TheSelfImprover, matejst, Damir Desevac and Prometheus like this post
Damir Desevac
Posts : 319 Join date : 2020-11-27 Age : 43 Location : Denmark
Subject: Re: Rebel 14 Wed Jan 12, 2022 4:24 pm
Thanks for the new Rebel Ed. If you find time please add MP support as I have 64 cores at my disposal, and would like to try your engine on multiple cores.
matejst likes this post
Mclane
Posts : 3009 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: Rebel 14 Wed Jan 12, 2022 8:07 pm
I am little irritated. This is a pimped fruit. The evaluation has been replaced with a nnue net. The net was generated out of Benjamin data. But why is it rebel14 ?!
It has a different evaluation. That was generated out of Benjamin data.
Imo its more a fruit-benjamin clone. I would think different if ed would implement nnue in pro deo or an old rebel. But this is not the case here, or ?!
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Wed Jan 12, 2022 8:07 pm
Damir Desevac wrote:
Thanks for the new Rebel Ed. If you find time please add MP support as I have 64 cores at my disposal, and would like to try your engine on multiple cores.
Maybe some day but I am not very interested to step into the arena again. Best would be to move the nnue stuff into a new engine. Toga IV comes to mind as the most easy job. But no hard promise.
Mclane
Posts : 3009 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: Rebel 14 Wed Jan 12, 2022 8:10 pm
I see that you experiment with neural nets. That what most programmers do in the moment.
Replacing the old human chess evaluation with neural net code.
And generating this net with your own engines.
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Wed Jan 12, 2022 8:13 pm
Mclane wrote:
I am little irritated. This is a pimped fruit. The evaluation has been replaced with a nnue net. The net was generated out of Benjamin data. But why is it rebel14 ?!
It has a different evaluation. That was generated out of Benjamin data.
Imo its more a fruit-benjamin clone. I would think different if ed would implement nnue in pro deo or an old rebel. But this is not the case here, or ?!
In some cases, I quite like irritating people who need to be irritated. -- Robert Smith
Mclane, TheSelfImprover and matejst like this post
Damir Desevac
Posts : 319 Join date : 2020-11-27 Age : 43 Location : Denmark
Subject: Re: Rebel 14 Thu Jan 13, 2022 5:02 pm
Admin wrote:
Damir Desevac wrote:
Thanks for the new Rebel Ed. If you find time please add MP support as I have 64 cores at my disposal, and would like to try your engine on multiple cores.
Maybe some day but I am not very interested to step into the arena again. Best would be to move the nnue stuff into a new engine. Toga IV comes to mind as the most easy job. But no hard promise.
Ed I have one small request. Can you add Threads options in GUI parameters, even if Rebel is just using 1 core. I have tried and run Rebel in Infinity GUI server, and engine there detects 0 CPU and stops responding after couple of moves. I think Infinity GUI can not detect number of Threads Rebel is using is because engine does not show a setting in the command window for threads. So engine is using 0 cores.. Some gui need the hash and cores in parameters so engine can function properly, if you can fix this would be great? Just by adding Threads to 1 will fix this.
Dio
Posts : 213 Join date : 2021-08-28
Subject: Re: Rebel 14 Thu Jan 13, 2022 8:58 pm
I am [J.B.] currently testing Rebel 14 for the 40/4 CEGT ratinglist together with Wolfgang Battig, the first results are looking very good, the Elo rating in CEGT is about 100 Elo lower than in the GRL-list. Thanks for this strong engine. https://cegt.forumieren.com/t1654-testing-rebel-14nn
Admin and Mclane like this post
Mclane
Posts : 3009 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: Rebel 14 Thu Jan 13, 2022 9:04 pm
After so many years a new rebel. Amazing.
TheSelfImprover likes this post
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Thu Jan 13, 2022 11:19 pm
Dio wrote:
I am [J.B.] currently testing Rebel 14 for the 40/4 CEGT ratinglist together with Wolfgang Battig, the first results are looking very good, the Elo rating in CEGT is about 100 Elo lower than in the GRL-list. Thanks for this strong engine. https://cegt.forumieren.com/t1654-testing-rebel-14nn
Boy, you are fast
The GRL is on a pause, still need my comps to test new NNUE magic
matejst
Posts : 612 Join date : 2020-11-26
Subject: Re: Rebel 14 Sat Jan 15, 2022 1:18 am
Ed, I have a few questions. Chris, Pawel also can answer, because it is a topic that I find deeply interesting, and any comment is welcome.
