All tests in my final assignment passed (Course 1, Week 4). The assignment was submitted and got 100% score.
However the final accuracy is lower than the expected one. Is it a problem with my code or something changed in the test/data?
I’d like to play with this code a bit more, so if there is a problem, I’d like to find it.
The optimized value of the cost function is also different (better than expected):
I run the code in the entire assignment once more (with no change from yesterday’s run), and now all figures match the expected values.
Yes, the reason is that in the previous instance, you had not run the cell that redefines layers_dims to be the 4 layer network. So you were training with the L layer code, but on the same 2 layer definition of the network. The result is not exactly the same as in the plain vanilla 2 layer case, because it turns out they gave you a different and more sophisticated initialization function for the L layer case.
When you ran the whole notebook, that included the cell that redefines layers_dims.
