ValueError: list.remove(x): x not in list in model.hierarchical_topic_reduction()

I created a model with

model= Top2Vec(documents_text2, min_count = 4,
                       speed = "fast-learn", 
                       document_ids=document_ids2, 
                       workers = workers_n,keep_documents=False)

Then I tried to reduce the number of topics with

model.hierarchical_topic_reduction()

and get this error

model10 = model.hierarchical_topic_reduction(1000)
Traceback (most recent call last):

  File "<ipython-input-12-4ee6e263e4a0>", line 1, in <module>
    model10 = model.hierarchical_topic_reduction(1000)

  File "C:\Users\anaconda\.conda\envs\top2vec_final\lib\site-packages\top2vec\Top2Vec.py", line 1215, in hierarchical_topic_reduction
    ix_keep.remove(most_sim)

ValueError: list.remove(x): x not in list

[p-dre]

First I load the model an tested the reducing again. Same error

model = Top2Vec.load("Top2vec_model")

test = model.hierarchical_topic_reduction(200)
Traceback (most recent call last):

  File "<ipython-input-37-4e020d4f68c8>", line 1, in <module>
    test = model.hierarchical_topic_reduction(200)

  File "C:\Users\anaconda\.conda\envs\top2vec_final\lib\site-packages\top2vec\Top2Vec.py", line 1215, in hierarchical_topic_reduction
    ix_keep.remove(most_sim)

ValueError: list.remove(x): x not in list

After that I run:

top_vecs = model.topic_vectors
sizes = model.topic_sizes
smallest = sizes.sort_values(ascending=True).index[0]
res = np.inner(top_vecs[smallest], top_vecs)
most_sim = np.flip(np.argsort(res))[1]

most_sim
57

smallest
19904

[MackieBlackburn]

Could it be that most_sim == smallest because two topics have a similarity of 1, so somehow “smallest” ends up in index 1 of flip(res)? In that case the topic would already be removed by line 1214. I am in the process of verifying this but I am only getting this error when the number of topics is very large and that takes a long time to run.