WargamesBench Leaderboard¶
Reproducibility criterion: win-rate deviation ≤ 2 % across identical-seed runs.
Summary¶
| Policy | Mean Win Rate | Std | Scenarios | Episodes |
|---|---|---|---|---|
| test_baseline | 100.0% | 0.0% | 1 | 2 |
Per-scenario Results¶
| Scenario | Win Rate | Episodes | Mean Steps | Elapsed (s) |
|---|---|---|---|---|
| sym_4v4_clear | 100.0% | 2 | 500 | 0.0 |
Generated by WargamesBench — 20 canonical scenarios.