Excercise 6 - UNQ_C5

get this error while running the train_step(generated_image) function… pls help

Two questions:
-Does your compute_layer_style_cost() function work correctly?
-In your compute_style_cost() function, are you calling compute_layer_style_cost()?

Yes, compute_layer_style_cost() function does works correclty and yeah the compute_layer_style_cost() function is called in the compute_style_cost() function(it was already there).

Hello @CSB_38_Waseem,

Let’s analyze the Traceback:

From it, we realize that the problem happens in tf.subtract, and that should bring our focus onto a_S and a_G. Remember a_S is returned from vgg_model_outputs(...) which was defined in Section 5.4. If you read that section, it said vgg_model_outputs

returns a list of the outputs for the middle layers.

Therefore, you are letting tf.subtract to work on a list of tensors of different shapes (since the middle layers have different shapes), and it is also why you see those shapes that I have underlined.

Now, your objective is to calculate the style cost, and you will further need to calculate the content cost and the total cost. However, remember we have defined some functions in previous exercises to handle them? I suggest you to revisit them and try to figure out how they handle a_S, a_G, and a_C which are lists, and then see if you can understand the problem behind this error. Then, use those defined functions to do the job.


1 Like

TYSM, helped me to solve the issue :blush:

You are welcome, @CSB_38_Waseem!