What was released was a pre-print, not a publication. Every top ML conference requires releasing of code, but this is not the norm for CS research nor for research in general.

The bar isn't low, this is just a pre-print.