Reinforcement Learning on Atari Full Suite 57 games
98.2% Games ImprovedKron
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| KronBase Algorithm=PQN2026.02 | 98.2 | 1,404 | 1.6 | |
| Baseline + MultiSkip + SIGRegBase Algorithm=PQN2026.02 | 92.9 | 1,695 | 11.4 | |
| Baseline + SIGRegBase Algorithm=PQN2026.02 | 89.5 | 889 | 9.3 | |
| Baseline + MultiSkip + SIGRegBase Algorithm=PPO2026.02 | 87.7 | 0.886 | 3,200 | |
| Baseline + SIGRegBase Algorithm=PPO2026.02 | 68.4 | 0.255 | 2,700 | |
| KronBase Algorithm=PPO2026.02 | 45.6 | -0.026 | 1,800 |