Closed GoogleCodeExporter closed 8 years ago
Hi guy,
I'd like to implement this change, but this might influence the recorded
Weka-Model of MCTSBot.
Do you know how CSPoker implemented NumRaises ?
Easiest thing might be to just recreate the WekaModel after I committed the
patch, but I don't know how to do it (nor do I have the original HandHistories
you used) ?
Original comment by bluegasp...@gmail.com
on 13 Nov 2010 at 9:39
I do not know how CSPoker works, but i i've planned to learn it. So you can
wait, until that happens, and then implement NumRaises without breaking MCTSBot.
Original comment by mihano...@gmail.com
on 14 Nov 2010 at 1:16
This issue was closed by revision abc30a54f8.
Original comment by bluegasp...@gmail.com
on 14 Nov 2010 at 10:52
I just realized, that MCTSBot isn't calling #getNumRaises at all as it keeps
its own statistics.
Change is thus safe (makes MCTSBot play a little bit less aggressive).
I checked against PA too and made same changes to your proposal:
- getNumRaise=1 already after small blind
- getNumRaises gets reset after each round, I implemented that too.
- some more adjustments to keep JUnit-Tests green (the logic to detect the
possible bigblind-action preflop, looks at numRaises).
Original comment by bluegasp...@gmail.com
on 14 Nov 2010 at 10:57
Original issue reported on code.google.com by
mihano...@gmail.com
on 6 Nov 2010 at 3:50