Week 1 - Gradient Checking Task - Wrong grade issue


I successfully finished the gradient checking task and all notebook paragraphs were successfully executed as the following picture portraits:

However, the grader is providing an output of 60/100 which doesn’t make sense and It is not providing any sort of additional feedback about what could be wrong and for obvious reasons with that score I didn’t approved the task.

Any hints and solutions for this non-sense issue?.




I had the same problem.
And if you have problem with the assertion “You are not using np.linalg.norm for numerator or denominator” in gradient_check_n, please check the type of numerator , denominator, and difference.
It turned out that, the test requires difference to be a float number. However, when I used np.linalg.norm with axis=0, it made numerator and denominator become np.ndarray. Thus, difference is a np.ndarray. My suggestion is to just keep axis=None as default.

Hope this can help.

1 Like