id2word returns weird numbers instead of words with get_topic_terms

[arshad115]

Problem description

While getting the top topic terms for my lda model, I get numbers instead of words. The corpus doesn't contain any numbers. Other models return the top words correctly, but for my corpus the results are always numbers.

Code for training:

    dictionary = Dictionary(texts)  # texts list has documents split by  words
    dictionary.filter_extremes(no_above=0.8, no_below=3)
    dictionary.compactify()

    corpus = [dictionary.doc2bow(text) for text in texts]
    model = LdaMulticore(corpus=corpus, num_topics=num_topics, passes=5)

Versions

version 3.8.1

Please provide the output of:
I used this code to get the top words, which works for other models.

def get_topic_top_words(lda_model, topic_id, nr_top_words=5):
    """ Returns the top words for topic_id from lda_model.
    """
    id_tuples = lda_model.get_topic_terms(topic_id, topn=nr_top_words)
    word_ids = np.array(id_tuples)[:,0]
    words = map(lambda id_: lda_model.id2word[id_], word_ids)
    return words

Output:

For topic 0, the top words are: 9143.0, 3313.0, 14517.0, 17358.0, 306.0, 8569.0, 2560.0, 20018.0, 320.0, 2807.0, 2838.0, 14388.0.
For topic 1, the top words are: 14.0, 405.0, 496.0, 90.0, 440.0, 270.0, 417.0, 143.0, 274.0, 145.0, 193.0, 817.0.
For topic 2, the top words are: 496.0, 405.0, 14.0, 90.0, 1713.0, 40.0, 199.0, 71.0, 193.0, 931.0, 248.0, 145.0.
For topic 3, the top words are: 47065.0, 143.0, 1080.0, 14.0, 1598.0, 90.0, 147.0, 272.0, 900.0, 248.0, 595.0, 2580.0.
For topic 4, the top words are: 14.0, 1335.0, 1061.0, 339.0, 959.0, 2366.0, 809.0, 352.0, 147.0, 1312.0, 268.0, 989.0.
For topic 5, the top words are: 9143.0, 19553.0, 14.0, 759.0, 437.0, 22.0, 456.0, 405.0, 1971.0, 1704.0, 1821.0, 27175.0.
For topic 6, the top words are: 2560.0, 1229.0, 14.0, 93649.0, 817.0, 496.0, 71.0, 143.0, 270.0, 231.0, 914.0, 455.0.
For topic 7, the top words are: 417.0, 817.0, 150.0, 2644.0, 14.0, 147.0, 405.0, 622.0, 183.0, 2535.0, 517.0, 496.0.
For topic 8, the top words are: 183.0, 469.0, 14.0, 1331.0, 143.0, 389.0, 696.0, 332.0, 39.0, 104.0, 405.0, 886.0.
For topic 9, the top words are: 14.0, 921.0, 22.0, 405.0, 545.0, 417.0, 183.0, 24.0, 484.0, 71.0, 231.0, 143.0.

[arshad115]

I forgot to pass the dictionary to the LDA model, so it prints out ids of the tokens instead of words. You can get the words from the dictionary by passing the id.