https://github.com/Eventual-Inc/Daft/blob/a945bba824281c374071d4851404ac22d02e3fca/tests/series/test_minhash.py#L74-L96 @Sammy Sidhu thoughts on

minhash

being different for empty strings? seems to be the same for everything else 😅

i think that is fine

Empty strings are actually quite important in minhashing because of the space delimiter which means that every space becomes a new empty string.. so it means that for texts with multiple whitespaces the minhash will definitely be changing. That said, i think between implementations the impact of this is not that big, unless people keep in disk some precomputed values from previous versions.