rime.models.zero_shot.bayes_lm

Classes

BayesLM(item_df[, max_num_candidates, ...])

p(y|x) propto lm(x|y) * p(y), where x is the last item in user history and y is a candidate item.