vincentFrancais

I second what @ezeli said. Using demux I often exceed the buffer_size on my datasets. Furthermore it's not always straightforward to define a classifier function to split datasets according to...