SedatChess

[Sedat Canbaz]

You are welcome )

Now I feel much better..I mean all in friendly way )

Agreed...500+ games (per opening) would be much better!
But in my guess, even several games (per same opening) can be
Ranked at higher places..than the ones which never tested..

Why I say like this (if check latest current test-400 games),
We see that even via small number of learned games
Eman's performance is increased...and I strongly believe that
Pre-learned Eman. exp file is helped a lot...

Btw, BIG thanks to you, for your idea: Experience MultiPV OFF
And it seems that Eman 8.91 is stronger under these settings...

On other hand,
Nowadays I have no much free time for chess but later,
Maybe in September, I may test RapTora 2.2 as without learning:
https://sites.google.com/site/computers ... t-nn-cs-33

Then we will have chances to compare, e.g
RapTora 2.2 NON-Learning will be Champ or not..
And frankly in my estimation: no much chances!

About your book question,
You can try to test the full version of Unique - 300 openings:
https://sites.google.com/site/computers ... test-suite

Greetings )

In friendly way much better for sure... I dont like arguments...

Well yeah several games (not much but some as you say) can help but normally are not enough, but for sure I believe they can help...

Regarding the Experience MultiPV OFF, you are welcome, this option is better turned off because of a simple but logical reason, if it is set to true, the engine saves a lot of useless, shallow depth entries, that are not even played in the game. The consequence of this is, useless experience data that makes the experience file bigger and slows down the engine in future games / matches. Thanks for testing it!

I hope that after this people stop doing tests with Experience MultiPV ON, because people mostly believe you, and you showed that ON is worse...

Thanks for the test-suite, now its time to train Stockfish + learning and Eman...

Regards,

Ibai

Not at all..it is my pleasure...

Happy testings )

[deeds]

EDIT:
Experience MultiPV OFF is performed as best (than all)

Engine's trainers know this since months/years :



That's why i confirm what i said previously :

... We all understand that Sedat-like people...have missed the train of engine's training...

So it means 100% of the fake tourneys called "Experience Files vs Books" managed by Sedat-like people are unfair and even lied to people about the superiority of learning engines. As these testers are unenlighted on learning feature/engines and unenlighted on experience data/file formats, we expect they continue to run their little "warmup" (2 games/opening) believing they ran true fair tourneys !

[Sedat Canbaz]

EDIT:
Experience MultiPV OFF is performed as best (than all)

Engine's trainers know this since months/years :

Hehe...you are here ))
Frankly I was wondering where are you..LOL...

Well and to be more clear,

As usual..it seems you missed,
My goal is to test the latest engines mainly via default settings
Plus I am a Tournament manager, not as Engine trainer!
That's the difference...if still not clear,
I am not as Eman beta tester...hopes helps...

Plus as other funny comment by you is that,
You love to say, it is not fair...e.g such as:
Exp files are NOT trained vs books...

But you are missing one important point:
Current participated books vers are not trained vs current exp !!!
And I wonder much,
What kind of BULLSHIT you'll say about this ?))

Ok..anyhow, SCCT Book/EXP tour is paused, but
It's planning to be resumed after 2 months...
So in that long period of time,

Anyone who is interested...have plenty of time to train...
Book vs EXP etc...

Ah one thing more, while you are here...
What is your opinion about parallel matches via Eman?
As far as I remember, the best expert (Deeds) said such as :
Eman is not able to learn properly (via Concurrent matches)
Maybe I am missing something ? LOL...

[deeds]

As usual..it seems you missed,
My goal is to test the latest engines mainly via default settings

As expected..it seems you missed that learned engines on their default settings are playing on training mode, not on tournament mode !

I am not as Eman beta tester...hopes helps...

But until now you didn't tested Eman, you only trained Eman at 2 games/opening loool

But you are missing one important point:
Current participated books vers are not trained vs current exp !!!

No, it's you who missed this important point : books = static chess when learning engines = dynamic chess !
And good luck to train a book against a learning engine which doesn't play the same moves (=unpredictable playing style of the learning engine), which learns in realtime the most efficient book's moves (=predictable playing style of the book).

Ah one thing more, while you are here...
What is your opinion about parallel matches via Eman?

Always the same opinion since my first contact with the world of learning engines : when i'm opening a door, no one is closing the same door !

When an Eman's instance is writing some experience data on the experience file, any other Eman's instance is reading these experience data from the same experience file. So all the other Eman's instances start to play with the old and incomplete experience data. The more Eman's instances, the more experience data is lost.

As your conditions lose experience data, you take great care not to announce to Rulleeeee the shameful rate of the non-conformity of the moves played at SCCT with the Rulleeeee's experience data.

Reminder : default settings = training mode = 70-75% of conformity with experience data = far from the true strength of the experience file. It's like you changed 25-30% of the data of a "blablabla" database and say to people that the result represents the "blablabla" database at 100%, it's a lie ! As you did such lies several time, you're a liar !

