Using Bert for Topic Modeling

BaruchG · December 3, 2019, 10:59pm

Does anyone know if it would be possible to use something like Bert for topic modeling? The main deep learning method that I have seen for topic modeling is lda2vec, but the pytorch port did not seem to have great things to say about the way that is was set up in the paper. Is there some other way to topic modeling using Deep Learning, preferably using fast.ai?

FeryET · December 31, 2019, 8:05am

Yes. In this code there is an approach to do topic modelling via Bert.

github.com

huseinzol05/NLP-Models-Tensorflow/blob/master/topic-model/2.bert-topic.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# !pip3 install bert-tensorflow"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Download BERT-Base model"
   ]
  },
  {
   "cell_type": "code",

This file has been truncated. show original