UnlockPickup on MiniGrid 8x8 (test)
1,886.8Reward AUCULPS
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ULPScalibration=uncertainty-aware2026.06 | 1,886.8 | 99.7 | 15.37 | 1.89 | 997 | 15,478 | |
| Uncalibrated LLMcalibration=none2026.06 | 1,113.77 | 72.3 | 38.57 | 1.12 | 723 | 38,568 | |
| Q-Learning (GRPO)Algorithm=GRPO2026.06 | 627.05 | 41.4 | 40.78 | 0.63 | 414 | 40,780 | |
| DQN (GRPO)Algorithm=GRPO2026.06 | 135.5 | 3 | 47.89 | 0.14 | 3 | 47,890 | |
| Linear RL (PPO)Algorithm=PPO2026.06 | -290 | 0.1 | 49.99 | -0.29 | 1 | 49,993 | |
| Unguided RL (PPO)Algorithm=PPO2026.06 | -358.18 | 0 | 50 | -0.36 | 0 | 50,000 |