Reinforcement Learning on Atari 49 ALE (test)
3,389Alien ScoreOPR
Evaluation Results
| Method | Links | |||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OPRSteps=10M, Epochs=1002026.03 | 3,389 | 2,571.2 | 5,285.7 | 3,410 | 2,632 | 880,040 | 506 | 33,800 | 5,701.2 | 60.8 | 83.4 | 111.7 | 23,521.5 | 20,160 | 150,110 | 79,421 | -0.6 | 434 | 21 | 33 | 9,216 | 1,648 | 1,845 | 33,831 | 3.2 | 17,505 | 12,300 | 6,791 | 6,830 | 2,500 | 2,970 | 8,754 | 14.2 | 100 | 36,652.5 | 7,998 | 21,210 | 23.4 | 1,720 | 1,945 | 3,700 | -0.6 | 7,560 | 127.1 | 67,114 | 1,380 | 28,558.3 | 5,470 | 6,720 | |
| SILSteps=50M, Frames=200M2026.03 | 2,242.2 | 1,362 | 1,812 | 17,984.2 | 2,259.4 | 3,084,781.7 | 1,137.8 | 25,075 | 2,366.2 | 31.1 | 99.6 | 452 | 7,559.5 | 6,710 | 130,185.8 | 10,140.5 | 21.5 | 1,205.1 | 55.8 | 32.2 | 6,289.8 | 23,304.2 | 1,874.2 | 33,156.7 | -2.4 | 310.8 | 2,888.3 | 10,614.6 | 34,449.2 | 1,100 | 4,025.1 | 14,958.2 | 20.9 | 661.2 | 104,975.6 | 14,306.1 | 57,071.7 | 10.5 | 2,456.5 | 2,951.7 | 31,309.2 | -17.3 | 10,811.7 | 340.5 | 53,314.6 | 0 | 461,522.4 | 7,088.3 | 9,164.2 | |
| A2C2026.03 | 1,859.2 | 739.9 | 1,981.4 | 16,083.3 | 2,056 | 3,032,444.2 | 1,333.7 | 10,683.3 | 3,931.7 | 31.2 | 99.7 | 501.6 | 3,857.8 | 3,464.2 | 129,715.8 | 18,331.4 | -0.5 | 0 | 39.1 | 0 | 339.5 | 9,358.5 | 329.2 | 28,008.1 | -4.3 | 399.2 | 1,563.3 | 8,883.9 | 32,507.5 | 5.8 | 2,843.4 | 11,174.2 | 20.8 | 210.8 | 17,605.2 | 13,036 | 39,874.2 | 3.2 | 1,795.2 | 2,466.1 | 57,371.7 | -10.3 | 5,346.7 | 305.6 | 48,131.8 | 0 | 391,241.6 | 4,196.7 | 124.2 | |
| PPO2026.03 | 1,475.6 | 757.2 | 4,970.6 | 4,531 | 2,195.1 | 3,010,196 | 1,118.8 | 16,053.3 | 5,046.1 | 40.9 | 94.6 | 339 | 4,418.1 | 2,837.2 | 104,169 | 11,370.4 | -13.5 | 694 | 23.4 | 32 | 314.1 | 4,546.1 | 414.7 | 29,104.9 | -2.1 | 511 | 5,318 | 7,488 | 26,640 | 0 | 2,650 | 7,241.6 | 20.7 | 100 | 14,104.2 | 7,963 | 34,400 | 15.4 | 1,680 | 1,000 | 35,000 | -14.4 | 4,500 | 200 | 35,000 | 0 | 5,000 | 4,000 | 5,000 | |
| ACPERSteps=50M, Frames=200M2026.03 | 390.2 | 424.8 | 818.2 | 3,533.1 | 1,780.1 | 58,012.5 | 1,203.2 | 15,025 | 2,602.4 | 59.3 | 100 | 118.5 | 7,790.1 | 1,307.5 | 19,918.8 | 4,777.5 | -9.8 | 3,113.3 | 59.8 | 31.4 | 2,342.5 | 3,919.5 | 627.5 | 13,299.1 | 0 | 598.1 | 5,875 | 11,323.2 | 20,485 | 0 | 1,016 | 2,888 | 20.9 | 100 | 657.2 | 2,224.5 | 8,925 | 7.7 | 804.5 | 729.5 | 1,107.5 | -17 | 3,952.5 | 270.7 | 9,562.5 | 0 | 21,797.7 | 1,550 | 4,278.8 | |
| DQNSteps=50M, Frames=200M2026.03 | 227.1 | 163.6 | 537.5 | 855 | 1,007.8 | 58,012.5 | 421.1 | 20,433.3 | 2,896.7 | 46.1 | 73.2 | 119.9 | 2,753.8 | 1,090 | 34,390 | 6,310.5 | -16.1 | 276.8 | -14.4 | 22.6 | 110 | 2,148 | 263.3 | 12,300 | -4.4 | 365.1 | 3,340 | 3,800 | 20,500 | 0 | 1,016 | 3,888 | 18.9 | 100 | 3,800 | 3,400 | 9,562.5 | 10.5 | 804.5 | 729.5 | 10,000 | -17 | 3,952.5 | 270.7 | 9,562.5 | 0 | 21,797.7 | 1,550 | 4,278.8 |