Zangdar - petit sorcier deviendra grand

Discussion of chess software programming and technical issues.

Moderator: Ras

Chessqueen
Posts: 5685
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

Re: Zangdar - petit sorcier deviendra grand

Post by Chessqueen »

Carbec wrote: Sun Jan 07, 2024 9:08 am Hello,

Many thanks for the lenghty explanation. I already got files from Zurichess. (quiet-labeled.epd, v6 and v7).
There are many things I am unfamiliar. I will go step by step, and try to understand all.
I have already read Andrew Grant paper, well, its like ancient greek :?

In effect, I began with the PeSTO tables, than later used the Weiss ones. They are very good, so I didn't
change for a time. But I think that now, I have to implement something if I want to test different parameters
in the evaluation.

I am a decent club player, although I don't play OTB now, age and illness came... I could put non-stupid
values for the parameters, but good ones ?? Also, Im not sure that "human" ideas in strategy fully apply
for computing. Its sure I prefer look at master games than computer games !

Philippe

I Set up Chesse-320 to Elo = 2975


Chessqueen
Posts: 5685
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

Re: Zangdar - petit sorcier deviendra grand

Post by Chessqueen »

Chessqueen wrote: Sun Jan 07, 2024 7:41 pm
Carbec wrote: Sun Jan 07, 2024 9:08 am Hello,

Many thanks for the lenghty explanation. I already got files from Zurichess. (quiet-labeled.epd, v6 and v7).
There are many things I am unfamiliar. I will go step by step, and try to understand all.
I have already read Andrew Grant paper, well, its like ancient greek :?

In effect, I began with the PeSTO tables, than later used the Weiss ones. They are very good, so I didn't
change for a time. But I think that now, I have to implement something if I want to test different parameters
in the evaluation.

I am a decent club player, although I don't play OTB now, age and illness came... I could put non-stupid
values for the parameters, but good ones ?? Also, Im not sure that "human" ideas in strategy fully apply
for computing. Its sure I prefer look at master games than computer games !

Philippe

Zangdar Vs Cheese-320 Elo=2975

Rank Engine Score Za Ch S-B
1 Zangdar-2.24-avx2 11.5/20 · ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· · ===0==1====01011=11= 97.75
2 Cheese-320-win 8.5/20 ===1==0====10100=00= · ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· ·· · 97.75


20 games played / Tournament is finished

Tournament start: 2023.12.24, 07:35:27
Latest update: 2024.01.07, 20:25:21
Site/ Country: DESKTOP-4QNC0GS, United States
Level: Blitz 3/2
Hardware: Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz with 15.9 GB Memory
Operating system: Windows 10 Home Home Edition (Build 9200) 64 bit
PGN-File: 1100=1600.pgn
Table created with: Arena 3.5.1

Carbec
Posts: 162
Joined: Thu Jan 20, 2022 9:42 am
Location: France
Full name: Philippe Chevalier

Re: Zangdar - petit sorcier deviendra grand

Post by Carbec »

Hi,

I started to tune Zangdar, but got some weird results.

Firstly, I noticed that the QueenValue was increasing a lot with iterations, and didn't seemed to stop.
So I redid the tuning, with 100000 iterations. Althought all parameters converged rapidly, I had to
wait at least 50000 iterations to see the QueenValue converge, at a very large value, like 2200+ (EG).

Is this normal ?
Also, Do I have to put the PieceValue, and the PSQT tables in the tuning ?
I noticed also that in several engines, people stop to tune at 10000 iterations;
In that case, QueenValue is not converged.

Thanks for help

Philippe
mathmoi
Posts: 290
Joined: Mon Mar 13, 2006 5:23 pm
Location: Québec
Full name: Mathieu Pagé

Re: Zangdar - petit sorcier deviendra grand

Post by mathmoi »

Bonjour Philippe,

I'm not an expert in engine tuning, only did it once with CLOP lots of years ago. Howerver I think that if you tune de PSQT and the piece values together you will get weird results. For exemple, the average value for a queen in the piece-square table could be -1300 and the value of a Queen need to be 2200 to compensate a real average value of 900. I'm not sure what is the prefered solution for this. You could comletely remove the piece values or rebalance the pieces values and PSQT in some ways. I would be interested to know what others do in this situation.

Bonne journée,