BGonline.org Forums

Bot evals (XGR bug)

Posted By: Matt Cohn-Geier
Date: Wednesday, 14 December 2011, at 10:42 a.m.

In Response To: Bot evals (XGR bug) (Bob Koca)

Note the relatively high gammons that White wins in position #2. Also note that this doesn't occur on n-ply evals, it only happens on the minirollouts. What is surprising is that XGR++ gets this wrong. My first thought was that the bot was breaking the 6 pt to hit loose or something in the face of the weaker board but even on 1-ply or 2-ply the bot appears competent of playing out some variations. So why should XGR++ get it wrong?

Messages In This Thread

• Two positions
  Matt Cohn-Geier -- Tuesday, 13 December 2011, at 8:44 a.m.
  • Two positions
    Snua -- Tuesday, 13 December 2011, at 12:34 p.m.
  • Two positions
    pontiac -- Tuesday, 13 December 2011, at 2:36 p.m.
    • Two positions
      Sam Pottle -- Tuesday, 13 December 2011, at 5:41 p.m.
  • Bot evals (XGR bug)
    Matt Cohn-Geier -- Tuesday, 13 December 2011, at 9:22 p.m.
    • Bot evals (XGR bug)
      Timothy Chow -- Tuesday, 13 December 2011, at 9:26 p.m.
      • Bot evals (XGR bug)
        Matt Cohn-Geier -- Tuesday, 13 December 2011, at 9:45 p.m.
      • Bot evals (XGR bug)
        Bob Koca -- Wednesday, 14 December 2011, at 3:43 a.m.
        • Bot evals (XGR bug)
          Matt Cohn-Geier -- Wednesday, 14 December 2011, at 10:42 a.m.
          • Bot evals (XGR bug: no, but surely need to be improved)
            eXtreme Gammon -- Wednesday, 14 December 2011, at 1:32 p.m.
            • Bot evals (XGR bug: no, but surely need to be improved)
              Matt Cohn-Geier -- Wednesday, 14 December 2011, at 9:31 p.m.
              • Bot evals (XGR bug: no, but surely need to be improved)
                Matt Cohn-Geier -- Wednesday, 14 December 2011, at 9:51 p.m.
  • Rollouts
    Matt Cohn-Geier -- Wednesday, 14 December 2011, at 10:44 a.m.