One thing that can be important but tricky with some tokenization libraries is to un-tokenize, i.e. recover the offsets of the tokens in the original string. Is

One thing that can be important but tricky with some tokenization libraries is to un-tokenize, i.e. recover the offsets of the tokens in the original string. Is this supported in your library? How?