C2W3: Tensorflow intro_Ex. 3 - one_hot_matrix : Error message while output is correct

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

Hello!
I am getting a huge error message while my output matches the expected output.
Here is my output:
tf.Tensor(
[[0.]
[1.]
[0.]
[0.]], shape=(4, 1), dtype=float32)

Here is the error message:

`InvalidArgumentError Traceback (most recent call last)
in
12 print("\033[92mAll test passed")
13
—> 14 one_hot_matrix_test(one_hot_matrix)

in one_hot_matrix_test(target)
7 assert result.shape[1] == 1, f"Reshape to have only 1 column"
8 assert np.allclose(result, [[0.], [1.], [0.], [0.]] ), “Wrong output. Use tf.one_hot”
----> 9 result = target(3, depth)
10 assert np.allclose(result, [[0.], [0.], [0.], [1.]] ), “Wrong output. Use tf.one_hot”
11

in one_hot_matrix(label, depth)
15 # YOUR CODE STARTS HERE
(deleted)
19 # YOUR CODE ENDS HERE`

…

/opt/conda/lib/python3.7/site-packages/tensorflow/python/eager/execute.py in quick_execute(op_name, num_outputs, inputs, attrs, ctx, name)
58 ctx.ensure_initialized()
59 tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,
—> 60 inputs, attrs, num_outputs)
61 except core._NotOkStatusException as e:
62 if name is not None:

InvalidArgumentError: Input to reshape is a tensor with 4 values, but the requested shape has 12 [Op:Reshape]

Any ideas? I have difficulties with this package.

Thanks in advance

2

Hi Viktoriia,

Check out the dimensions of your call to tf.reshape, they look wrong to me.
Think about what dimension the resulting one_hot matrix should be? The exercise says:
" return the one hot encoding in a column wise[sic] matrix"

The case that is failing for you is when the one_hot function is being called with label=3, depth=4 for which the result should be (according to test_case):
[[0.],
[0.],
[0.],
[1.]]

Hope that helps!

3 Likes
3

Hi @Viktoriia,

@neurogeek is right, check the second argument to tf.reshape.

If still in doubt, print the shape of the one_hot tensor right after calling tf.one_hot and you’ll immediately see why you need to call tf.reshape :slight_smile:

If you don’t mind, delete the code between # YOUR CODE STARTS HERE and # YOUR CODE ENDS HERE once you get it to work.

Good luck!

3 Likes
4

Dear @neurogeek and @nramon
Many thanks for your replies. It took me time to figure out…
So in this particular exercise we need to compute a one_hot_matrix for a single label therefore the shape must be number of classes by 1. :woman_facepalming:
That was really helpful ! I would probably not get it today without your hints. :pray:
Thanks!

2 Likes
5

Yep you are right!
Glad that worked out!

1 Like
6

Hello, I am stuck on the same one_hot_matrix exe. but my problem is I can not manage shape.
Here Is my code.

##YOUR CODE STARTS HERE

one_hot = tf.one_hot(label, depth)

##YOUR CODE ENDS HERE

and here is my error :---------------------------

tf.Tensor([0. 1. 0. 0.], shape=(4,), dtype=float32)

IndexError Traceback (most recent call last)
in
12 print("\033[92mAll test passed")
13
—> 14 one_hot_matrix_test(one_hot_matrix)

in one_hot_matrix_test(target)
5 print(result)
6 assert result.shape[0] == depth, “Use the parameter depth”
----> 7 assert result.shape[1] == 1, f"Reshape to have only 1 column"
8 assert np.allclose(result, [[0.], [1.], [0.], [0.]] ), “Wrong output. Use tf.one_hot”
9 result = target(3, depth)

/opt/conda/lib/python3.7/site-packages/tensorflow/python/framework/tensor_shape.py in getitem(self, key)
885 else:
886 if self._v2_behavior:
→ 887 return self._dims[key].value
888 else:
889 return self._dims[key]

IndexError: list index out of range