Does the search impact the evaluation at the same depth? I use both Rebel 14, the Rebel NN with Marvin [and Marvin with the previous Benjamin net]. But I often get different evaluations, let's say at depth 14, between Rebel 14 and Marvin_Rebel. What could cause it? Pruning? Extensions? An example, to make myself clear:
In this position, Rebel 14 proposes 11. Nd5, while Marvin_Rebel and Marvin_Benjamin (the older net) both proposes 11. Re1 (before ultimatively switching to other moves); at move 7, Rebel 14 chooses Nde2, again when Marvin_Rebel goes with Nb3. So I thought that it was mainly caused by the search. There also seems that there is a certain randomness in Marvin (non-deterministic engine?), because when I reanalized the same position it chose a different move (11.h3), while Rebel 14 sticked with 11. Nd5.
How big is the influence of the search on the final evaluation? I understand theoretically the effects (both the benefits and the dangers) of pruning, but how important is it? You spoke about the negative effects of pruning on Pro Deo's style. You mentioned something like "pruning knowledge". Could you explain a bit more?
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Sat Jan 15, 2022 6:52 am
Yes, search matters, sometimes even the size of the hash table. Marvin has a better search, if I were you I would use Marvin.
matejst
Posts : 612 Join date : 2020-11-26
Subject: Re: Rebel 14 Sat Jan 15, 2022 11:12 am
Yes, I use Marvin_Rebel most of the time, and also Marvin_Benjamin. I can compare the two nets, the style of play is the one I want (very active, aggressive, often spectacular), and it has multiPV.
But for me, it is now clear that the net IS the engine, and that Marvin_Rebel is much more Rebel than Marvin. And it is really a great engine for analysis, incredibly human. It reminds me the way the evaluation was developed before Texel tunning and similar automated ways of tuning.
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Sat Jan 15, 2022 7:07 pm
That's the point some people on Talkchess don't understand. Rebel 14 isn't Fruit, it's Benjamin, Marvin with Benjamin net isn't Marvin any longer. What Fruit and Marvin contribute is the elo that comes with its search. In theory I could add Benjamin NN to SF and SF is no SF any longer. I don't know what Dietrich did with Toga, if he trained a NN with Toga it is still Toga, if he used SF it is SF with the strength of the Toga search.
Mclane
Posts : 3009 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: Rebel 14 Sat Jan 15, 2022 8:39 pm
Ryzen 5 3600
Platz
Motor
Punkte
St
Lc
Dr
St
Lc
St
Lc
Dr
Lc
Lc
Dr
St
Et
St
Lc
St
Et
Ho
Ko
Ig
De
Ko
Wa
Ch
Wa
Se
De
De
Re
To
Eq
Se
De
Ma
Dr
Bo
Sm
Za
Cr
Pr
Be
Ro
Pr
Or
Co
De
Co
Co
De
K2
De
Ro
De
SO
S-B
01
Stockfish_13_win_x64_avx2 [t6]
179,5/212
· ·· ·
====
==1=
====
1===
====
====
==1=
===1
=1=1
====
