This project implements and evaluates ProdLDA (Product of Latent Dirichlet Allocation), a topic model based on the Variational Autoencoder (VAE) architecture. Unlike traditional models, ProdLDA uses neural networks to more flexibly capture the latent topic distribution in documents.
python nlp data-science machine-learning natural-language-processing deep-learning pytorch lda pyro topic-modelling prodlda
-
Updated
Sep 26, 2025 - Python