Multi-Agent Reinforcement Learning on SMAC Super Hard (test)
83.926h_vs_8z Win RateDDN
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DDNMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 83.92 | — | 97.22 | 91.48 | 95.4 | |
| DMIXMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 49.43 | — | 95.11 | 85.45 | 90.45 | |
| DPLEXMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 43.75 | — | 96.88 | 90.62 | 81.25 | |
| QMIXMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 12.78 | — | 92.44 | 84.77 | 37.61 | |
| IQLMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 0 | — | 68.92 | 2.27 | 84.87 | |
| VDNMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 0 | — | 89.2 | 63.12 | 85.34 | |
| QPLEXMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 0 | — | 96.88 | 78.12 | 75 | |
| DIQLMetric=Median win rate percentage (%), Test Runs=5, Training Timesteps=8 million2023.06 | 0 | — | 85.23 | 6.02 | 91.62 |