dbravender / tricksterstable-rs

Rust implementations of game engines featured in Trickster's Table
Other
4 stars 0 forks source link

Hotdog: tweak reward function and iterations #14

Closed dbravender closed 3 months ago

dbravender commented 3 months ago

When I initially ran experiments to improve the reward function the rules of the game were not correctly implemented. Try again!

dbravender commented 3 months ago

This was an easy fix. I had the default AI iterations set to 500. I will switch to a new reward function that takes into account scores and increase the iterations to 2,000.

 occurrences [current (1.0.65) bot score, next release bot score]
  13 [0, 3] 
  17 [3, 0] 
  63 [5, 0] 
  71 [0, 1] 
 148 [1, 0] 
 161 [2, 0] 
 193 [0, 2] 
 334 [0, 5]