Skip to content

This repository contains the source code of the EMNLP 2020 paper Interactive Fiction Game Playing as Multi-Paragraph Reading Comprehension with Reinforcement Learning.

Notifications You must be signed in to change notification settings

XiaoxiaoGuo/rcdqn

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Multi-Passage Reading Comprehension Deep Q-Network

About this repository

This repository contains the source code of the EMNLP 2020 paper Interactive Fiction Game Playing as Multi-Paragraph Reading Comprehension with Reinforcement Learning.

Dependencies

To get started with the framework, install the following dependencies:

  • Python 3.7
  • PyTorch (>=1.4.0)
  • Jericho (2.4.3 or 2.4.2)
  • spaCy (2.2.3)

Set up interactive fiction games simulator

The code base is built for Jericho 2.4.3 and 2.4.2. To install Jericho 2.4.2, run:

pip install jericho==2.4.2

The game roms are in the roms folder.

Training the models for interactive fiction games

  1. Prepare and pre-process the glove word embeddings.

1.1 Download the 100d GLoVe word embeddings and copy the file to the agents/glove folder.

1.2 Pre-process the glove word embeddings; run:

python glove_utils.py

or download the processed embeddings from here.

  1. Train the models for interactive fiction games. For example, to train on the game Zork1, run:
python train.py --batch_size 64 --env_id "zork1.z5"

License

MIT License

About

This repository contains the source code of the EMNLP 2020 paper Interactive Fiction Game Playing as Multi-Paragraph Reading Comprehension with Reinforcement Learning.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages