TextWorld: A Learning Environment for Text-based Games
arXiv:1806.11532
Abstract
We introduce TextWorld, a sandbox learning environment for the training and evaluation of RL agents on text-based games. TextWorld is a Python library that handles interactive play-through of text games, as well as backend functions like state tracking and reward assignment. It comes with a curated list of games whose features and challenges we have analyzed. More significantly, it enables users to handcraft or automatically generate new games. Its generative mechanisms give precise control over the difficulty, scope, and language of constructed games, and can be used to relax challenges inherent to commercial text games like partial observability and sparse rewards. By generating sets of varied but similar games, TextWorld can also be used to study generalization and transfer learning. We cast text-based games in the Reinforcement Learning formalism, use our framework to develop a set of benchmark games, and evaluate several baseline agents on this set and the curated list.
arXiv commentPresented at the Computer Games Workshop at IJCAI 2018, Stockholm
References in corpus (6)
- Grounded Language Learning in a Simulated 3D World
- Language Understanding for Text-based Games Using Deep Reinforcement Learning
- Neural Map: Structured Memory for Deep Reinforcement Learning
- Gated-Attention Architectures for Task-Oriented Language Grounding
- MazeBase: A Sandbox for Learning from Games
- CommAI: Evaluating the first steps towards a useful general AI
Cited by in corpus (46)
- Learning Dynamic Belief Graphs to Generalize on Text-Based Games
- Graph Constrained Reinforcement Learning for Natural Language Action Spaces
- ALFWorld: Aligning Text and Embodied Environments for Interactive Learning
- MeetUp! A Corpus of Joint Activity Dialogues in a Visual Environment
- Interactive Fiction Games: A Colossal Adventure
- Language Tasks and Language Games: On Methodology in Current Natural Language Processing Research
- Enhancing Text-based Reinforcement Learning Agents with Commonsense Knowledge
- How To Avoid Being Eaten By a Grue: Exploration Strategies for Text-Adventure Agents
- Action Assembly: Sparse Imitation Learning for Text Based Games with Combinatorial Action Spaces
- Playing Text-Adventure Games with Graph-Based Deep Reinforcement Learning
- How to Query Language Models?
- Comprehensible Context-driven Text Game Playing
- Why Build an Assistant in Minecraft?
- WordCraft: An Environment for Benchmarking Commonsense Agents
- Reinforcement Learning with External Knowledge by using Logical Neural Networks
- Building Dynamic Knowledge Graphs from Text-based Games
- I love your chain mail! Making knights smile in a fantasy game world: Open-domain goal-oriented dialogue agents
- Keep CALM and Explore: Language Models for Action Generation in Text-based Games
- Toward Automated Quest Generation in Text-Adventure Games
- I'm sorry Dave, I'm afraid I can't do that, Deep Q-learning from forbidden action
- NLPGym -- A toolkit for evaluating RL agents on Natural Language Processing Tasks
- Rank the Episodes: A Simple Approach for Exploration in Procedurally-Generated Environments
- Towards Solving Text-based Games by Producing Adaptive Action Spaces
- Ecological Semantics: Programming Environments for Situated Language Understanding
- Language is Power: Representing States Using Natural Language in Reinforcement Learning
- What Would Jiminy Cricket Do? Towards Agents That Behave Morally
- TeachMyAgent: a Benchmark for Automatic Curriculum Learning in Deep RL
- Towards Socially Intelligent Agents with Mental State Transition and Human Utility
- Zero-Shot Learning of Text Adventure Games with Sentence-Level Semantics
- Generating Interactive Worlds with Text
- Language (Re)modelling: Towards Embodied Language Understanding
- Process-Level Representation of Scientific Protocols with Interactive Annotation
- Instance based Generalization in Reinforcement Learning
- SocialAI: Benchmarking Socio-Cognitive Abilities in Deep Reinforcement Learning Agents
- A survey of benchmarking frameworks for reinforcement learning
- Modeling Worlds in Text
- Bootstrapped Q-learning with Context Relevant Observation Pruning to Generalize in Text-based Games
- Interactive Fiction Game Playing as Multi-Paragraph Reading Comprehension with Reinforcement Learning
- Playing by the Book: An Interactive Game Approach for Action Graph Extraction from Text
- Neuro-Symbolic Reinforcement Learning with First-Order Logic
- Sentiment Analysis for Reinforcement Learning
- Feudal Reinforcement Learning by Reading Manuals
- Generalization in Text-based Games via Hierarchical Reinforcement Learning
- LOA: Logical Optimal Actions for Text-based Interaction Games
- Learning to Generalize for Sequential Decision Making
- A Study of State Aliasing in Structured Prediction with RNNs