Research map / Multi-agent systems and games / Human-AI and goal-conditioned
One Life to Learn: Inferring Symbolic World Models for Stochastic Environments from Unguided Exploration
Zaid Khan, Archiki Prasad, Elias Stengel-Eskin, Jaemin Cho, Mohit Bansal
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add gradient-based parameter inference over law weights (ablation) | OneLife · Crafter-OO (state ranking and fidelity) | Rank@1: 13.0% -> 18.7% | Improved |
| add gradient-based parameter inference over law weights (ablation) | OneLife · Crafter-OO (state ranking and fidelity) | MRR: 0.429 -> 0.479 | Improved |
| add gradient-based parameter inference over law weights (ablation) | OneLife · Crafter-OO (state ranking and fidelity) | normalized edit distance: 0.057 -> 0.058 | No clear effect |
| use OneLife inference instead of PoE-World inference (ablation) | OneLife law synthesizer · Crafter-OO (state ranking and fidelity) | Rank@1: 10.8% -> 18.7% | Improved |
| use OneLife inference instead of PoE-World inference (ablation) | OneLife law synthesizer · Crafter-OO (state ranking and fidelity) | raw edit distance: 10.634 -> 8.764 | Improved |
| replace deterministic monolithic program synthesis with mixture of laws | WorldCoder vs OneLife · Crafter-OO (state ranking and fidelity) | MRR: 0.264 -> 0.479 | Improved |
| optimize fidelity only (PoE-World) without state ranking (ablation) | PoE-World · Crafter-OO (state ranking and fidelity) | Rank@1 vs random: 8.5% -> 10.8% | No clear effect |
| use learned world model for simulating plans | OneLife · Zombie Fighter, Stone Miner, Sword Maker | plan preference match with true env: 3 of 3 scenarios matched | Improved |
Closest papers by meaning
- Agent2World: Learning to Generate Symbolic World Models via Adaptive Multi-Agent Feedback (NeurIPS 2026 · similarity 0.88)
- Beyond Prediction: Steering VLM Agents with Retrospective World Modeling (NeurIPS 2026 · similarity 0.88)
- Simplifying Latent Dynamics with Softly State-Invariant World Models (NeurIPS 2024 · similarity 0.88)
- World Models Should Prioritize the Unification of Physical and Social Dynamics (NeurIPS 2025 · similarity 0.87)
- Learning POMDP World Models from Observations with Language-Model Priors (NeurIPS 2026 · similarity 0.87)
- Zero-shot World Models via Search in Memory (NeurIPS 2025 · similarity 0.87)
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.
