Planning and Control on maze2d-umaze v1 (100 episodes, 300 steps/ep)
165.19ScoreInverter
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| InverterK=64, #NN-passes=52026.05 | 165.19 | 0.8 | 3.14 | 2.9 | 18.7 | 1.6 | |
| InverterK=32, #NN-passes=102026.05 | 164.77 | 0.34 | 2.67 | 4.3 | 31 | 2.7 | |
| InverterK=16, #NN-passes=192026.05 | 164.25 | 0.54 | 3.36 | 9.3 | 73.2 | 6.4 | |
| InverterK=128, #NN-passes=32026.05 | 161.64 | 2.17 | 3.19 | 1.9 | 11.4 | 1 | |
| DecisionLLM#NN-passes=3002026.05 | 145.2 | 35.32 | 33.3 | — | 10 | 873.9 | |
| SAC-N#NN-passes=3002026.05 | 130.59 | 16.52 | 1.8 | — | 539.3 | 47.1 | |
| Diffuser#NN-passes=642026.05 | 116.32 | 34.7 | 6.64 | 3 | 427.8 | 37.4 | |
| ReBRAC#NN-passes=3002026.05 | 106.87 | 22.16 | 0.21 | — | 63 | 5.5 | |
| EDAC#NN-passes=3002026.05 | 95.26 | 6.39 | 1.74 | — | 522.1 | 45.6 | |
| AWAC#NN-passes=3002026.05 | 82.67 | 28.3 | 1.28 | — | 382.5 | 33.4 | |
| IQL#NN-passes=3002026.05 | 42.11 | 0.58 | 1.69 | — | 508.4 | 44.4 | |
| TD3+BC#NN-passes=3002026.05 | 29.41 | 12.31 | 1.67 | — | 502.1 | 43.9 | |
| DT#NN-passes=3002026.05 | 18.08 | 25.42 | 3.57 | — | 1.07 | 93.7 | |
| BC-10%#NN-passes=3002026.05 | 12.18 | 4.29 | 1.25 | — | 374.9 | 32.8 | |
| BC#NN-passes=3002026.05 | 0.36 | 8.69 | 1.7 | — | 511.3 | 44.7 | |
| CQL#NN-passes=3002026.05 | -8.9 | 6.11 | 1.84 | — | 551.2 | 48.2 |