I completed this assignment but even though my numbers match the expected output, there is no grader annotations and when submitted I get a grade of 0/100. Is there something wrong with the grader functions in this notebook?
Those numbers do look correct. But there may be other things wrong which have nothing to do with the actual code. I just submitted my Face Recognition assignment again and got full marks, so there is nothing wrong with the grader. Have you checked what it says under “Show Grader Output”? That may shed some light on what is causing your 0 score. Here’s what my output looks like:
Grader outptu: Cell #5. Can’t compile the student’s code. Error: NameError(“name ‘y_true’ is not defined”)
y_true is defined as (None, None, None) within the pre existing evaluation code. As the comment indicates this variable is not used anywhere so I just deleted it from the function as well as evaluation code but there are still no grader notes anywhere in my code.
I don’t know if there are course administrators that can help me get the grading function to show up in my code. It is holding up my course completion.
Wait, ok, I just went and looked at the code again to remind myself what it looks like. Are you telling us that you changed the function definition of triplet_loss to remove the y_true argument? If so, that will not end well and could easily cause problems with the grader. Changing anything outside the “YOUR CODE HERE” blocks is not a good idea. It’s not illegal, but it needs to be done with extreme care.
In this particular case, it means you completely misinterpreted what that comment in the “docstring” of the function says. It says you don’t actually use that value here, but you need it because this function is used as a Keras loss function that we are writing. And Keras loss functions have a standard function signature. You changed the function signature, so that’s why everything failed with that error.
If you want to get a clean copy of the notebook to restore the original code, there is a topic about that on the FAQ Thread.