https://github.com/akashgit/autoencoding_vi_for_topic_models

Tensorflow implementation for prodLDA and NVLDA.
https://github.com/akashgit/autoencoding_vi_for_topic_models

Last synced: about 2 months ago
JSON representation

Tensorflow implementation for prodLDA and NVLDA.

• Host: GitHub
• URL: https://github.com/akashgit/autoencoding_vi_for_topic_models
• Owner: akashgit
• License: mit
• Created: 2016-11-15T19:51:50.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2021-04-19T20:39:53.000Z (over 3 years ago)
• Last Synced: 2024-05-21T00:49:49.985Z (4 months ago)
• Language: Python
• Homepage: http://openreview.net/forum?id=BybtVK9lg
• Size: 5.46 MB
• Stars: 250
• Watchers: 13
• Forks: 52
• Open Issues: 8
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-topic-models - ProdLDA - Original TensorFlow implementation of Autoencoding Variational Inference (AEVI) for Topic Models [:page_facing_up:](https://arxiv.org/pdf/1703.01488.pdf) (Models / Embedding based Topic Models)

README

        
# Autoencoding Variational Inference for Topic Models

__UPDATE__

>> Pyro added a prodlDA tutorial: https://pyro.ai/examples/prodlda.html

>> AVITM is now available in OCTIS at https://github.com/MIND-Lab/OCTIS

Please consider using OCTIS and Pyro versions as they are more upto date.

1. As pointed out by [@govg](https://github.com/govg), this code depends on a slightly older version of TF. I will try to update it soon, in the meantime you can look up a quick fix [here](https://github.com/akashgit/autoencoding_vi_for_topic_models/issues/5) for working with newer version of TF or (3) and (2) below if you'd rather prefer Keras or PyTorch.

2. [@nzw0301](https://github.com/nzw0301) has implemented a [Keras](https://github.com/nzw0301/keras-examples/blob/master/prodLDA.ipynb) version of prodLDA.

3. [@hyqneuron](https://github.com/hyqneuron) recently implemented a [PyTorch](https://github.com/hyqneuron/pytorch-avitm) version of AVITM. So check out his repo.

4. Added `topic_prop` method to both the models. Softmax the output of this method to get the topic proportions.

---

#### Code for the ICLR 2017 paper: Autoencoding Variational Inference for Topic Models

---

#### > [Arxiv](https://arxiv.org/abs/1703.01488)

#### > [OpenReview](http://openreview.net/forum?id=BybtVK9lg)

---

###### This is a tensorflow implementation for both of the Autoencoded Topic Models mentioned in the paper.  

---

To run the `prodLDA` model in the `20Newgroup` dataset:

> `CUDA_VISIBLE_DEVICES=0 python run.py -m prodlda -f 100 -s 100 -t 50 -b 200 -r 0.002 -e 200`

Similarly for `NVLDA`:

> `CUDA_VISIBLE_DEVICES=0 python run.py -m nvlda -f 100 -s 100 -t 50 -b 200 -r 0.005 -e 300`

Check `run.py` for other options.