This is naturally due to having large-ish dictionaries embedded: https://github.com/dropbox/zxcvbn/tree/master/data

Some people in the thread you linked to recommend lazy loading, another idea would be to use a bloom filter – I used this in a project to check server-side for hashes of (MMs of) common passwords.

Excuse me, it's less than one MB for actually useful (and rarely used) functionality, in a time when many websites routinely weigh in at several MegaBytes for tracking, advertisements, silly pictures and other cruft.

It is very noble that you are concerned about page size (seriously, I frequently travel and am on awfully slow WiFi or mobile networks (remember GPRS and Edge?) and hate wasteful websites with a passion) - but I'd much rather you remove all the other cruft (seriously, how much can two fields for "Enter username and password" take?) and include the most excellent zxcvbn.

It's pretty gigantic, so lazy-loading it is pretty much a hard requirement.

If you're using React/Redux, you probably have a build tool like Webpack setup, which makes loading async dependencies pretty painless.

If you want to use a server side tool, I wrote a port of it in Java (with quite a few extra features): https://github.com/GoSimpleLLC/nbvcxz

At least on the server side, size doesn't matter quite as much.