NN Deep feed forward answers

The answer to the Deep FFNN exercise can be found

FFNN_from_scratch_answer.ipynb

The answer to the NNAlign exercise can be found

NNAlign_from_scratch_answer.ipynb
NOTE: The answers to the NNAlign notebook fix some inconsistencies in the code regarding the weight initialization, and shows the performance on the 8,10,11 mer subset of the A0301 test data, which is where the performance gain should be pronounced most.