These are just notebooks for visualizing weights in Keras layers. One notebook pulls the weights out and directly visualizes them. This is a pictorial representation of the convolution filter kernel. The other notebook displays the evaluation of one or more internal layers with a grayscale image as input. This is the projection of the grayscale image through the collection of convolutions, which also gives some insight into the weights applied to the incoming data. In particular, it shows how the kernels modify data from the mean value given it.
ColinShaw/jupyter-keras-weight-visualization
Visualization of the VGG16 weights in a Jupyer notebook
Jupyter Notebook