Loading topic model with multiple topic aspects changes their format

[zilch42]

Hi Maarten,

Really enjoying the multi aspect feature. Great work with that. I am finding that when loading a topic model saved with safetensors, the display format of the extra aspects changes. It's including a large float with each keyword.

from bertopic.representation import KeyBERTInspired
from bertopic.representation import PartOfSpeech
from bertopic.representation import MaximalMarginalRelevance
from sklearn.datasets import fetch_20newsgroups
from bertopic import BERTopic

# Documents to train on
docs = fetch_20newsgroups(subset='all',  remove=('headers', 'footers', 'quotes'))['data']

# The main representation of a topic
main_representation = KeyBERTInspired()

# Additional ways of representing a topic
aspect_model1 = PartOfSpeech("en_core_web_sm")
aspect_model2 = [KeyBERTInspired(top_n_words=30), MaximalMarginalRelevance(diversity=.5)]

# Add all models together to be run in a single `fit`
representation_model = {
   "Main": main_representation,
   "Aspect1":  aspect_model1,
   "Aspect2":  aspect_model2 
}
topic_model = BERTopic(representation_model=representation_model).fit(docs[0:200])

topic_model.save("example_model/", "safetensors")

topic_model.get_topic_info()

Restart python and load model...

from bertopic import BERTopic
topic_model = BERTopic.load("example_model/", embedding_model="sentence-transformers/all-MiniLM-L6-v2")
topic_model.get_topic_info()

[MaartenGr]

Thanks! It seems that I check for a tuple (which it was initially) but after loading in from json it converted it to a list instead. That should be a one-liner to fix.

[MaartenGr]

It should be fixed in the main branch 😄

[zilch42]

Yep looks good! Thanks!