Reputation: 869
I am using gensim for topic modeling. After training the lda model I call get_document_topics on a new document to get the topic distribution. However, for some documents, the return value is an empty list. Here is my code. Any idea what could have gone wrong?
topic_vector = [ x[1] for x in self.ldamodel.get_document_topics(new_doc_bow , minimum_probability=
0.0, per_word_topics=False)]
Upvotes: 0
Views: 466
Reputation: 869
Here is how I solved this issue:
First, make sure your gensim version is gensim-3.6.0 you can check this by running this command in your terminal:
pip freeze | grep gensim
If not, you can uninstall your gensim and install the newer version. Then in the file gensim/models/ldamodel.py you need to edit value of epsilon to a larger value.
DTYPE_TO_EPS = {
np.float16: 1e-5,
np.float32: 1e-35, # modify the value and set it to 1e-5
np.float64: 1e-100,
}
Upvotes: 1