How to Motivate Your Dragon: Teaching Goal-Driven Agents to Speak and Act in Fantasy Worlds

TL;DR

This paper develops goal-driven agents capable of natural language communication and actions in fantasy worlds by extending a large-scale text-game with quests, pre-training, and a factorized action space, evaluated through zero-shot tests.

Contribution

It introduces a novel reinforcement learning system that combines language modeling, commonsense reasoning, and a factorized action space for fantasy agents.

Findings

Agents act consistently with their motivations

Agents talk naturally in line with goals

Zero-shot evaluation shows promising results

Abstract

We seek to create agents that both act and communicate with other agents in pursuit of a goal. Towards this end, we extend LIGHT (Urbanek et al. 2019) -- a large-scale crowd-sourced fantasy text-game -- with a dataset of quests. These contain natural language motivations paired with in-game goals and human demonstrations; completing a quest might require dialogue or actions (or both). We introduce a reinforcement learning system that (1) incorporates large-scale language modeling-based and commonsense reasoning-based pre-training to imbue the agent with relevant priors; and (2) leverages a factorized action space of action commands and dialogue, balancing between the two. We conduct zero-shot evaluations using held-out human expert demonstrations, showing that our agents are able to act consistently and talk naturally with respect to their motivations.