Action-component payoff optimization on L3 warmup (off-diagonal)
4.17Per-Interaction PayoffReciprocity Gradient
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Reciprocity Gradient2026.05 | 4.17 | 93 | |
| Reciprocity Gradient2026.05 | 4.08 | 91 | |
| DPGTouter=15, Seeds=32026.05 | 1.63 | 36 | |
| DDPGTouter=15, Seeds=32026.05 | 1.58 | 35 | |
| TD3Touter=15, Seeds=32026.05 | 1.48 | 33 | |
| DPGTouter=15, Seeds=32026.05 | 0.55 | 12 | |
| TD3Touter=15, Seeds=32026.05 | 0.32 | 7 | |
| DDPGTouter=15, Seeds=32026.05 | 0.25 | 6 |