train_lda.py 文件源码

def LDAPageVctorizer(*,
                     n_topics: int,
                     min_df: int,
                     max_features: int,
                     max_iter: int,
                     ngram_range: Tuple[int, int],
                     vocabulary=None,
                     batch_size: int=4096,
                     verbose=1):
    vec = _vectorizer(min_df=min_df, max_features=max_features,
                      ngram_range=ngram_range, vocabulary=vocabulary)
    lda = LatentDirichletAllocation(
        learning_method='online',
        n_topics=n_topics,
        batch_size=batch_size,
        evaluate_every=2,
        verbose=verbose,
        max_iter=max_iter,
        n_jobs=1,
    )
    return make_pipeline(vec, lda)