Honey XI is now available for download.
The package includes 10 different flavors of engines, those marked with FD also have Fortress Detection code by Joe Ellis.
This included version of Bluefish includes code by Eelco de Groot and is not related to the high threaded Bluefish that shows up occasionally at broadcasted chess matches.
https://github.com/MichaelB7/Stockfish/releases
Scroll to the bottom for all flavors of binaries.
With Linux officially dropping support for 32 bit software in their next major release due in April of 2020, I have not provided any 32 bit Linux exe's
The Pi exe's are provided by Allan Cooper and Lucas Monge provides the 32 bit exe's for Windows, everything else by yours truly.
Honey XI
Moderators: hgm, Rebel, chrisw
-
- Posts: 5228
- Joined: Thu Mar 09, 2006 9:40 am
- Full name: Vincent Lejeune
Re: Honey XI
Thanks for all these versions, mike !
Vincent
Vincent
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: Honey XI
It is my pleasure of course to share.
What follows below is for anyone that may find it useful.
I often get questions on analysis. First, I respect the fact that everyone may have their own method , and I am not saying my way is the best way or only way.
First ,we know every engine has its strengths and weaknesses -so it's best to use multiple engines that are dissimilar enough. I would use three engines:
1. Stockfish or Honey ( but not both as they are too similar)
2. Matefinder or Black-Diamond ( again, pick one or the other as they are too similar)
3. Anyone of the Strong Neural Nets - at the moment I am using a Sergio net as they appear to be quite strong at LTC ( and it's free)
1. The goal is to get a consensus with all 3 engines - may not always happen. One the two A/B engines disagree, will force them to think for while using the other engine's best move , then go back to the original position and see if one of them change it best move to the engine.
2. Pay attention to the evaluation from the NN engine, if the NN engine is scored noticeably higher than the AB engines , that is definitely a move one would want to search more more deeply with the AB engines. It usually indicates that there is a strong positional advantage with the move that one of the AB engines may eventually find if left to search or may not find at all. Search that move in the AB engines, more often than not, the NN is more correct on strong positional moves that may involve a small sacrifice.
How long to search -
It varies by position , some positions are simply dead - neither side can do anything. I do not spend a lot of time on those positions.
Other positions are very sharp and can go either way, you would want spend at least 15 or 20 minutes, and sometime that is not enough, you may have to do an overnight analysis to see a breakthrough.
-
- Posts: 3186
- Joined: Sat Feb 16, 2008 7:38 am
- Full name: Peter Martan
-
- Posts: 582
- Joined: Wed May 10, 2006 7:28 pm
- Location: Birmingham, England
Re: Honey XI
Great work Mike!
Indeed, what does Defensive do? And what happened to Honey's Brute Force option?
Marek Soszynski
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: Honey XI
fewer losses when playing against equal or stronger opponentsMarek Soszynski wrote: ↑Thu Feb 13, 2020 8:25 amGreat work Mike!
Indeed, what does Defensive do? And what happened to Honey's Brute Force option?
Code: Select all
MichaelB7@VM-894787 MINGW64 ~/home/Github/bayeselo/src (master)
$ bayeselo
version 0058, Copyright (C) 1997-2016 Remi Coulom and updated by Michael Byrne.
compiled Feb 13 2020 09:50:25.
This program comes with ABSOLUTELY NO WARRANTY.
This is free software, and you are welcome to redistribute it
under the terms and conditions of the GNU General Public License.
See http://www.gnu.org/copyleft/gpl.html for details.
ResultSet>rp C:\cluster.mfb\pgn\Big6.pgn
1500 game(s) loaded
ResultSet>elo
ResultSet-EloRating>mm 0 1
00:00:00,00
ResultSet-EloRating>r
Rank Name Rating Δ + - # Σ Σ% W L D W% =% OppR
---------------------------------------------------------------------------------------------------------
1 Honey-XI 3146 0.0 15 15 500 292.0 58.4 115 31 354 23.0 70.8 3091
2 Stockfish-XI 3141 5.2 16 16 500 286.0 57.2 121 49 330 24.2 66.0 3092
