Andscacs nnue 0.1

[Dann Corbit]

Of course, my comments too were tongue in cheek

Naturally, I do not believe that SF has done something wrong (I think it is pretty clear, but I state the obvious because there are people who are immune to smileys)

The point I was trying to make is that SF borrowed something and it was ok. (and quite frankly, it really and truly was OK)

But if someone else tries to do the same thing, it is looked upon with shame and suspicion.

Is there, or is there not, something wrong with this picture?

Math belongs to everyone. Algorithms belong to everyone. Data tables belong to everyone.
Except in computer chess.

[Guenther]

Of course, my comments too were tongue in cheek

Naturally, I do not believe that SF has done something wrong (I think it is pretty clear, but I state the obvious because there are people who are immune to smileys)

The point I was trying to make is that SF borrowed something and it was ok. (and quite frankly, it really and truly was OK)

But if someone else tries to do the same thing, it is looked upon with shame and suspicion.

Is there, or is there not, something wrong with this picture?

Math belongs to everyone. Algorithms belong to everyone. Data tables belong to everyone.
Except in computer chess.

yawn...
SF did not ask for borrowing, actually it was gifted back for borrowing a lot from SF.

[Dann Corbit]

Of course, my comments too were tongue in cheek

Naturally, I do not believe that SF has done something wrong (I think it is pretty clear, but I state the obvious because there are people who are immune to smileys)

The point I was trying to make is that SF borrowed something and it was ok. (and quite frankly, it really and truly was OK)

But if someone else tries to do the same thing, it is looked upon with shame and suspicion.

Is there, or is there not, something wrong with this picture?

Math belongs to everyone. Algorithms belong to everyone. Data tables belong to everyone.
Except in computer chess.

yawn...
SF did not ask for borrowing, actually it was gifted back for borrowing a lot from SF.

Thank goodness that this math which belongs to the creators of that math was gifted to SF because otherwise they would be dirty rotten cheaters like other people who use that math.

The computer chess community is a strange flock of birds. All of them invented the feather.

[Frank Quisinsky]

The computer chess community is a strange flock of birds.

I am one of the birds!

Originality is most important but modern copyrights produced a really big, new and own problem!

Over many years chess programmers are working!
A new programmer added 100% of ideas, comes from others, in his new engine.
The programmer add the engine on github with a copyright.

That's the reason why ...
The community of freeware have a big problem with the community of open source.
Internet is full from such material. Freeware supporter vs. Open-Source supporter!

Open-Source Copyright's are _often_ the "epitome from a clone" ... very measured!
Sinice many years the copyrights itself build the software virus!
The causal-chain completly forgotten!!

I have the opinion that the idea from open-source is good-will but what the "World" do with the good-will idea is a shame.
Today not to repair, it is to late!

I have all the time problems in question, what can I add for engines for a rating list or a tournament. Game-statistics helps me in the past. Today not possible because Elo is to high. I am to end with better stats ideas. Possible, but to many time needed and time can be better use.

The decision for Engine x or y "no" or "yes" produced anytime a bad situation and discussion in the community!
"Birds", ... I am one of the birds!
Each persons with a clear opinion to topic x or y is a bird!!!
For each situation in our life we can build more as one opinion ... pro and contra, more times!

The secret is to try to understand people with an other opinion.
And here, a lot of people have a big problem.

This forum is full of birds!

Best
Frank

[Frank Quisinsky]

To many gods for a small group of "believers".

Nobody claims the wisdom to be eaten with spoons.

No, no ...
I have the feeling that to many programmers claims the wisdom to be eaten with spoons.

And this produced:
_If_ in programming ...
To many gods for a small group of "believers".

And all this for a chicken feed of Elo.

[AndrewGrant]

I wish someone created a very detailed tutorial on how to design/implement/train NNUE from scratch -
unfortunately there's a abyss gap on this question - those smart enough just doing their own networks
while noobs are doomed to HCE. For me this is discrimination.
I would've definitely create a youtube tutorial on NNUE from scratch if only I was smart enough(((

All of the source code is out there. I'll happily walk you through all of it at an hourly rate, starting from the basics of an NN and its backprop, to building the system to train the NNs by hand with C, to building datasets using Python and Cutechess, to executing those NNs in practice inside the engine.

Here is some extra knowledge you might not have already: https://chess.stackexchange.com/questio ... 3000_33736

[maksimKorzh]

I wish someone created a very detailed tutorial on how to design/implement/train NNUE from scratch -
unfortunately there's a abyss gap on this question - those smart enough just doing their own networks
while noobs are doomed to HCE. For me this is discrimination.
I would've definitely create a youtube tutorial on NNUE from scratch if only I was smart enough(((

All of the source code is out there. I'll happily walk you through all of it at an hourly rate, starting from the basics of an NN and its backprop, to building the system to train the NNs by hand with C, to building datasets using Python and Cutechess, to executing those NNs in practice inside the engine.

