Multi-Agent Path Finding on 4-agent MAPF-FrozenLake 6x6
36.67Valid RateQwen3-4B + GRPO + LLM-as-Environment-Engineer
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen3-4B + GRPO + LLM-as-Environment-EngineerBackbone=Qwen3-4B, Training=GRPO, Environment Design=Ours2026.06 | 36.67 | 22 | |
| Grok-4.22026.06 | 32 | 18 | |
| Qwen3-4B + GRPO (random)Backbone=Qwen3-4B, Training=GRPO, Config=single randomly sampled Round-02026.06 | 31.33 | 18.67 | |
| Kimi-K2.52026.06 | 28.67 | 16.67 | |
| GPT-5.42026.06 | 16 | 10 | |
| Gemini-3.1-Pro2026.06 | 12.67 | 10.67 | |
| Qwen3-4B (base)Backbone=Qwen3-4B, Status=untrained2026.06 | 2.67 | 2.67 |