Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: REBEL NUUE BLOG Tue Nov 14, 2023 11:52 am
As you possibly have already read the NNUE development of REBEL has come to a halt because my available hardware has become insufficient to make further progress in an acceptable time. My NNUE develop PC, has a Ryzen 12 cores, 24 threads plus a fast RTX 3080. That was good in the early days with Rebel 14 and 15 with small net sizes and few data but with REBEL-EAS using 28B data into a large net size the development time for a net increased from 2 days to 10-12 days and the fun slowly started to shrink.
After switching to Leela data I made quickly 20-25 elo progress but thereafter only 2 new ideas worked (see the good-data folder) and the fun reached a low point, spoiling 10-12 days too many times. That's no way of living. It's not something new, the stronger your engine becomes the harder it is to make further progress.
I still love NNUE programming, so much new to discover, so many new ideas to test, see the new-data folder. And so I setup a test environment of 4B data and small net size, test it with the new ideas. In this way testing a new idea usually takes only 2 days. And as my experience with NNUE programming has learned me, if new data works on 4B it will usually work adding the new data to the current best data of 26.5B as shown in the 2 improvements in the good-data folder.
So, there is this blog you can follow, usually updated once or twice a day with the latest results.
Subject: Re: REBEL NUUE BLOG Tue Nov 14, 2023 7:59 pm
nice improvement wie epoch 80 (compared with epoch 40), about 26 Elo ?
I of course hope that Rebel EAS will not be the final version, the results of the previous DEV version looked good (+30). I also hope that there will be a change of mindset with Chris, at this strong level it is very difficult to make progress.
I also can't imagine that there aren't ways to improve the current search. There seem to be numerous opportunities for improvement in the new programs (of which I am only aware of a fraction), which may not individually bring about major improvements in rating, but should have an impact as a whole. Without these very small advances, there would be no measurable progress at Stockfish.
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Tue Nov 14, 2023 11:44 pm
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Wed Nov 15, 2023 12:00 am
Dio wrote:
nice improvement wie epoch 80 (compared with epoch 40), about 26 Elo ?
Early epochs learn fast.
Dio wrote:
I of course hope that Rebel EAS will not be the final version, the results of the previous DEV version looked good (+30). I also hope that there will be a change of mindset with Chris, at this strong level it is very difficult to make progress.
Temporarily moving to another game can bring new ideas.
Dio wrote:
I also can't imagine that there aren't ways to improve the current search. There seem to be numerous opportunities for improvement in the new programs (of which I am only aware of a fraction), which may not individually bring about major improvements in rating, but should have an impact as a whole. Without these very small advances, there would be no measurable progress at Stockfish.
True.
Dio likes this post
texium
Posts : 112 Join date : 2022-07-19
Subject: Re: REBEL NUUE BLOG Thu Nov 16, 2023 8:40 am
what's different in the cstal data from the rebel eas that makes them so much different
texium
Posts : 112 Join date : 2022-07-19
Subject: Re: REBEL NUUE BLOG Thu Nov 16, 2023 8:42 am
i also think your method while "strengthening" rebel/cstal, has also weakened its conversion skills. if you test rebel eas vs rebel 16 i'd be curious how it does in already winning position, as if you look at ccrl it has a very high drawrate like leela
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Thu Nov 16, 2023 10:16 am
texium wrote:
what's different in the cstal data from the rebel eas that makes them so much different
Chris and I talk a lot, exchange ideas and then we go different paths. Meaning, hard to tell the differences.
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Thu Nov 16, 2023 10:26 am
texium wrote:
i also think your method while "strengthening" rebel/cstal, has also weakened its conversion skills. if you test rebel eas vs rebel 16 i'd be curious how it does in already winning position, as if you look at ccrl it has a very high drawrate like leela
Draw rates are high in balanced positions, like in CCRL testing, go to Stefan's new list, or my anti-draw pilot for less draws.
texium
Posts : 112 Join date : 2022-07-19
Subject: Re: REBEL NUUE BLOG Thu Nov 16, 2023 10:42 am
Admin wrote:
texium wrote:
i also think your method while "strengthening" rebel/cstal, has also weakened its conversion skills. if you test rebel eas vs rebel 16 i'd be curious how it does in already winning position, as if you look at ccrl it has a very high drawrate like leela
Draw rates are high in balanced positions, like in CCRL testing, go to Stefan's new list, or my anti-draw pilot for less draws.
Perhaps I worded it wrong and misinterpreted data. I think cstal has an issue with generating advantages, be that it's aggressive style, or something else, but since ccrl tests balanced rather than winning openings, higher draw rate would be more indicative of not getting advantages for it to convert which is why it does better in adrl. I notic d when playing against weaker engines with normal openings it might struggle to get a win
Nezhman likes this post
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Thu Nov 16, 2023 11:30 am
No progress measured feeding the net with 10% high mobility situations, it only will make a net somewhat more aggressive but that I already have done with EAS. Now I want elo
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Thu Nov 16, 2023 11:36 am
texium wrote:
Admin wrote:
texium wrote:
i also think your method while "strengthening" rebel/cstal, has also weakened its conversion skills. if you test rebel eas vs rebel 16 i'd be curious how it does in already winning position, as if you look at ccrl it has a very high drawrate like leela
Draw rates are high in balanced positions, like in CCRL testing, go to Stefan's new list, or my anti-draw pilot for less draws.