==11
=011
111=
0=11
1=1=
=1=1
1=11
===1
1=1=
01=1
1111
1111
1111
=111
1111
1110
1111
1111
1=11
1111
1111
1111
0111
1111
1101
1111
1111
1111
1=11
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
17211,50
02
Lc0 0.28.0 dx12 radeon570 [j94-130]
173,5/212
====
· ·· ·
===1
====
=1==
====
11==
====
===1
0==1
=1==
1===
==1=
=1==
=1=0
=111
====
1=11
=111
1111
==11
==11
==11
1===
1111
1111
===1
1111
1111
1111
1=11
1==1
1=11
1=1=
111=
1111
11=1
1111
1=11
111=
1111
1111
11=1
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
16530,00
03
Dragon-2.5-64bit-avx2 [t6]
173,0/212
==0=
===0
· ·· ·
====
10==
===0
=1=1
1===
==0=
=11=
====
1=11
1==1
1=10
=00=
==10
11=1
==11
1==1
===1
1=1=
1111
0111
=11=
1111
1111
11=1
1011
11=1
1101
1111
1111
111=
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1101
1111
16255,50
04
Stockfish14_x64_avx2 [t6]
171,5/212
====
====
====
· ·· ·
====
====
==01
==00
====
0===
====
==01
111=
=1=1
1=1=
==11
1===
100=
==1=
1=11
==11
1111
1=1=
1111
1111
1111
1=11
1111
1111
1111
111=
0111
1111
1111
1=01
1111
1111
1111
1111
1111
1111
1111
1101
1111
1111
1111
1110
1111
1111
1111
1111
1111
1111
1111
16108,75
05
Lc0 0.27.0 dx12 radeon570 [j94-130]
171,0/212
0===
=0==
01==
====
· ·· ·
=1==
====
====
==0=
====
===0
1=11
11==
10=1
==11
01=1
=1==
=111
=1==
==1=
1==1
==11
1=1=
1=1=
1=1=
111=
1111
1==1
1111
1111
1111
1111
1111
111=
1111
1=11
1111
1111
=111
11=1
1=11
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
15940,00
06
Stockfish_12_x64 [t6]
171,0/212
====
====
===1
====
=0==
· ·· ·
====
=1=1
=11=
=10=
0===
==1=
====
=1=0
110=
====
==1=
1=01
===1
=1=1
0==1
1===
1=11
==11
11=1
=11=
1111
1111
1111
1=11
111=
1111
11=1
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
15815,25
07
Lc0 0.27.0 dx12 radeon570 [68286]
169,5/212
====
00==
=0=0
==10
====
====
· ·· ·
====
0===
===0
====
111=
=11=
1==1
1===
01==
=1=1
1111
==11
1=1=
=1=1
11=0
11==
11=1
1==1
1=11
111=
1111
1111
111=
==11
1111
1111
1111
1111
=111
=111
1111
1111
1==1
1111
1111
1111
1111
1111
1111
1111
=11=
1111
1111
1111
1111
1111
1111
15796,75
08
Dragon-2.5-64bit-avx2 [t6, mcts]
169,0/212
==0=
====
0===
==11
====
=0=0
====
· ·· ·
====
==10
0===
1===
==10
1=00
11==
=1==
11=1
==1=
1=1=
===1
11=1
1111
=11=
1111
1=10
1111
1111
1111
1111
1111
1111
=1=1
11=1
11=1
1111
1111
1111
111=
=111
1=11
1=11
1111
11=1
1111
1111
1=11
1111
1111
1111
1111
1111
1111
1111
1=11
15796,75
09
Lc0 0.27.0 dx12 radeon570 [j94-120]
168,0/212
===0
===0
==1=
====
==1=
=00=
1===
====
· ·· ·
===1
===1
=1==
1===
=01=
====
==1=
1=1=
1=1=
==1=
1==0
=11=
=1=1
=111
=1=1
1=11
1111
11=1
11=1
1111
1111
1=10
1111
1==1
1111
1111
1111
1111
111=
=111
1111
1111
1=11
1111
1111
1111
1111
1111
1=1=
1111
1111
1111
1=11
1111
1111
15714,75
10
Lc0 0.27.0 dx12 radeon570 [j94-100]
168,0/212
=0=0
1==0
=00=
1===
====
=01=
===1
==01
===0
· ·· ·
=0==
1==0
1=1=
=1==
==11
=110
1===
====
=11=
