Reinforcement Learning on Atari 2600 (41 Game Evaluation Set)
2,152Alien ScoreRIMs-PPO
Evaluation results
| Method | Links | |||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RIMs-PPOArchitecture=RIMs, Algorithm=PPO, kA=52019.09 | 2,152 | 1,800 | 5,400 | 21,040 | 3,801 | 3,500,000 | 1,195 | 22,000 | 5,320 | 42 | 95 | 590 | 5,534 | 12,303 | 132,039 | 230,324 | -3.8 | 2,800 | 38 | 33 | 33,000 | 1,090 | -4 | 800 | 1,800 | 7,900 | 51,000 | 6,800 | 20 | 100 | 22,500 | 12,000 | 53,430 | 11 | 2,800 | 70,000 | 10,000 | 390,000 | 220,000 | 10,800 | 15,000 | |
| LSTM-PPOArchitecture=LSTM, Algorithm=PPO2019.09 | 1,612 | 1,000 | 4,000 | 3,090 | 1,611 | 3,280,000 | 1,153 | 21,000 | 698 | 30 | 80 | 593 | 4,600 | 11,000 | 138,000 | 26,320 | -3 | 1,600 | 20 | 29 | 7,000 | 500 | -5 | 425 | 13,000 | 10,000 | 28,000 | 4,200 | 20 | 90 | 22,000 | 7,500 | 53,000 | 3 | 1,600 | 35,000 | 4,000 | 70,000 | 90,000 | 3,833 | 200 |