I'm trying to fully understand the Leela's verbose-move-stats output:
Are the statements below correct?
- The move that will be played is
e2e4, since it's the one with the greatest count of visits (
- It also happens to have the greatest probability of being the best (
P: 10.01%), the best single-node evaluation (
V: 0.0556) and the best average child nodes evaluation (
- However, if the search continues, then the next node to be visited at this level of the tree will be
b1a3, because it has the best PUCT score (
- The number
+63besides the visit count of
e2e4is the win/draw/loss combined score of playout simulated games, such as in
Also, I have some questions:
- When/where will the playout score be used by the searching algorithm? I don't see it in the PUCT formula.
- What is the
e2e4? Is it just a move index?