Faster Initialization

[SaarthShah]

Hi, I am trying to deploy this package as one of the core dependencies of my custom spelling correction algorithm. The problem is that i am loading a corpus of 5 million sentences into this which adds ~ 10 seconds to our main backend startup time. Do you have any ideas on how this startup time can be reduced?

I feel like instead of reading a txt file we can do it even faster using a json?