Furthermore, we won't be surprise if the opening's list used at SCCT showed a very poor experience rate with the Rulleeeee's experience file.

[Sedat Canbaz]

As expected..it seems you missed that learned engines on their default settings are playing on training mode, not on tournament mode !

Hey Chameleon (Deeds),

You'll never stop with BULLSHIT comments?

Anyhow,
I see again that you are in action, but unfortunately in
WRONG direction..hmm, please take compass next time..LOL

And what a pity also that (as usual),
You are NOT checking/NOT reading/NOT learning etc. Why ?
I already told you several days ago a few useful tips that
Before contacting Me: you should study all the lessons,
Which I suggested to you...please try again, just because
Hard, it doesn't mean to give up! Hope is a such thing that
Can help you to get through the difficult times!

Look, what I can tell you more,

I might not know that Eman is in training mode
Or Eman's experience MultiPV OFF is better...
Or about Eman's learning techniques etc.

But...what I know,

Many of Top engines are capable to learn
Properly... and even in parallel matches !!

What I know also that,
I relay and use Author's suggested settings!

Besides, what I know,
You are not Author of current Top chess engines,
So... I don't care much about what you say...LOL

As Final Words,
While we talk about TRAINING mode etc.
Listen well, I am on HOLIDAY mode..LOL

Hope helps...

Bye Bye )

[Sedat Canbaz]

Hello there,

Well, another new testings, but very likely
This is as last testings, next time I hope to
Run/Resume tours in September 2023 )

1st of all,
Almost same as recent test..e.g main diff.
RaptoRa 2.2 Eduard Vs SF-PB 060623

As main target and to be more clear,
To see again learning influence but via RapTora....
Plus to see Cutechess GUI's concurrent games: 10
That means, 10 games are played simultaneously..
Plus to see RapTora 2.2 + Eman exp's influence

Code: Select all

1st Test, Elo difference: 14 / DrawRatio: 70%
                                     
1   SF-PB 060623      +17/-13/=70 52.00%   52.0/100
2   RapTora 2.2 NoExp +13/-17/=70 48.00%   48.0/100

Note: No any learned exp used, but learning was on
----------------------------------------------------
2nd Test, Elo difference: 24 / DrawRatio: 71%
                    
1 RapTora 2.2, Eman.exp +18/-11/=71 53.50%  53.5/100
2 SF-PB 060623          +11/-18/=71 46.50%  46.5/100

Note: Eman pre-learned exp is used, plus learning on
----------------------------------------------------
3nd Test, Elo difference: 42 / DrawRatio: 78%
               
1 RapTora 2.2, Eman.exp +17/-5/=78 56.00%  56.0/100
2 SF-PB 060623          +5/-17/=78 44.00%  44.0/100

Note: Same Eman pre-learned exp is used, plus learning on
And it's clear that (in next re-run): RaptoRa is better!!

Conditions:
2x Epyc 7B12, CuteChess, 1 Core, Ponder OFF, 1m+0.6s, Unique, 128 MB Hash, 4-MEN

As Fınal Words,
RaptTora's learning is valid (via Concurrent games: 10)
In first test, SF-PB 060623 is managed to score better!
But in 2nd and especially in 3rd test, results changed!
E.g RapTora 2.2 Eduard + Eman exp is performed better..
No doubt, Eman exp is played as most important role!
And good news that RaptoRa is not so bad via Eman exp!
Note: Eman exp is same vers, which played in recent test..
A very small exp but quite effective (under these cond.)
Even I can say, e.g if we are going to compare all results:
RaptoRa 2.2 Eduard performed better than Eman 8.91 engine!

Thanks for your interest..

GAMES:
https://mega.nz/file/z1oUTQqI#Xqvk5tBO8 ... cbFoJa26_E

Greetings

[Anton101]

So, did you use the exp file from Eduard?

[Sedat Canbaz]

EDIT
Conditions:
Ryzen7 4800H, CuteChess, 1 Core, Ponder OFF, 1m+0.6s, 128 MB Hash, 4-MEN
Note: Unique v50 opening suite is used (played as twice)

[Sedat Canbaz]

So, did you use the exp file from Eduard?

No, I used RapTora 2.2 compiled by Eduard Nemeth
As Exp file is used the same Eman exp which played in recent test:
Eman 8.91 vs SF-PB 060623

In other words (via current test is simply that):
The target is to see Eman exp via Raptora engine

And as we noticed that,
Raptora engine + Eman exp are performed unbelievable so strong!
That means also that, in case of Eman Exp entries..I can test them
Via Raptora 2.2 engine, you know, for Me, the stability is most important part!
As you may aware, Enan 8.91 crashes via concurrent 64, where Raptora not...

Greetings

[Anton101]

Oh, ok!

I miss understood.

Great test!

When you come back from vacations, I'll send you a very updated RapTora with its own exp file.

Best regards.