111=
==1=
1===
==11
11==
11=1
111=
111=
1111
1111
11=1
1111
11=1
111=
1==1
1111
1111
1111
=111
1111
1111
1111
1111
11=1
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
15463,50
11
Dragon-64bit-avx2 [t6]
164,0/212
====
=0==
====
====
===1
1===
====
1===
===0
=1==
· ·· ·
0=1=
=11=
0===
==1=
=01=
===1
==1=
=1==
11==
11==
=111
==11
=011
1111
1111
=110
1011
1111
=111
1=11
1111
1111
1111
0111
1111
1111
111=
1=11
1=1=
1111
11=1
1101
1111
1111
0111
1101
1111
111=
1111
1111
1111
1111
1111
15469,75
12
Stockfish_11_20011801_x64 [t6]
159,5/212
==00
0===
0=00
==10
0=00
==0=
000=
0===
=0==
0==1
1=0=
· ·· ·
0=1=
=0=1
11==
====
===0
==1=
==0=
==1=
=111
=11=
=111
11=1
1==1
011=
111=
1111
1111
11=1
1111
1111
1111
1111
1111
1011
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1110
1111
14103,25
13
Ethereal-13.00-pext-avx2 [t6, e13standard]
159,5/212
=100
==0=
0==0
000=
00==
====
=00=
==01
0===
0=0=
=00=
1=0=
· ·· ·
===1
==0=
01==
====
====
10==
1=1=
01==
1==1
=1=1
1111
1=11
11=1
111=
111=
1111
1111
1=11
1111
=111
1111
1111
1111
1111
1111
1111
1111
11=1
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
13939,50
14
Stockfish14_x64_avx2 [t6, no NNUE]
158,5/212
000=
=0==
0=01
=0=0
01=0
=0=1
0==0
0=11
=10=
=0==
1===
=1=0
===0
· ·· ·
=01=
00=0
0=0=
===1
====
=1==
=1=1
1==1
01==
1111
0111
=111
1111
111=
1110
1111
=111
1111
=11=
1111
11=1
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
13894,00
15
Lc0 0.27.0 dx12 radeon570 [703810]
157,5/212
1=00
=0=1
=11=
0=0=
==00
001=
0===
00==
====
==00
==0=
00==
==1=
=10=
· ·· ·
==1=
===1
===1
0===
101=
==10
====
1==1
11=1
1=0=
1=11
1=11
11==
1111
11=1
1111
1111
11=1
1111
1111
11=1
111=
1111
1111
1011
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
13920,25
16
Stockfish_13_win_x64_avx2 [t6, no NNUE]
157,0/212
0=0=
=000
==01
==00
10=0
====
10==
=0==
==0=
=001
=10=
====
10==
11=1
==0=
· ·· ·
01==
=1==
0=11
===1
1111
==01
=1=1
1===
0011
=111
1011
1111
111=
0=11
=111
1111
1111
1111
1011
1111
1111
1101
1111
111=
1011
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
13995,75
17
Ethereal-13.00-avx2 [t6, e13standard]
153,0/212
=0=0
====
00=0
0===
=0==
==0=
=0=0
00=0
0=0=
0===
===0
===1
====
1=1=
===0
10==
· ·· ·
===1
=0==
1===
0===
1===
1==1
=1==
====
=1=1
1111
1111
1=11
11=1
111=
111=
1=11
1111
1111
=111
1111
1=11
=111
=111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
13234,25
18
Houdini 6 [t6]
147,5/212
0=00
0=00
==00
011=
=000
0=10
0000
==0=
0=0=
====
==0=
==0=
====
===0
===0
=0==
===0
· ·· ·
==1=
=0=1
===1
=1==
=0=0
=1=1
11=1
1111
=111
1111
1111
===1
11==
1111
==11
1111
=111
1111
=111
1=11
1111
1==1
11==
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
12489,00
19
Komodo-14.1-64bit-bmi2 [t6]
144,0/212
===0
=000
0==0
==0=
=0==
===0
==00
0=0=
==0=
=00=
=0==
==1=
01==
====
1===
1=00
=1==
==0=
· ·· ·
===1
=0==
====
==0=
=101
1==1
=111
1101
=1==
0=11
0=11
