For large scale evaluation (~1 million poker hands) involving students poker bots, we may need a time restriction for the action selection.
Estimation:
- 3 students, 1 baseline
- ~5 actions per set and poker hand
- 1 million poker hands
- parallel on 10 machines
- maximum of 24 hours for the evaluation
=> 4 * 5 * (1 million / 10) * x / 3600 = 24 => x = ~0.04 seconds per action
For large scale evaluation (~1 million poker hands) involving students poker bots, we may need a time restriction for the action selection.
Estimation:
=> 4 * 5 * (1 million / 10) * x / 3600 = 24 => x = ~0.04 seconds per action