3 Blue-Honey-XI 3123 17.9 16 16 500 270.5 54.1 96 55 349 19.2 69.8 3095
4 Bluefish-XI 3104 18.4 16 16 500 254.0 50.8 85 77 338 17.0 67.6 3099
5 Black-Diamond-XI 3076 28.0 16 16 500 228.0 45.6 63 107 330 12.6 66.0 3105
6 komodo-13.2.5-64bit 3010 66.0 15 15 500 169.5 33.9 37 198 265 7.4 53.0 3118
---------------------------------------------------------------------------------------------------------
Δ = delta from the next higher rated opponent
# = number of games played
Σ = total score, 1 point for win, 1/2 point for draw
ResultSet-EloRating>los
Ho St Bl Bl Bl ko
Honey-XI 69 98 99 99100
Stockfish-XI 30 95 99 99100
Blue-Honey-XI 1 4 96 99100
Bluefish-XI 0 0 3 99100
Black-Diamond-XI 0 0 0 0 99
komodo-13.2.5-64bit 0 0 0 0 0
ResultSet-EloRating>d
1 Honey-XI 3146 500.0 ( 292.0 : 208.0)
100.0 ( 50.5 : 49.5) Stockfish-XI 3141
100.0 ( 54.5 : 45.5) Blue-Honey-XI 3123
100.0 ( 54.0 : 46.0) Bluefish-XI 3104
100.0 ( 61.0 : 39.0) Black-Diamond-XI 3076
100.0 ( 72.0 : 28.0) komodo-13.2.5-64bit 3010
2 Stockfish-XI 3141 500.0 ( 286.0 : 214.0)
100.0 ( 49.5 : 50.5) Honey-XI 3146
100.0 ( 50.0 : 50.0) Blue-Honey-XI 3123
100.0 ( 56.5 : 43.5) Bluefish-XI 3104
100.0 ( 59.5 : 40.5) Black-Diamond-XI 3076
100.0 ( 70.5 : 29.5) komodo-13.2.5-64bit 3010
3 Blue-Honey-XI 3123 500.0 ( 270.5 : 229.5)
100.0 ( 45.5 : 54.5) Honey-XI 3146
100.0 ( 50.0 : 50.0) Stockfish-XI 3141
100.0 ( 50.0 : 50.0) Bluefish-XI 3104
100.0 ( 57.5 : 42.5) Black-Diamond-XI 3076
100.0 ( 67.5 : 32.5) komodo-13.2.5-64bit 3010
4 Bluefish-XI 3104 500.0 ( 254.0 : 246.0)
100.0 ( 46.0 : 54.0) Honey-XI 3146
100.0 ( 43.5 : 56.5) Stockfish-XI 3141
100.0 ( 50.0 : 50.0) Blue-Honey-XI 3123
100.0 ( 54.5 : 45.5) Black-Diamond-XI 3076
100.0 ( 60.0 : 40.0) komodo-13.2.5-64bit 3010
5 Black-Diamond-XI 3076 500.0 ( 228.0 : 272.0)
100.0 ( 39.0 : 61.0) Honey-XI 3146
100.0 ( 40.5 : 59.5) Stockfish-XI 3141
100.0 ( 42.5 : 57.5) Blue-Honey-XI 3123
100.0 ( 45.5 : 54.5) Bluefish-XI 3104
100.0 ( 60.5 : 39.5) komodo-13.2.5-64bit 3010
6 komodo-13.2.5-64bit 3010 500.0 ( 169.5 : 330.5)
100.0 ( 28.0 : 72.0) Honey-XI 3146
100.0 ( 29.5 : 70.5) Stockfish-XI 3141
100.0 ( 32.5 : 67.5) Blue-Honey-XI 3123
100.0 ( 40.0 : 60.0) Bluefish-XI 3104
100.0 ( 39.5 : 60.5) Black-Diamond-XI 3076
ResultSet-EloRating>
so if you like that type of output above for your games , feel free "rp" is short for readpgn, one of several keyboard shortcuts I added
commands:
rp pgnfile
elo
mm 0 1
###then you can do one or more of these commands
r ( shortcut for rating)
los ( likelihood of superiority)
d (shortcut details)
link: https://github.com/MichaelB7/bayeselo/b ... yeselo.exe
no dll required -which is why it is so huge
Also, the Weakfish engine is a bruteforce engine - so that replaced the option
-
- Posts: 385
- Joined: Sat Feb 04, 2017 11:57 pm
- Location: USA
Re: Honey XI
Did you apply the SF11 mate score problem?
it fixes #2533, #2543, #2423.
it fixes #2533, #2543, #2423.
i7-6700K @ 4.00Ghz 32Gb, Win 10 Home, EGTBs on PCI SSD
Benchmark: Stockfish15.1 NNUE x64 bmi2 (nps): 1277K
Benchmark: Stockfish15.1 NNUE x64 bmi2 (nps): 1277K
-
- Posts: 4889
- Joined: Thu Mar 09, 2006 6:34 am
- Location: Pen Argyl, Pennsylvania
Re: Honey XI
Not yet , that just got pushed. Will do so and issue an updated bleeding edge binary when I get back home ima fee days. I had noticed that issue as well, doesn’t really affect game play , but it’s very annoying from a user perspective. Hence the need to update the bleeding edge binaries . Look for it by Wednesday.
-
- Posts: 3
- Joined: Mon Mar 02, 2020 4:05 pm
- Full name: Tobias Mueller
Re: Honey XI
Hi, first of all thank you very much Mike and all the others involved in the project for providing uns with the Honey engine! I am not sure if someone has already done it, but I have managed to compile binaries for Android and so far they seem to work well in DroidFish and an old version of AcidApeChess on Android 8.0.0. In case someone is interested to test them I can provide the binaries or instructions how to compile (some changes in the Makefile and also minor changes in the code was needed).
-
- Posts: 20
- Joined: Sat Aug 31, 2019 10:38 pm
- Full name: Mariusz Jankowski
Re: Honey XI
Please share binaries for Android. I have been waiting for this for a long time.