I actually disagree with you here. I don't think the code is unreadable, it follows standard notation used in Machine Learning. If you read scientific papers yo

I actually disagree with you here. I don't think the code is unreadable, it follows standard notation used in Machine Learning. If you read scientific papers you will notice that e.g. variable names are the same as those used in mathematical formulas that everyone in the field is familiar with. The same goes for parameters, function names, and so on. They are standard notation/naming and only look confusing to people outside of the ML field. Giving them long uncommon names would actually be more confusing. As someone with experience in ML research I think this code is quite well written compared to what you typically see (a single function with hundreds of lines and dozens of if statements). I can immediately see what any of the functions does, and I haven't even read the paper.