Week 2, exercise 8: TypeError: 'tuple' object cannot be interpreted as an integer

Course Q&A Deep Learning Specialization Neural Networks and Deep Learning
1

Hello – I’m working on exercise 8, and am getting an error - TypeError: ‘tuple’ object cannot be interpreted as an integer. Would appreciate insight on why this is breaking. Thank you!

Screen Shot 2022-08-13 at 6.34.09 PM
Screen Shot 2022-08-13 at 6.34.09 PM2234×728 132 KB

2

Apparently, you have a wrong value for num_px, which needs to be 64. This is what you set at Exercise 1. Please double check.

If we look at the shape for your num_px, it is exactly same as the shape of train_set_x_orig.
So, please check your result of Exercise 1, and also any incorrect updates for num_px after that in your notebook.

3

it looks like you’re using a tuple as the parameter for shape.

Instead, use it like

plt.imshow(test_set_x[:, index].reshape(numpx, num_num_px, 3))

I’m not taking this course, so it’s my gut feel looking into the stacktrace.

4

Good try, but, I suppose your code gets the same error.
There are some errors in your both approach and implementation.

Approach:

it looks like you’re using a tuple as the parameter for shape.

This is a good starting point. But, you need to dig one step why it became a tuple.
There are good clues in Tricia’s trace (print).

  • The shape of num_px is (209,64,64,3). Even only looking at a trace, you should be aware, from the last 3 dimensions, that this is most likely a set of color images.
  • And, this code picks up 1 sample (index=1) from “test_set_x”, which should be (12288,1) from a trace.
  • Then, look at the 3rd parameter for reshape(). It’s 3. So, num_px * num_px must be 4096 and the value of num_px should be 64. Then, we understand that this code is trying to reshape one slice of a test set to (64,64,3), which exactly matches to our first guess from given num_px, i.e, a color image.

This is the logical flow for debugging. num_px needs to be 64. Then, we can start to think why num_px became to 64.

Implementation:

For reshape, you set numpx and num_num_px. How you get numpx and num_num_px ?
You may want to show both are scalar value obtained from somewhere, not num_px. Even in that case, from an original code, both needs to be the same value. So, selecting one variable is the right way to guide.
If this is just a typo, the shape of num_px is still (209,64,64,3). Even you remove one pair of parenthesis, reshape((209,64,64,3), (209,64,64,3), 3) does not work.

In net, you have a good idea to start. What you need is to dig one step further to see “the truth”.

2 Likes
5

Thank you so much! I really appreciate your help.