LRN, does it work?

You can test the training by yourself, i decided to implement lrn for 2 reasons:

• Edge popup algorithm can't have a normalization during the training except for lrn, or cosine normalization, or other normalizations that do not use learning parameters

• The effectiveness of lrn has been proven (in practice) by this paper https://paperswithcode.com/paper/imagenet-classification-with-deep. Implementing it lead to a better performance of 1.4% more accuracy