=1=1
=111
11=1
111=
0=11
1111
111=
1111
1111
111=
1101
1111
1111
1111
1110
1111
1111
1111
1111
1111
1111
1111
1111
1111
12364,25
20
Igel-x64_popcnt_avx2_3_0_0 [t6]
141,5/212
0=0=
0000
===0
0=00
==0=
=0=0
0=0=
===0
0==1
000=
00==
==0=
0=0=
=0==
010=
===0
0===
=1=0
===0
· ·· ·
===1
==1=
==01
1===
=111
=11=
1011
111=
1=1=
111=
110=
1111
==11
11=1
1111
1111
=1==
=11=
1111
1=11
1111
111=
1111
1111
1111
1111
11=1
1111
1111
1111
1111
1111
1111
1111
11763,75
21
DeepHiarcs15 [t6]
141,5/212
10=0
==00
0=0=
==00
0==0
1==0
=0=0
00=0
=00=
==0=
00==
=000
10==
=0=0
==01
0000
1===
===0
=1==
===0
· ·· ·
0===
1===
1==1
1=1=
==1=
1===
11=1
==10
1101
1=1=
111=
1=11
1111
1111
1111
=101
1111
=111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
11667,25
22
Komodo-14.1-64bit-bmi2 [t6, mcts]
141,0/212
0000
==00
0000
0000
==00
0===
00=1
0000
=0=0
0===
=000
=00=
0==0
0==0
====
==10
0===
=0==
====
==0=
1===
· ·· ·
=1==
==11
11=1
111=
=10=
11=1
==11
1==1
1=11
1111
=111
1111
1111
1111
=11=
1111
11=1
1111
1111
1111
1111
1111
1111
1111
111=
1111
1111
1111
1111
1111
1111
1111
11376,25
23
Wasp520avx [t6]
136,5/212
0000
==00
1000
0=0=
0=0=
0=00
00==
=00=
=000
==00
==00
=000
=0=0
10==
0==0
=0=0
0==0
=1=1
==1=
==10
0===
=0==
· ·· ·
===1
1101
==01
111=
1011
1=11
111=
1111
1101
0===
=111
1111
111=
1111
11=1
=111
1111
1111
11=1
1111
1=11
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
10946,00
24
Chiron-5-64bit [t6]
126,5/212
0000
0===
=00=
0000
0=0=
==00
00=0
0000
=0=0
00==
=100
00=0
0000
0000
00=0
0===
=0==
=0=0
=010
0===
0==0
==00
===0
· ·· ·
=111
1==0
1===
=111
1=0=
==11
1===
=111
1=1=
1=11
1111
1111
111=
1111
111=
1101
=111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
9527,00
25
Wasp450-x64-modern [t6]
118,0/212
=000
0000
0000
0000
0=0=
00=0
0==0
0=01
0=00
00=0
0000
0==0
0=00
1000
0=1=
1100
====
00=0
0==0
=000
0=0=
00=0
0010
=000
· ·· ·
0==1
0===
01==
1=11
==10
1111
11=1
====
==11
=111
1111
1011
=11=
1=11
1111
111=
111=
1111
=111
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
8624,25
26
Seer_znver2.0.1 [t6]
112,5/212
0000
0000
0000
0000
000=
=00=
0=00
0000
0000
000=
0000
100=
00=0
=000
0=00
=000
=0=0
0000
=000
=00=
==0=
000=
==10
0==1
1==0
· ·· ·
0===
==11
1111
01=1
11==
1001
1=10
1=11
1=10
111=
1111
1110
1111
=110
1111
1111
11=1
1011
1111
1111
1111
1111
1111
1111
1111
1111
1111
1111
7782,50
27
DeepHiarcs14WCSC [t6 ]
105,5/212
0001
===0
00=0
0=00
0000
0000
000=
0000
00=0
000=
=001
000=
000=
0000
0=00
0100
0000
=000
0010
0100
0===
=01=
000=
0===
1===
1===
· ·· ·
10==
110=
0=01
=0=0
==1=
11=0
====
====
1=01
1=11
1110
1111
=111
11=1
=11=
1111
111=
1111
11=1
=111
=111
11=1
1111
=11=
1111
1111
1111
7567,50
28
Deep Rybka 4.1 x64 [t6]
103,5/212
0000
0000
0100
0000
0==0
0000
0000
0000
00=0
0000
0100
0000
000=
000=
00==
0000
0000
0000
=0==
000=
00=0
00=0
0100
=000
10==
==00
01==
· ·· ·
100=
=1==
1=00
==1=
1111
=111
