deeds_chess_system

[deeds]

*** DeeDs Chess System ***

For all the positions of a game, the evaluation of the engine does not have the same precision.
The same goes for experience data, sometimes there are many, sometimes there are few.
Sometimes they are reinforced with refined scores at very great depths.
So with values fixed for all the positions of a game like those of the "Experience Book Min Depth / Eval Importance" options, it is not optimal.
Engine's moves are not always in conformity with its experience data.

D.C.S has been designed to make the best use of experience data in each position.
As i don't read the C/C++ language and i don't know anyone who can modify the source codes of the private engines,
i coded a tool which drives several engines simultaneous, which chooses the most efficient moves or the moves of the reinforced data or the moves from the opening book.
The move's choice depends on the position of the game, the among/quality of experience data, what contains the opening book, etc.

D.C.S passes UCI commands regarding opening book options to the "book_engine" configured to choose moves from the opening book.
D.C.S passes other UCI commands regarding engines settings to the "playing_engine" configured to analyze the positions, choose moves from the experience file, save new moves and update the experience data, etc.

D.C.S needs an INI file which looks like this :


Throughout this "proof-of-concept" phase, i used Eman 8.40 as playing_engine, asmFish 291118 as book_engine and Depth4_180423.exp as the experience file.
The rest of the D.C.S. features/details remains under NDA...

[deeds]

All the openings lists :
- GHPPN's 15 most popular openings
- 20 first legal moves
- Albert Silver's 25 gambit suit
- Jeroen Noomen's 50 tcec season 22 superfinal openings

[deeds]

- Stefan Pohl's 100 SALC_V3_10moves openings
- Jeroen Noomen's 100 best 2017 openings
- YHWH's 130 openings
- Jeroen Noomen's 2017 selection 100 openings

[hungtranvn]

Dear Deeds
Why are you using Eman 8.40 and not using a newer version, or another engine.
Can I use Little Beast 3 sl with Eman experience?

[deeds]

Dear Deeds
Why are you using Eman 8.40 and not using a newer version, or another engine.

Sorry but i'm training 5 different engines at the time and i trained until 6 different engines if i count JudaS...

Dear Deeds
Why are you using Eman 8.40 and not using a newer version, or another engine.

Because :
- since the Normalize evaluation patch of Nov 5, 2022, the scores of Eman 8.50 (and newer release) are based on a WDL model (based on fishtest's games = Stockfish selfplay)
- this WDL model is not very accurate
- this WDL model is modified several times/year
- at this period (November, 2022), i had already got between 14M to 17M "pawn-based" Eman scores and i don't want to pollule my Eman.exp file with "ephemeral wdl-based" scores from Stockfish selfplay games
- depending on the chosen opening list, learning engines showed gains from +50 to +200 elo and since there aren't +50 elo difference between Eman 8.40 and Eman 8.92, there wasn't urgency to loose the fruits of my 7 learning sessions
- i reinforced (and i'm reinforcing yet) at D40 the experience data of Eman 8.40 which has proven its fantastic resistance on odd TC and against books, here again, there wasn't urgency

[deeds]

Can I use Little Beast 3 sl with Eman experience?

I advise you to use a non-merged experience file trained by its own engine.

There are already too many unfair people like :
- Alex who merged igor.exp (eman engine) with sarona.exp (sarona engine) and renamed it to ProteusSF.exp
- Sedat who renamed Khalid's eman.exp to StockfishMZ.exp
- Sedat who renamed ProteusSF.exp to StockfishMZ.exp
- etc

[hungtranvn]

Thanks a lot.
But I use the Little Beast 3 SL with the new Experience File from the beginning, not Merged Experience File, according to you, is it good?

[deeds]

If your engine manages to learn, that's the basis.

Then :

- if your engine scores are based on an ephemeral WDL model, each new WDL model will pollute more and more your experience file.
I know at least one engine trainer in this situation => he no longer updates his engine at training.

- if your engine scores are based on the pawn value, as it is an international standard fixed value, no problem.
The pawn-based scores are intelligible, comparable between most databases/experience files/chess literature, editable by all the tools that I have already "generously shared at least 27 times" on github.

Good training !

[hungtranvn]

Dear Deeds!
I decided to use Eman 8.40, but when I find it again, some old links are dead, can you send it to me?
Thank you very much

[deeds]

Known limitations with Eman 8.40 :

- [bug] moves like Ra1-b1, Ra1-c1, etc. are not always written in Eman.exp
- [net] latest compatible network is nn-e1fb1ade4432 (22-04-2023)
- [exp] "SugaR Experience version 2" file format. If the experience file is written under the old v1 format => Eman will convert it to v2 at startup
- [exp] max filesize checked until 10 GB, 450M moves, etc. (see 22pieces_d10.exp)
- [cutechess-cli] concurrency=1 (Eman only reads/writes an unique exp file at the time). Create a learning farm for several Eman's instances at the time.
- [hash] min hash 1 GB when engine has to load/save an exp file

If you agree with all that, here is the link.

Good training !