We explore using latent natural language instructions as an expressive and
compositional representation of complex actions for hierarchical decision
making. Rather than directly selecting micro-actions, our agent first generates
a latent plan in natural language, which is then executed by a separate model.
We introduce a challenging real-time strategy game environment in which the
actions of a large number of units must be coordinated across long time scales.
We gather a dataset of 76 thousand pairs of instructions and executions from
human play, and train instructor and executor models. Experiments show that
models using natural language as a latent variable significantly outperform
models that directly imitate human actions. The compositional structure of
language proves crucial to its effectiveness for action representation. We also
release our code, models and data.
Description
[1906.00744] Hierarchical Decision Making by Generating and Following Natural Language Instructions
%0 Journal Article
%1 hu2019hierarchical
%A Hu, Hengyuan
%A Yarats, Denis
%A Gong, Qucheng
%A Tian, Yuandong
%A Lewis, Mike
%D 2019
%K hierarchical nlp
%T Hierarchical Decision Making by Generating and Following Natural
Language Instructions
%U http://arxiv.org/abs/1906.00744
%X We explore using latent natural language instructions as an expressive and
compositional representation of complex actions for hierarchical decision
making. Rather than directly selecting micro-actions, our agent first generates
a latent plan in natural language, which is then executed by a separate model.
We introduce a challenging real-time strategy game environment in which the
actions of a large number of units must be coordinated across long time scales.
We gather a dataset of 76 thousand pairs of instructions and executions from
human play, and train instructor and executor models. Experiments show that
models using natural language as a latent variable significantly outperform
models that directly imitate human actions. The compositional structure of
language proves crucial to its effectiveness for action representation. We also
release our code, models and data.
@article{hu2019hierarchical,
abstract = {We explore using latent natural language instructions as an expressive and
compositional representation of complex actions for hierarchical decision
making. Rather than directly selecting micro-actions, our agent first generates
a latent plan in natural language, which is then executed by a separate model.
We introduce a challenging real-time strategy game environment in which the
actions of a large number of units must be coordinated across long time scales.
We gather a dataset of 76 thousand pairs of instructions and executions from
human play, and train instructor and executor models. Experiments show that
models using natural language as a latent variable significantly outperform
models that directly imitate human actions. The compositional structure of
language proves crucial to its effectiveness for action representation. We also
release our code, models and data.},
added-at = {2019-09-04T16:06:11.000+0200},
author = {Hu, Hengyuan and Yarats, Denis and Gong, Qucheng and Tian, Yuandong and Lewis, Mike},
biburl = {https://www.bibsonomy.org/bibtex/26c24ac5ae89341c80ed2e919891178d4/kirk86},
description = {[1906.00744] Hierarchical Decision Making by Generating and Following Natural Language Instructions},
interhash = {0dd4931cb8c743c495ff5d62bf039fb6},
intrahash = {6c24ac5ae89341c80ed2e919891178d4},
keywords = {hierarchical nlp},
note = {cite arxiv:1906.00744},
timestamp = {2019-09-04T16:06:11.000+0200},
title = {Hierarchical Decision Making by Generating and Following Natural
Language Instructions},
url = {http://arxiv.org/abs/1906.00744},
year = 2019
}