PRIVATE ProteusSFX-AI tests

Code, algorithms, languages, construction...
User avatar
dorsz
Posts: 273
Joined: Mon Aug 30, 2021 8:43 pm

Re: PRIVATE ProteusSFX-AI tests

Post by dorsz » Thu Jan 25, 2024 4:39 pm

1. Yes. Besides the same time it must be the same number of cores and hash for both engine.
2. Yes.
3. Yes.
4. No. The same positions for both sides, but it does not have to be unbalanced to be fair. Unbalanced openings test suite are more fun and to reduce draw ratio.
5. Yes. The smaller the difference in strength, the more games are needed.

AlexChess
Posts: 504
Joined: Sat Jan 08, 2022 9:25 am
Real Name: Alex Morales

Re: PRIVATE ProteusSFX-AI tests

Post by AlexChess » Thu Jan 25, 2024 5:24 pm

MY SECOND PUBLIC LETTER TO CHESSBASE: NO ANSWER AGAIN :oops:

Chessbase Fritz 18 GUI and Playchess.com are really GARBAGE, ABANDONWARE. I have paid 99$ Chessbase Dragon 3.2 mainly for 6 months Premium Account access to Playchess.com The software has NEVER been fixed and Dragon 3.3 update is unavailable! If I play with it on Banksiagui orArena he starts only in demo mode 400 ELO lower than the Official Dragon 3.2 by Komodo! :shock: I cannot even play offline matches with it since di engine config loses UCI settings. I set 4 engines and it start using only 1. I must hardcode the setting on ProteusSFX-AI as a workaround. There is Spanish
Ruben user cherry-picking / farming / cheating on Playchess.com and nobody cares. I have reported him also for illegaly promoting Bitcoins on Chessbase servers no vital sign from Chessbase support, admins or moderators. Now also Solista il cherrypicking with his siblings accounts Lermerchess Powerbeast, Solista, Solistachess, EddyAlone and nobody cares, even TheChessTurk Sysop online!

Image

AlexChess
Posts: 504
Joined: Sat Jan 08, 2022 9:25 am
Real Name: Alex Morales

Re: PRIVATE ProteusSFX-AI tests

Post by AlexChess » Thu Jan 25, 2024 5:48 pm

dorsz wrote:
Thu Jan 25, 2024 4:39 pm
1. Yes. Besides the same time it must be the same number of cores and hash for both engine.
2. Yes.
3. Yes.
4. No. The same positions for both sides, but it does not have to be unbalanced to be fair. Unbalanced openings test suite are more fun and to reduce draw ratio.
5. Yes. The smaller the difference in strength, the more games are needed.
Thank you for the suggestions. After 300 games I will play the same match against BrainLearn 26.5 or better and Tactical 231223 or better, that like Swordfish 15.5 have scored +1 on previous match against ProteusSFX-AI. Then I'll restart the carousel with positions 301-600 and so on ;)

User avatar
dorsz
Posts: 273
Joined: Mon Aug 30, 2021 8:43 pm

Re: PRIVATE ProteusSFX-AI tests

Post by dorsz » Thu Jan 25, 2024 6:00 pm

At the end match your engine against new Stockfish.

User avatar
dorsz
Posts: 273
Joined: Mon Aug 30, 2021 8:43 pm

Re: PRIVATE ProteusSFX-AI tests

Post by dorsz » Thu Jan 25, 2024 6:01 pm

I don't comment playchess. This forum shows it is a dirty place.

AlexChess
Posts: 504
Joined: Sat Jan 08, 2022 9:25 am
Real Name: Alex Morales

Re: PRIVATE ProteusSFX-AI tests

Post by AlexChess » Thu Jan 25, 2024 9:02 pm

@dorsz I have made an error spending so much time on chess servers. They cause dependance like drugs and like drugs they are TOXIC and induce compulsive behaviors. I'm progressively abandoning them. Now I understand why some skilled friends avoid them. Instead I really would like to code something good with AI for chess ;)

YES, I will also test last Stockfish-dev builds with DOTS 8 suite. Note: After the first 36 games, next positions lead to much more won games! Right now 8-6 / 53 for me, but the match is still long. An update with all pgn games after every 100 games :)

ConRog
Posts: 32
Joined: Thu Jun 15, 2023 6:12 am
Real Name: Roger Constantinesco
Location: Tahiti, French Polynesia

Re: PRIVATE ProteusSFX-AI tests

Post by ConRog » Sat Jan 27, 2024 2:05 am

I have no confidence on results and ranking in online bot tournaments, because some guys have multiple bots.

1 PC with there best bot, and another PC playing with bots much less strong connected with different names, and they play each other to make there best bot gain ELO points!

They don't care about there weakest bots that are playing online only to give points to others.

If they have 2 PCs connected with different bots, no problem at all to do that.

AlexChess
Posts: 504
Joined: Sat Jan 08, 2022 9:25 am
Real Name: Alex Morales

Re: PRIVATE ProteusSFX-AI tests

Post by AlexChess » Sat Jan 27, 2024 9:15 am

Image

AlexChess
Posts: 504
Joined: Sat Jan 08, 2022 9:25 am
Real Name: Alex Morales

Re: PRIVATE ProteusSFX-AI tests

Post by AlexChess » Sat Jan 27, 2024 10:14 am

dorsz wrote:
Thu Jan 25, 2024 4:39 pm
1. Yes. Besides the same time it must be the same number of cores and hash for both engine.
2. Yes.
3. Yes.
4. No. The same positions for both sides, but it does not have to be unbalanced to be fair. Unbalanced openings test suite are more fun and to reduce draw ratio.
5. Yes. The smaller the difference in strength, the more games are needed.
ALL ProteusSFX-AI 240101 vs Swordfish 15.5 PGN games until now
I'm still leading with the minimal advantage ;)

https://fileport.io/N8Egh5sJZPnL

AlexChess
Posts: 504
Joined: Sat Jan 08, 2022 9:25 am
Real Name: Alex Morales

Re: PRIVATE ProteusSFX-AI tests

Post by AlexChess » Sat Jan 27, 2024 10:25 am

ConRog wrote:
Sat Jan 27, 2024 2:05 am
I have no confidence on results and ranking in online bot tournaments, because some guys have multiple bots.

1 PC with there best bot, and another PC playing with bots much less strong connected with different names, and they play each other to make there best bot gain ELO points!

They don't care about there weakest bots that are playing online only to give points to others.

If they have 2 PCs connected with different bots, no problem at all to do that.
I completely agree with you. I have reported bugs and bad users to Chessbase and Lichess more times, but they even not answer :roll:

Image

Post Reply