Research map / Multi-agent systems and games / Human-AI and goal-conditioned
BabyAI: A Platform to Study the Sample Efficiency of Grounded Language Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use RL-generated demonstrations instead of bot-generated demonstrations | Behavioral cloning · GoToRedBallGrey | demonstrations required to solve: 8.43 - 12.4 -> 1.53 - 2.11 | Helped |
| Use RL-generated demonstrations instead of bot-generated demonstrations | Behavioral cloning · GoToRedBall | demonstrations required to solve: 49.7 - 62 -> 36.6 - 44.5 | Helped |
| Use RL-generated demonstrations instead of bot-generated demonstrations | Behavioral cloning · GoToLocal | demonstrations required to solve: 148 - 193 -> 74.2 - 81.8 | Helped |
| Replace IL with RL training | BabyAI level learners · GoToRedBallGrey, GoToRedBall, GoToLocal, PickupLoc, PutNextLocal, GoTo | training episodes or demonstrations to solve | Hurt |
| Add GoToLocal pretraining (ablation) | IL on GoTo · GoTo | demonstrations required to solve: 341 - 409 -> 183 - 216 | Helped |
| Add GoToObjMaze pretraining (ablation) | IL on GoTo · GoTo | demonstrations required to solve: 341 - 409 -> 444 - 602 | Hurt |
| Add GoToLocal and GoToObjMaze pretraining (ablation) | IL on GoTo · GoTo | demonstrations required to solve: 341 - 409 -> 173 - 216 | Helped |
| Add GoToLocal pretraining (ablation) | IL on PickupLoc · PickupLoc | demonstrations required to solve: 204 - 241 -> 71.2 - 88.9 | Helped |
| Add GoToLocal pretraining (ablation) | IL on PutNextLocal · PutNextLocal | demonstrations required to solve: 245 - 323 -> 188 - 231 | Helped |
| Provide bot demonstrations adaptively for missions the learner failed (ablation) | IL from Bot · GoToRedBallGrey | demonstrations required to solve: 8.43 - 12.4 -> 1.71 - 1.88 | Helped |
| Provide bot demonstrations adaptively for missions the learner failed (ablation) | IL from Bot · GoToRedBall | demonstrations required to solve: 49.7 - 62 -> 31.8 - 36 | Helped |
| Provide bot demonstrations adaptively for missions the learner failed (ablation) | IL from Bot · GoToLocal | demonstrations required to solve: 148 - 193 -> 93 - 107 | Helped |
Closest papers by meaning
- Ground-Compose-Reinforce: Grounding Language in Agentic Behaviours using Limited Data (NeurIPS 2025 · similarity 0.86)
- Self-Challenging Language Model Agents (NeurIPS 2025 · similarity 0.85)
- Offline Training of Language Model Agents with Functions as Learnable Weights (ICML 2024 · similarity 0.85)
- Preparing Lessons for Progressive Training on Language Models (AAAI 2024 · similarity 0.85)
- Hypothesis, Verification, and Induction: Grounding Large Language Models with Self-Driven Skill Learning (AAAI 2024 · similarity 0.85)
- Scaling Prompt Synthesis for Large Language Model Reasoning (ICML 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.