Skip to content
We release dataset collected for our research, code that implement neural network models described in the paper, and scripts to reproduce all of our results, and visualization tool for visualize dataset.
C++ Python JavaScript Lua Shell HTML Other
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data add copyright Aug 6, 2019
game add copyright Aug 6, 2019
pretrained_models add copyright Aug 6, 2019
scripts add copyright Aug 6, 2019
third_party
tube add copyright Aug 6, 2019
visual init commit Aug 5, 2019
.clang-format init commit Aug 5, 2019
.gitignore update readme Aug 5, 2019
.gitmodules init commit Aug 5, 2019
CMakeLists.txt init commit Aug 5, 2019
CODE_OF_CONDUCT.md init commit Aug 5, 2019
CONTRIBUTING.md init commit Aug 5, 2019
LICENSE init commit Aug 5, 2019
README.md Update README.md Aug 22, 2019

README.md

Hierarchical Decision Making by Generating and Following Natural Language Instructions

This is the repo for paper Hierarchical Decision Making by Generating and Following Natural Language Instructions.

Dependencies

We write our model and training code using PyTorch and its C++ interface. It is a known issue that some strange behaviors can happen if the compiler used for compiling this repo is differnet from the compiler used by the pre-built PyTorch due to the incompatibility between different versions of gcc. Therefore we recommand to build PyTorch from scratch before compiling this project.

We recommand to use conda and follow the instruction here to compile and install PyTorch from source first. Then install dependency for this project:

conda install lua numpy tqdm
conda install -c conda-forge tensorboardx

Get Started

Clone repo

git clone ...
git submodule sync && git submodule update --init --recursive

Download dataset & pretrained models

To downalod and unzip the original replays, processed json files, and dataset, from the following command. Note that it will take a while for the command to finish.

cd data
sh download.sh

To download some pretrained models used in the paper:

cd pretrained_models
sh download.sh
python update_path.py

Visualize dataset

We build a visualization tool that works directly with json file so that people can get a more intuitive view of the dataset and start working on it without compiling the game. Please go to the visual folder for detailed instructions on how to use it.

Train models

We put the shell scripts that can be used to re-train the model with configurations used in the paper in scripts/behavior_clone/scripts. Simply run command like

sh scripts/coach_rnn500.sh

to start training. The command needs to be run under behavior_clone folder. Normally it will take quite a while to load the dataset. For quick testing and debugging, one can add --dev at the end of the shell script to use the dev dataset instead, which contains only 2000 entries and thus much faster to load.

Run matches between models

To run matches between trained models, we first need to compile the game. Please see the "Build" and "Set env var" section for details. After the game is compiled, the following command can be used to launch matches between an RNN coach + RNN executor and zero executor (the one that does not use latent language).

python match2.py --coach1 rnn500 --executor1 rnn \
        --coach2 rnn500 --executor2 zero \
        --num_thread 500 --seed 9999

Structure

scripts

This is the main folder for our algorithm, containing code for data processing, model definition & training, and evaluation. See the readme file for each subfolder for more details.

visual

This contains a web tool for visualizing dataset from json so that we can have a peek of the dataset without compiling the game.

game

This folder contains the implementation of the game, including game logic, some built-in AIs used for collecting data, as well as necessary backends to extract features from game state for model evaluation.

tube

This folder defines a set of infra that dynamically batches data from various C++ game threads and transfer them between C++ and Python.

Build

mkdir build
cd build
export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
cmake ..
make

Note that we need to set the following before running any multi-threading program that uses the C++ torch::Tensor. Otherwise a simple tensor operation will use all cores by default.

export OMP_NUM_THREADS=1

Citation

If you use this repo in your research, please consider citing the paper as follows:

@article{DBLP:journals/corr/abs-1906-00744,
  author    = {Hengyuan Hu and
               Denis Yarats and
               Qucheng Gong and
               Yuandong Tian and
               Mike Lewis},
  title     = {Hierarchical Decision Making by Generating and Following Natural Language
               Instructions},
  journal   = {CoRR},
  volume    = {abs/1906.00744},
  year      = {2019},
  url       = {http://arxiv.org/abs/1906.00744},
  archivePrefix = {arXiv},
  eprint    = {1906.00744},
  timestamp = {Thu, 13 Jun 2019 13:36:00 +0200},
  biburl    = {https://dblp.org/rec/bib/journals/corr/abs-1906-00744},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

Copyright

Copyright (c) Facebook, Inc. and its affiliates. All rights reserved.

This source code is licensed under the license found in the LICENSE file in the root directory of this source tree.

You can’t perform that action at this time.