There’s nothing inherently important to classful networking you learn that’s necessary for VLSM. They amount to common convention based on subnet size, and even then nearly nobody actually uses A or B sized subnets except as summary routes, which again, is not inherent to classful networking.
Classful networking has been obsolete for thirty years for good reason, you gain nothing from restricting yourself in that way.
Nothing actually uses classful networking anymore. Any situation where classful network concepts are implemented is necessarily limiting the capabilities of the network. As such it’s completely useless to bother spending time learning it.