Multi-Agent Pathfinding on 5-agent 6x6 (evaluation set)
26Valid RateQwen3-4B + GRPO + LLM-as-Environment-Engineer
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Qwen3-4B + GRPO + LLM-as-Environment-Engineer2026.06 | 26 | 17.33 | |
| Qwen3-4B + GRPOtraining_config=random2026.06 | 21.33 | 14 | |
| Kimi-K2.52026.06 | 18 | 12 | |
| Grok-4.22026.06 | 13.33 | 9.33 | |
| GPT-5.42026.06 | 10 | 6 | |
| Gemini-3.1-Pro2026.06 | 6.67 | 4.67 | |
| Qwen3-4Bvariant=base2026.06 | 2.67 | 2.67 |