Grid-based maze navigation on MazeNamo 15x15 Hard
100Success RateLLM-Flax
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LLM-FlaxRules=LLM-Flax2026.04 | 100 | — | 10.979 | 36 | |
| ManualRules=Manual2026.04 | 90 | 10 | 11.901 | 36.2 | |
| LLM-FlaxConfig=LLM-Flax2026.04 | 1 | — | 10.979 | 36 | |
| LLM-FlaxConfig=LLM-Flax2026.04 | 1 | — | 10.979 | — | |
| ManualConfig=Manual2026.04 | 0.9 | — | 11.901 | 36.2 | |
| ManualConfig=Manual2026.04 | 0.9 | — | 11.901 | — | |
| PLOIConfig=PLOI2026.04 | 0.7 | — | 3.75 | 32 | |
| Pure FDConfig=Pure FD2026.04 | 0.333 | — | 34.37 | 29.3 | |
| Full LLM-FlaxConfig=Full LLM-Flax2026.04 | 0.2 | — | 23.1 | — |