2019/06/03 by Hengyuan Hu, Hu, Hengyuan, Denis Yarats +7 · 1 voice · 4 citations
Computer Science · #AI-based Problem Solving and Planning #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #FOS: Computer and information sciences #Reinforcement Learning in Robotics #Speech and dialogue systems #cs.AI #cs.CL
paper · pdf · doi:10.48550/arxiv.1906.00744
openalex publication_date 2019/06/03 · arxiv published 2019/06/03 · openalex created_date 2019/06/07 · arxiv updated 2019/10/02 · openalex updated_date 2026/07/28
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.