ValueError: Distance matrix cannot contain negative values

First off, massive congrats and a huge thank you for the library - I can't even express in words how good it is ;)

I'm working on a fairly large (~388k) multilingual dataset and am encountering a strange error when calculating the hierarchical topics relate to the distance matrix.

Here is my code:

`texts = data['text'].astype(str).to_list()
ids = data.index.to_list()

sentence_model = SentenceTransformer("paraphrase-multilingual-MiniLM-L12-v2", cache_folder=cache_folder)
embeddings = sentence_model.encode(texts, show_progress_bar=True)
np.save(f'{results_path}/embeddings', embeddings)

umap_model = UMAP(
n_neighbors=100,
n_components=5,
min_dist=0.01,
metric="cosine",
low_memory=True,
n_jobs=5,
random_state=42,
)

sws = list(get_stopwords(language='multi'))
vectorizer_model = CountVectorizer(stop_words=sws, ngram_range=(1,2), min_df=10)

bertopic_args = {
"language":'multilingual',
"embedding_model": sentence_model,
"vectorizer_model": vectorizer_model,
"umap_model": umap_model,
"verbose":True,
"low_memory":True,
"calculate_probabilities":False,
}
bertopic_model = BERTopic(**bertopic_args)
topic, probs = bertopic_model.fit_transform(texts, embeddings)
test_tuned = bertopic_model.get_topic_info()

data['topic'] = topic
hierarchical_topics = bertopic_model.hierarchical_topics(texts)
`

Here the error I get:

ValueError                                Traceback (most recent call last)
File <timed exec>:2

File....../lib/python3.11/site-packages/bertopic/_bertopic.py:939, in BERTopic.hierarchical_topics(self, docs, linkage_function, distance_function)
    937 embeddings = self.c_tf_idf_[self._outliers:]
    938 X = distance_function(embeddings)
--> 939 X = validate_distance_matrix(X, embeddings.shape[0])
    941 # Use the 1-D condensed distance matrix as an input instead of the raw distance matrix
    942 Z = linkage_function(X)

File ..../lib/python3.11/site-packages/bertopic/_utils.py:141, in validate_distance_matrix(X, n_samples)
    139 # Make sure its entries are non-negative
    140 if np.any(X < 0):
--> 141     raise ValueError("Distance matrix cannot contain negative values.")
    143 return X

ValueError: Distance matrix cannot contain negative values.

This code works fine for all the other datasets I've used it on.

Am I doing something wrong, or is there some issue?

Start by reproducing the report with the supplied BERTopic, UMAP, CountVectorizer, and embedding setup. Trace BERTopic.hierarchical_topics in bertopic/_bertopic.py into validate_distance_matrix in bertopic/_utils.py and inspect the distance matrix produced for this dataset. Done means the cause of the negative values is identified and hierarchical_topics handles the reported case without raising this error.