Skip to content

Transformer-based Conditional Variational Autoencoder for Controllable Story Generation

Notifications You must be signed in to change notification settings

fangleai/TransformerCVAE

Repository files navigation

TransformerCVAE

This repository contains source code for paper Transformer-based Conditional Variational Autoencoder for Controllable Story Generation:

@article{fang2021transformer,
  title={Transformer-based Conditional Variational Autoencoder for Controllable Story Generation},
  author={Fang, Le and Zeng, Tao and Liu, Chaochun and Bo, Liefeng and Dong, Wen and Chen, Changyou},
  journal={arXiv preprint arXiv:2101.00828},
  year={2021}
}
  1. get source data (Arxiv, Yelp, WritingPrompts, WikiPlots).
  2. data pre-processing (data/).
  3. training (choose from several different implementations on parallelism and precision: train.py, train_dist.py, train_dist_half.py).
  4. generation, evaluation and analysis (generate.py/generate_prefix.py, eval_ppl.py/eval_ppl_prefix.py, tsne_plot.py).

Contact: lefang@buffalo.edu

Update on 2022: If you encounter package version issue, sorry for that I don't have a requirements.txt with exact versions. I used this package: https://github.com/nvidia/apex and an old pytorch version compatible with it at that time, say pytorch=0.4 (not 100% sure).

About

Transformer-based Conditional Variational Autoencoder for Controllable Story Generation

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages