Is this a grading error or am I missing something?

Course 5, week 3, “True or False” quiz question - “Continuing the example from Q4, suppose you work on your algorithm for a few more weeks, and now find that for the vast majority of examples on which your algorithm makes a mistake, P(y* | x) > P({y(hat) | x). This suggests you should not focus your attention on improving the search algorithm.”

I think this is “True” and marked it as such in the quiz. However, it graded it incorrect with the error:
“P(y^* \mid x) > P(\hat{y} \mid x)$$ indicates the error should be attributed to the search algorithm rather than to the RNN.”

To me this looks like a bug in the grader… But also very possible that I am missing something. Thought to check here.


I’m going to guess you’re referring to the “error analysis on beam search” topic.

Please review that video, around time mark 5:16.