BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning

Abstract

Allowing humans to interactively train artificial agents to understand language instructions is desirable for both practical and scientific reasons, but given the poor data efficiency of the current learning methods, this goal may require substantial research efforts. Here, we introduce the BabyAI research platform to support investigations towards including humans in the loop for grounded language learning. The BabyAI platform comprises an extensible suite of 19 levels of increasing difficulty. The levels gradually lead the agent towards acquiring a combinatorially rich synthetic language which is a proper subset of English. The platform also provides a heuristic expert agent for the purpose of simulating a human teacher. We report baseline results and estimate the amount of human involvement that would be required to train a neural network-based agent on some of the BabyAI levels. We put forward strong evidence that current deep learning methods are not yet sufficiently sample efficient when it comes to learning a language with compositional properties.

arXiv commentAccepted at ICLR 2019

Cited by in corpus (29)

• A Survey of Zero-shot Generalisation in Deep Reinforcement Learning
• A Benchmark for Systematic Generalization in Grounded Language Understanding
• The NetHack Learning Environment
• CausalWorld: A Robotic Manipulation Benchmark for Causal Structure and Transfer Learning
• Unnatural Language Processing: Bridging the Gap Between Synthetic and Natural Language Data
• Approximate information state for approximate planning and reinforcement learning in partially observed systems
• Learning Language-Conditioned Robot Behavior from Offline Data and Crowd-Sourced Annotation
• Learning General Policies from Small Examples Without Supervision
• WordCraft: An Environment for Benchmarking Commonsense Agents
• A Consciousness-Inspired Planning Agent for Model-Based Reinforcement Learning
• Emergent Language: A Survey and Taxonomy
• ELLA: Exploration through Learned Language Abstraction
• Learning Behaviors with Uncertain Human Feedback
• Safe Reinforcement Learning with Natural Language Constraints
• Systematic Evaluation of Causal Discovery in Visual Model Based Reinforcement Learning
• Entropy Minimization In Emergent Languages
• TeachMyAgent: a Benchmark for Automatic Curriculum Learning in Deep RL
• Zero-Shot Compositional Policy Learning via Language Grounding
• Interactive Learning from Activity Description
• Bridging the Imitation Gap by Adaptive Insubordination
• Language Conditioned Imitation Learning over Unstructured Data
• Instance based Generalization in Reinforcement Learning
• Don't Do What Doesn't Matter: Intrinsic Motivation with Action Usefulness
• Working Memory Graphs
• Systematic Generalization on gSCAN: What is Nearly Solved and What is Next?
• Human-In-The-Loop Document Layout Analysis
• YES SIR!Optimizing Semantic Space of Negatives with Self-Involvement Ranker
• Compositional Networks Enable Systematic Generalization for Grounded Language Understanding
• Grounding 'Grounding' in NLP