 print(np.around(A4.numpy()[:,(0,-1),:,:].mean(axis = 3), 5))
---> 30 public_tests.identity_block_test(identity_block)

File c:\CNN\Assignments\W2A1\public_tests.py:56, in identity_block_test(target)
     54 A4np = A4.numpy()
     55 resume = A4np[:,(0,-1),:,:].mean(axis = 3)
---> 56 assert np.allclose(resume, 
     57                      np.array([[[0.,         0.,         0.,         0.,        ],
     58                               [0.,         0.,         0.,         0.,        ]],
     59                              [[0.37390518, 0.37390518, 0.37390518, 0.37390518],
     60                               [0.37390518, 0.37390518, 0.37390518, 0.37390518]],
     61                              [[3.2380054,  4.1391973,  4.1391973,  3.2380054 ],
     62                               [3.2380054,  4.1391973,  4.1391973,  3.2380054 ]]]),
     63                    atol = 1e-5 ), "Wrong values with training=True"
     65 print(colored("All tests passed!", "green"))

AssertionError: Wrong values with training=True
2

That’s an unexpected error, because the identity_block() function does not use a training parameter.

Are you using the latest copy of the notebook? It was updated about a month ago.

3

I have downloaded notebook today. It is latest I believe

4

What do you mean by “downloaded”? You don’t need to download any files. You should work on the notebooks in the Coursera Labs environment.

5

Even on the Coursera lab environment I get same error. Can you please send me version which works for you.

With training=False

[[[   0.         0.         0.         0.     ]
  [   0.         0.         0.         0.     ]]

 [[ 768.9997   768.9997   768.9997   384.99985]
  [ 384.99985  384.99985  384.99985  192.99992]]

 [[2306.999   2306.999   2306.999   1154.9995 ]
  [1154.9995  1154.9995  1154.9995   578.99976]]]
384.99985

With training=True

[[[0.      0.      0.      0.     ]
  [0.      0.      0.      0.     ]]

 [[0.40732 0.40732 0.40732 0.40732]
  [0.40732 0.40732 0.40732 0.40732]]

 [[5.00011 5.00011 5.00011 3.25955]
  [3.25955 3.25955 3.25955 2.40732]]]

---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)
Input In [5], in <cell line: 30>()
     26 A4 = identity_block(X, f=2, filters=[3, 3, 3],
     27                    initializer=lambda seed=0:constant(value=1))
     28 print(np.around(A4.numpy()[:,(0,-1),:,:].mean(axis = 3), 5))
---> 30 public_tests.identity_block_test(identity_block)

File /tf/W2A1/public_tests.py:38, in identity_block_test(target)
     34 assert np.floor(resume[1, 1, 0]) == 2 * np.floor(resume[1, 1, 3]), "Check the padding and strides"
     36 assert resume[1, 1, 0] - np.floor(resume[1, 1, 0]) > 0.7, "Looks like the BatchNormalization units are not working"
---> 38 assert np.allclose(resume, 
     39                    np.array([[[  0.,        0.,        0.,        0.,     ],
     40                               [  0.,        0.,       0.,        0.,     ]],
     41                              [[192.99992, 192.99992, 192.99992,  96.99996],
     42                               [ 96.99996,  96.99996,  96.99996,  48.99998]],
     43                              [[578.99976, 578.99976, 578.99976, 290.99988],
     44                               [290.99988, 290.99988, 290.99988, 146.99994]]]), 
     45                    atol = 1e-5 ), "Wrong values with training=False"
     47 tf.keras.backend.set_learning_phase(True)
     48 np.random.seed(1)

AssertionError: Wrong values with training=False
6

Use the current version from Coursera Labs.

7

Ok got it. Looks like I had an additional component in my code.
Thank you for your reply.