Research map / Multi-agent systems and games / Human-AI and goal-conditioned
The Neural MMO Platform for Massively Multiagent Research
Joseph Suarez, Yilun Du, Clare Zhu, Igor Mordatch, Phillip Isola
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 |
|---|---|---|---|
| train on 1 map vs 32 procedurally generated maps (ablation) | PPO LSTM · unseen maps | average lifetime: 52.73 -> 61.56 | Improved |
| 32 vs 256 training maps (ablation) | PPO LSTM · unseen maps | average lifetime: 61.56 -> 61.91 | No clear effect |
| 16384 training maps vs 32 (ablation) | PPO LSTM · unseen maps | average lifetime: 61.56 -> 53.25 | No clear effect |
| larger training population cap (4 -> 256) (ablation) | PPO LSTM · SmallMaps resource+progression | lifetime: 89.26 -> 227.36 | Improved |
| larger training population cap (4 -> 32) (ablation) | PPO LSTM · SmallMaps resource+progression | explore: 7.66 -> 11.41 | Improved |
| remove exploration routine from scripted forage bot (ablation) | Scripted Forage · SmallMaps | achievement: 7.56 -> 4.34 | Hurt |
| remove exploration routine from scripted combat bot (ablation) | Scripted Combat · SmallMaps | achievement: 3.45 -> 2.72 | Hurt |
| add combat heuristic to scripted forage bot | Scripted bots · SmallMaps | lifetime: 252.38 -> 76.52 | Hurt |
| transfer SmallMaps policy to LargeMaps zero-shot | Neural MMO policy · LargeMaps | lifetime: 57.12 -> 73.48 | Mixed |
| evaluate LargeMaps policy zero-shot on SmallMaps | Neural MMO policy · SmallMaps | lifetime: 51.86 -> 35.24 | Hurt |
| scale maps from SmallMaps to LargeMaps for neural forage | Neural Forage · Neural MMO | lifetime: 132.14 -> 356.13 | Improved |
| increase test-time population cap for agents trained in small populations | PPO LSTM · Neural MMO | lifetime | Hurt |
Closest papers by meaning
- CrafterDojo: A Suite of Foundation Models for Building Open-Ended Embodied Agents in Crafter (NeurIPS 2026 · similarity 0.84)
- MindForge: Empowering Embodied Agents with Theory of Mind for Lifelong Cultural Learning (NeurIPS 2025 · similarity 0.84)
- AgentOdyssey: Open-Ended Long-Horizon Text Game Generation for Test-Time Continual Learning Agents (NeurIPS 2026 · similarity 0.84)
- ADAM: An Embodied Causal Agent in Open-World Environments (ICLR 2025 · similarity 0.83)
- VirtualEnv: A Platform for Embodied AI Research (AAAI 2026 · similarity 0.83)
- AgentSquare: Automatic LLM Agent Search in Modular Design Space (ICLR 2025 · similarity 0.83)
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.