Perhaps I worded it wrong and misinterpreted data. I think cstal has an issue with generating advantages, be that it's aggressive style, or something else, but since ccrl tests balanced rather than winning openings, higher draw rate would be more indicative of not getting advantages for it to convert which is why it does better in adrl. I notic d when playing against weaker engines with normal openings it might struggle to get a win
It's hard to draw such a conclusion, current top engines hardly make mistakes, look at the number of SF losses, there are rare. Perhaps you could try the EAS version Chess System Tal.
Last edited by Admin on Thu Nov 16, 2023 7:52 pm; edited 1 time in total (Reason for editing : typo)
Nezhman likes this post
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Fri Nov 17, 2023 12:15 pm
Subject: Re: REBEL NUUE BLOG Fri Nov 17, 2023 6:42 pm
This is great, very unique approach by improving strength and style through training data. Curious if it's possible for you to find positions that top engines struggle to evaluate due to complexity and train on that
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Fri Nov 17, 2023 9:38 pm
texium wrote:
This is great, very unique approach by improving strength and style through training data. Curious if it's possible for you to find positions that top engines struggle to evaluate due to complexity and train on that
I don't even know the type of positions Rebel struggles with.
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Sat Nov 18, 2023 12:20 am
Finished Rook-Imbalance
Code:
Rook Imbalnce EPOCH PERC GAMES 40 46.9% 2100 60 47.2% 2100 80 49.4% 2100 100 50.8% 2100 126 50.4% 2100 MAIN 50.7% 2000
Testing Rook Imbalance, adding 400M positions to the 4B MAIN data. Conclusion: nice but not convincing, but good enough to create 2B of those, merge them into the 26.5B best data, start the trainer and go fishing for 10-12 days.
The days of fast progress are over, gathering is what's left.
Dio and Ghppn like this post
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Fri Nov 24, 2023 5:54 pm
I got some more small improvements than expected and while I promised myself -- no more training sessions of 10-12 days -- it's going to happen again
Adding New Raw Data [2] in total 5.9B positions to the main base of 26.4B positions, making the data in total 32.3B positions and 1.25Tb of data. Then a full shuffle of the data which takes about one day.
Then start the trainer with some new setting, for insiders increasing the start LR from 0.000875 to 0.000975 and a gamma of 0.994.
Dio, Eelco and Ghppn like this post
texium
Posts : 112 Join date : 2022-07-19
Subject: Re: REBEL NUUE BLOG Mon Nov 27, 2023 12:35 am
So I noticed the rebel site is up, do you plan to update it
Ghppn likes this post
Admin Admin
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Mon Nov 27, 2023 9:00 am
texium wrote:
So I noticed the rebel site is up, do you plan to update it
First epoch (70) looks good, score is not only higher (49.0%) but it also needed less data 21% whereas Rebel-Dev used 27% of the data.
Nevertheless with only 2000 games the error bar (sometimes) is without mercy and scores have a tendency to swing. But because we play so many games the end result will be pretty reliable with and error margin of -5 | +5 elo maximum and likely less.
Ghppn likes this post
texium
Posts : 112 Join date : 2022-07-19
Subject: Re: REBEL NUUE BLOG Wed Nov 29, 2023 8:58 am
Admin wrote:
texium wrote:
So I noticed the rebel site is up, do you plan to update it
[F12] - Tuning Piece Division, this is how our 4B test data looks like when we make an overview from the 4B data with NU [F8]. As you see some are not well synchronized, specially the high odd piece counts. With NU [F12] we create extra data to fill the gaps in order that the data is better synchronized. Note the data before and after.
Operation Edit the file balance-piece-count.txt and in millions enter the data that should be added for each of the low piece-counts. And good for ~10 elo, in this particular case.
Another use of NU [F12] is to set everything (for instance) on "1" in balance-piece-count.txt witch will create a new epd where every piece count has exactly 1.000.000 positions. In other words you can design any piece-count division data you wish, from midgame to endgame.
Posts : 2528 Join date : 2020-11-17 Location : Netherlands
Subject: Re: REBEL NUUE BLOG Mon Dec 04, 2023 1:52 pm
Some statistics of the data I am using.
Part-one - Pawnsless positions
Code:
Count Pawnless epd\l2-1.epd
Reading 1.305.000.000
Queen vs Queen : 2.742.316 Queen vs two Rooks : 156.644 Queen vs Rook : 329.140 Rook vs Rook : 28.463.072 KBNK mates : 401.139 Rook vs Minor : 4.500.619 Pawnless Total : 53.337.819 [4.09%]
Done...
This is 5% of my current (26.5B) best data and KR vs KR (28M) is laughable and in total 20x28=560M pointless positions that safely can (and will be deleted) and unnecessarily take space in the NN.
I wonder if databases of rating lists can beat my 4% of pawnless positions
Last edited by Admin on Mon Dec 04, 2023 8:51 pm; edited 1 time in total