Honey-XPro Release

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
ChiefPushesWood
Posts: 58
Joined: Thu Nov 08, 2018 5:30 pm
Full name: Chief PushesWood

Re: Honey-XPro Release

Post by ChiefPushesWood » Tue Jun 04, 2019 9:21 pm

As a follow up, here is a screenshot of settings. I want this engine to have contempt for black only... Are these correct?
Image

CPW

MikeB
Posts: 3559
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Honey-XPro Release

Post by MikeB » Wed Jun 05, 2019 5:53 am

ChiefPushesWood wrote:
Tue Jun 04, 2019 9:21 pm
As a follow up, here is a screenshot of settings. I want this engine to have contempt for black only... Are these correct?
Image

CPW
Yes - you have it correct for black only with contempt set to 22 during analysis. I personally prefer contempt off during analysis. With respect to an earlier question, the scores in both McCain and Honey than from SF - one the search aND EVAL is slightly different , and two I have scaled the scoring down as it my opinion that the standard SF centipawn scoring is inflated and thus overly optimistic ( and likewise wise overly pessimistic if on the losing end). That is my judgment based on watching chess computer engines for over 30 years. I'm not saying I'm right and SF is wrong, I simply prefer a scoring evaluation that is more toned down - it's a matter of personal preference for me . since I also use the very same engines for my ICCF games. If see a plus 1 score and change , I expect to win the game. With the toned down scoring that is true , with SF scoring , not quite as true. The difference is about 1/3 ( not 1/3 precisely) . In the example scores, (sans position which I understand why) , most people will think a plus 7 is a rook and two pawns or 2 minor pieces plus a pawn advantage, a plus 5 score is usually rook or a minor piece with two pawns - a better way to evaluate the scoring function is to simply look at the static score before the search - take the given position and type "eval" without doing a search. Also high contempt tends to inflate the score as well and SF will be less likely to take a draw ( through repetition ) even though the draw might be its best result. A high contempt is used to maximize ratings against weaker opponents , but in ICCF chess ( where generally your opponent is equal) , it will cost you games. This is especially true if you use high contempt for black.

ChiefPushesWood
Posts: 58
Joined: Thu Nov 08, 2018 5:30 pm
Full name: Chief PushesWood

Re: Honey-XPro Release

Post by ChiefPushesWood » Wed Jun 05, 2019 2:22 pm

Thanks Mike. I appreciate the feedback. I don't mind the more conservative eval... just needed to know that's what it was.

As far as contempt goes, I think contempt is super useful if it is properly utilized. A bit of a different paradigm than what we're all 'used to' for sure. It's up to the human to put in the work and decide on draw lines. JMO.

Would anyone mind commenting on the merits of MPV vs 'NEXT BEST' with these newer engines?

CPW

ChiefPushesWood
Posts: 58
Joined: Thu Nov 08, 2018 5:30 pm
Full name: Chief PushesWood

Re: Honey-XPro Release

Post by ChiefPushesWood » Wed Jun 05, 2019 4:31 pm

Hey Mike,
I was messing around again this morning and I noticed something in the UCI setup for Honey in Chessbase 15. The UCI setup I posted yesterday was for a different program and it shows the variable 'Analysis Contempt' which I had set for black.
Trying to do the same thing in CB15, but the variable 'Analysis Contempt' is nowhere to be found. Why would this be and isn't it important if I want contempt for black in this setup?
Image

CPW

ChiefPushesWood
Posts: 58
Joined: Thu Nov 08, 2018 5:30 pm
Full name: Chief PushesWood

Re: Honey-XPro Release

Post by ChiefPushesWood » Thu Jun 06, 2019 2:55 pm

ChiefPushesWood wrote:
Wed Jun 05, 2019 4:31 pm
Hey Mike,
I was messing around again this morning and I noticed something in the UCI setup for Honey in Chessbase 15. The UCI setup I posted yesterday was for a different program and it shows the variable 'Analysis Contempt' which I had set for black.
Trying to do the same thing in CB15, but the variable 'Analysis Contempt' is nowhere to be found. Why would this be and isn't it important if I want contempt for black in this setup?
Image

CPW
Anyone on this?

Chief

MikeB
Posts: 3559
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Honey-XPro Release

Post by MikeB » Thu Jun 06, 2019 5:37 pm

ChiefPushesWood wrote:
Thu Jun 06, 2019 2:55 pm
ChiefPushesWood wrote:
Wed Jun 05, 2019 4:31 pm
Hey Mike,
I was messing around again this morning and I noticed something in the UCI setup for Honey in Chessbase 15. The UCI setup I posted yesterday was for a different program and it shows the variable 'Analysis Contempt' which I had set for black.
Trying to do the same thing in CB15, but the variable 'Analysis Contempt' is nowhere to be found. Why would this be and isn't it important if I want contempt for black in this setup?
Image

CPW
Anyone on this?

Chief
Each GUI decides on its own how many UCI options it elects to show and which ones it does show. You have that bring that up with the chessbase people, as there is nothing I can do about it, fwiw , I use xBoard and it's fine. They charge money for their products and have funds to provide support. I collect no money and have no funds or time to provide support on apps which I do not use.

ChiefPushesWood
Posts: 58
Joined: Thu Nov 08, 2018 5:30 pm
Full name: Chief PushesWood

Re: Honey-XPro Release

Post by ChiefPushesWood » Thu Jun 06, 2019 5:45 pm

Well ok then. Sorry to take up your time.

For the record, pretty much every other Stockfish and derivative that has Analysis Contempt available as an option shows in Chessbase GUI. Honey and McBrain and McCain do not... I'm guessing it's not CB but something on your end. <shrug>

CPW

MikeB
Posts: 3559
Joined: Thu Mar 09, 2006 5:34 am
Location: Pen Argyl, Pennsylvania

Re: Honey-XPro Release

Post by MikeB » Thu Jun 06, 2019 6:25 pm

ChiefPushesWood wrote:
Thu Jun 06, 2019 5:45 pm
Well ok then. Sorry to take up your time.

For the record, pretty much every other Stockfish and derivative that has Analysis Contempt available as an option shows in Chessbase GUI. Honey and McBrain and McCain do not... I'm guessing it's not CB but something on your end. <shrug>

CPW
I'll take a look - might be an easy fix.

Dann Corbit
Posts: 10206
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Honey-XPro Release

Post by Dann Corbit » Thu Jun 06, 2019 6:32 pm

Seemed to work when I tried it in Arena:
honey-options.png
honey-options.png (34.27 KiB) Viewed 1244 times
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

ChiefPushesWood
Posts: 58
Joined: Thu Nov 08, 2018 5:30 pm
Full name: Chief PushesWood

Re: Honey-XPro Release

Post by ChiefPushesWood » Thu Jun 06, 2019 6:55 pm

Dann Corbit wrote:
Thu Jun 06, 2019 6:32 pm
Seemed to work when I tried it in Arena:
Hey Dann.
Yeah it's working for me in other GUI's as well. Just not Chessbase which is my main software. Curious is all.

Chief

Post Reply