kardosdrur/latent-terms_paraphrase-MiniLM-L6-v2

This repository contains a topic model trained with the Turftopic Python library.

To load and use the model run the following piece of code:

from turftopic import load_model

model = load_model(kardosdrur/latent-terms_paraphrase-MiniLM-L6-v2)
model.print_topics()

Model Structure

The model is structured as follows:

LatentTermsVectorizer(autoencoder=TopKAutoEncoder(n_epochs=1, n_latent=10000,
                                                  random_state=42, top_k=15),
                      concept_labels=array(['going|process|high|given',
       'expense|financial|finance|documentation',
       'northern|tern|fort|enlisted', ..., '', 'wan|dong|won|hale',
       'started|began|needs|percent'], shape=(10000,), dtype='<U52'),
                      encoder='paraphrase-MiniLM-L6-v2')

Topics

The topics discovered by the model are the following:

None

Package versions

The model in this repo was trained using the following package versions:

Package	Version
scikit-learn	1.8.0
sentence-transformers	5.4.0
turftopic	0.25.1
joblib	1.5.2

We recommend that you install the same, or compatible versions of these packages locally, before trying to load a model.