Build a Laplace-smoothed bigram language model from tokenized sentences and score supplied token sequences.