Here is some extra knowledge you might not have already: https://chess.stackexchange.com/questio ... 3000_33736

Hi Andy

The link in your post gave me more understanding than all NNUE threads I've been reading through.
Finally I know what HalfKP is.
Why? Because it explains the WHOLE PIPELINE instead of focusing on details.
Details start making sense only when the entire pipeline is clear.
I just can't understand why didn't you share it with all in a single thread??
Or you did and I missed that?

re: hourly rate lessons
- the problem is not in a lack mentoring, the gap was due to the lack of understanding the basic pipeline.
Now when you've finally explained it I can work on my own. And just to give an idea why mentoring sucks im my case -
here's a generalized sketch from my life (when I tried to learn via being mentored):

Mentor (after explained the subject in truly great details, did his best): did you understand?
Me: NO
Mentor: ... you're idiot... I don't know how to teach you...

The problem is not in teaching, it's in learning. It's a student's effort to learn. He has to do it on his own.
So I just don't want to waste your time and my money, because of this scenario.

But when someone provides me the whole pipeline in explanations or pseudo code or whatever - I
suddenly start understanding. For instance when Ronald Friederich did explain Texel's tuning to me (and your optimizations btw)
I just sat down and implemented it:
https://github.com/maksimKorzh/wukongJS ... eval_tuner
I know it:
1. Sucks
2. Slow as hell
3. Gives a very miserable improvement (despite the fact you're criticizing me for the lack of basic knowledge in statistics )
But it's a proof-of-concept implementation that works.
I've even written a "paper"-like doc for my own self breaking down the pipeline into parts for better understanding.
https://github.com/maksimKorzh/wukongJS ... _TUNING.MD

Now you might say (well at least think) that it's a total pointless and useless piece of crap and most likely it is,
but you know what it has changed and what opportunities it gave to me???

I can now read through your eval tuning PDF and it suddenly makes perfect sense to me because now
I have SOMETHING to optimize and make it better.

I know you don't care, but you just try to understand it from noob's position - the difference between absolute
dead point stuck and moving off that dead point. What I did is moving off the dead point. If you could ever
imagine how happy I was when realized the principle of turning scalar value to probabilistic via sigmoid???
And then subtract the result from actual game outcome and then sum the error and finally get mean square error, oh man...
That was such a rocket science but now I clearly understand it.

Same with NNUE - as soon as I "feel" the entire pipeline I'll be able to make my own.
Most likely it would be like it's done in Tomitank chess - to "compensate" eval and not to replace it.
Tamas's coding style is much easier to learn from compared to top engine authors, he also incorporates
all the weights into the JS source which is awesome, so bearing in mind that my current engine is in JS as well
I have literally all I need.

Absolute strength of an engine means nothing to me, only proof of concept implementations that I can later
share with my youtube subscribers.

You know what's the most fun of it? It's when your subscriber taking your code and then he writes to you:
"man, I tried to implement movegen many time before but couldn't now with your explanations I finally did it"
And he made damn 10x faster movegen using exactly the same algorithms I did but not using objects (they are slow as hell in JS as it turned out).
It's incredible feeling when you see like someone uses your proof-of-concept work and then optimizes it so eventually it
becomes much better than your initial work. And I'm doing that for free btw. Videos are for all.

Ok, it was a bit too much.
Just thank you for sharing that stack exchange answer, it was very insightful.

[cdani]

Another disappointing aspect of this release is that it's in blatant violation of the GPL. Andscacs, as a closed source engine, cannot rely on the GPL'd NNUE code it copies unless the entirety of the source code is also distributed to the users along with the binaries...

Sorry!
Was not my intention.
I removed the zip file.
Best wishes.

[cdani]

Hi,

I am sure Daniel José Queraltó will read the thread ...

Daniel, which engines you like most?
I am sure we can learn a bit if you told us your opinion?

Or better ...
From which programs and ideas you learn most?

Best
Frank

I'm too practical. I like the engines that help me obtain whatever is my current goal.

From the point of view of programming, I learnt a lot of Gull, Stockfish, Talkchess and many others.
From the point of view of playing, mostly from OTB chess, Icc, Lichess and Chessbase.
From the point of view of philosophy, understanding the big picture of chess is like trying to understand to which point finding order int the chaos makes sense; if there is some intrinsic order farther than our mental creations.

[JohnW]

Here the Windows executable and latest Stockfish nnue net:
https://1drv.ms/u/s!Aj8wlSmr_Hq3islynT1 ... Q?e=FVH5CN

Is the one drive link broken? Where can I get the Windows executable now?