Starting point to really understand Mxnet and GluonTS throughtfully

[rurjais]

As I read everyone's questions on DeepAR, and their doubts about models and everything, I am always guessing if someone has access to some information, videos, or tutorials that I don't...

I personally feel that the GluonTS library is very poorly organized, even the layout I don't think it is very easy to understand and, since I have all these struggles, I am asking if anyone here can pass me some info (not the basic tutorials that are available in the webpage) about types of data, what every transformation means, some graphic or drawn example of what each command do, how does DeepAR handles training, inference, etc., etc...

This is mostly because I get this overall feeling that everyone is so much step ahead of me that I am either too unprepared for this, or some folks are catching up with something that I don't...even the way functions work and are called with GluonTS are very unnatural to me, and that exacerbates my feelings even more...

Thank you to all those that may help