Week 4 - L-Layer neuron network has different result

Course Q&A Deep Learning Specialization Neural Networks and Deep Learning
1

I passed the L-layer-model-test but got different costs and final scores.

dl_nb
dl_nb1075×486 42.2 KB

2

That means that you forgot to run the cell that redefines layers_dims to be a 4 layer network. So you ran the L layer code with the specification for the 2 layer net. Notice that the result is very similar to the output you got from the 2 layer model training earlier. The reason that it is slightly better here is that they used a more sophisticated initialization algorithm for the L layer logic.

So your code is correct (as indicated by the grader), but you get a different answer because you ran it with different inputs.

3

Thank you so much Paul. I am so clumsy haha.