DL4 -Week1,Assignment 1,Excercise3

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

ipython-input-47-e9a8fc127504> in conv_forward(A_prev, W, b, hparameters)
84 weights= W[:,:,:,c]
85 bias= b[:,:,:,c].reshape(1,1,1)
—> 86 Z[i, h, w, c]=conv_single_step(a_slice_prev, weights, bias)
87 print(Z[i, h, w, c])
88 # YOUR CODE ENDS HERE

IndexError: index 4 is out of bounds for axis 3 with size 4
My question here is why is it saying out of bounds .
Is my code is correct?I am not able to understand the mistake.
Need Help!!!

2

Either a_slice_prev or weights or bias are the wrong shape.

I don’t think you need to reshape the bias.

Consider whether your code for “bias = …” is giving the correct value.

3

In fact this I got as shape of a_slice_prev and value of Z[i,h,w,c] .It had gone well with index 3 but it is not oroceeding further.I think I have a mistake in logic.
(3, 3, 4)
-2.651123629553914
(3, 3, 4)
-0.37849176551833935
(3, 3, 4)
-1.970549285291896
(3, 3, 4)
-1.9623529930298282
(3, 3, 4)
Can you please give me a hint?

4

You’ve got several threads going about this same problem. Did you look at the data I gave you on the other one:

I added a bunch of print statements to my conv_forward logic and here’s what I see when I run the test case for that:

stride 2 pad 1
New dimensions = 3 by 4
Shape Z = (2, 3, 4, 8)
Shape A_prev_pad = (2, 7, 9, 4)
Z[0,0,0,0] = -2.651123629553914
Z[1,2,3,7] = 0.4427056509973153
Z's mean =
 0.5511276474566768
Z[0,2,1] =
 [-2.17796037  8.07171329 -0.5772704   3.36286738  4.48113645 -2.89198428
 10.99288867  3.03171932]
cache_conv[0][1][2][3] =
 [-1.1191154   1.9560789  -0.3264995  -1.34267579]
First Test: All tests passed!
stride 1 pad 3
New dimensions = 9 by 11
Shape Z = (2, 9, 11, 8)
Shape A_prev_pad = (2, 11, 13, 4)
Z[0,0,0,0] = 1.4306973717089302
Z[1,8,10,7] = -0.6695027738712113
stride 2 pad 0
New dimensions = 2 by 3
Shape Z = (2, 2, 3, 8)
Shape A_prev_pad = (2, 5, 7, 4)
Z[0,0,0,0] = 8.430161780192094
Z[1,1,2,7] = -0.2674960203423288
stride 1 pad 6
New dimensions = 13 by 15
Shape Z = (2, 13, 15, 8)
Shape A_prev_pad = (2, 17, 19, 4)
Z[0,0,0,0] = 0.5619706599772282
Z[1,12,14,7] = -1.622674822605305
Second Test: All tests passed!

Notice that the shape of Z is wider than it is tall. Maybe you switched the meaning of h and w. Remember that h is for “height” not “horizontal”.

Also it would be good to print the shape of your Z value, since that is what the error is complaining about.

5

I don’t think so my final Z value is the problem because c in n_C is not completing its 5th iteration. That’s the problem here .
I am not getting why its not completing its iteration.
That’s why if there is any logical mistakes
h and w is fine here

6

I’m not talking about the value of Z, but the shape. Please add this statement early in your conv_forward:

print(f"Z.shape = {Z.shape}")

Note that there are 8 output channels, so the highest c value in your loop will be 7, right? So if your Z is the correct shape, then why does it fail when c is 4 or 5 (whichever you mean by “fifth iteration”)?

7

sorry ,I got Z.shape = (2, 5, 7, 4)
(3, 3, 4)
-2.651123629553914
Z.shape = (2, 5, 7, 4)
(3, 3, 4)
-0.37849176551833935
Z.shape = (2, 5, 7, 4)
(3, 3, 4)
-1.970549285291896
Z.shape = (2, 5, 7, 4)
(3, 3, 4)
-1.9623529930298282
Z.shape = (2, 5, 7, 4)
(3, 3, 4)
Is my Z shape is correct?
Now I am getting little confused.

8

I showed the correct shape in my earlier post:

stride 2 pad 1
New dimensions = 3 by 4
Shape Z = (2, 3, 4, 8)
Shape A_prev_pad = (2, 7, 9, 4)

So it looks like your h and w dimensions are wrong as well as the c dimension. For c you’ve used the input channels, but it should be the output channels (8 in this case).

For h and w, the formula is:

n_{out} = \displaystyle \lfloor \frac {n_{in} + 2p - f}{s} \rfloor + 1

9 
for h in range(n_H):
 {moderator edit: code removed}

this is basically I have done

n_H=3
n_W=4
-2.651123629553914
Z.shape = (2, 5, 7, 4)
-0.37849176551833935
Z.shape = (2, 5, 7, 4)
-1.970549285291896
Z.shape = (2, 5, 7, 4)
-1.9623529930298282
Z.shape = (2, 5, 7, 4)

I don’t know why Z.shape is not showing (2,3,4,8)

10

Where is the shape of Z determined? You have to predefine it before you start the nested “for” loops, right?

What is your n_C value?

Note that I edited your post to use the “</>” formatting option for the code and output, so that you can actually read it more easily.

11

Actually it looks like you just used all the input dimensions for Z. But Z is the output, right? So the only dimension it shares with the input is dimension 0: the “samples” dimension.

12

Thank you so much I could complete this Week1 assignment1 with colours of wonders.