1011
0==1
1011
1011
1111
11=1
1111
1=1=
1111
1111
1111
11=1
1111
1111
1111
1111
1111
1111
1111
1111
6694,75
29
Rebel-14-avx2 [pb]
102,5/212
0000
0000
00=0
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0001
0000
000=
0=00
0000
1=00
0=0=
==01
==00
0=00
0=1=
0=00
0000
001=
011=
· ·· ·
0111
1==1
01==
==1=
===1
1111
=11=
1=1=
1101
0111
=111
1111
1111
1111
=111
11=1
1111
1111
1111
1111
1111
1111
1111
1111
1111
6477,50
30
Toga3_0.3 net06 avx2 [t6]
100,0/212
0=00
0000
0010
0000
0000
0=00
000=
0000
0000
00=0
=000
00=0
0000
0000
00=0
1=00
00=0
===0
1=00
000=
0010
0==0
000=
==00
==01
10=0
1=10
=0==
1000
· ·· ·
10=0
=1=0
00=0
0111
====
1==1
11=0
1=11
011=
1=11
1111
1111
1111
1111
1111
1111
==11
1=11
1=11
1111
1111
1111
1111
1111
6676,25
31
Equinox 3.30-x64 [t6]
99,0/212
0000
0=00
0000
000=
0000
000=
==00
0000
0=01
0000
0=00
0000
0=00
=000
0000
=000
000=
00==
=0=0
001=
0=0=
0=00
0000
0===
0000
00==
=1=1
0=11
0==0
01=1
· ·· ·
=100
0===
====
1==1
===1
1111
=111
0===
1=11
1111
1111
1=11
=1==
1111
1111
=111
111=
1111
1111
1111
1111
1111
1111
6557,50
32
Senpai_20-x64 [t6]
95,0/212
0000
0==0
0000
1000
0000
0000
0000
=0=0
0000
00=0
0000
0000
0000
0000
0000
0000
000=
0000
=000
0000
000=
0000
0010
=000
00=0
0110
==0=
==0=
10==
=0=1
=011
· ·· ·
10=0
010=
=111
===1
10=1
=11=
=111
1111
=111
1111
1111
1111
111=
1111
1111
1111
1=11
1111
1111
1111
1111
1111
5726,00
33
DeepShredder13UCIx64 [t6]
89,5/212
0000
0=00
000=
0000
0000
00=0
0000
00=0
0==0
000=
0000
0000
=000
=00=
00=0
0000
0=00
==00
00=0
==00
0=00
=000
1===
0=0=
====
0=01
00=1
0000
==0=
11=1
1===
01=1
· ·· ·
111=
1=00
1=11
01=0
11=0
1110
0110
1100
1011
1101
1111
1111
0000
1001
1111
0011
1111
1111
1111
1111
1111
6211,75
34
Marvin5.1.0_x86_64_avx2 [t6]
86,5/212
1000
0=0=
0000
0000
000=
0000
0000
00=0
0000
0==0
0000
0000
0000
0000
0000
0000
0000
0000
000=
00=0
0000
0000
=000
0=00
==00
0=00
====
=000
===0
1000
====
101=
000=
· ·· ·
1111
==0=
1=10
==11
=1=1
1111
0=11
=11=
1101
111=
1111
1=11
1111
1111
111=
1111
1111
1111
1111
1111
5076,75
35
Drofa_3.1.0 [t6]
79,5/212
0000
000=
0000
0=10
0000
0000
0000
0000
0000
0000
1000
0000
0000
00=0
0000
0100
0000
=000
1=00
0000
0000
0000
0000
0000
=000
0=01
====
0100
0000
====
0==0
=000
0=11
0000
· ·· ·
0=1=
1=11
1010
1=01
=111
=111
11=0
1111
0=11
1111
11=1
1111
1=11
1111
1110
1=11
=111
1101
1111
4726,50
36
Booot6
78,0/212
0010
0000
0000
0000
0=00
0000
=000
0000
0000
0000
0000
0100
0000
0000
00=0
0000
=000
0000
0000
0000
0000
0000
000=
0000
0000
000=
0=10
1==0
=00=
0==0
===0
===0
0=00
==1=
1=0=
· ·· ·
==1=
=0=1
00=1
1==1
===1
==11
1111
11==
1111
1111
111=
11=1
1111
1111
1111
1111
1=11
=111
4350,50
37
SmarThink_v198_x64_standalone
77,0/212
0000
00=0
0000
0000
0000
0000
=000
0000
0000
0000
0000
0000
0000
0000
000=
0000
0000
=000
000=
=0==
=010
=00=
0000
000=
0100
0000
0=00
0100
0=0=
00=1
0000
01=0
10=1
0=01
0=00
==0=
· ·· ·
=01=
1101
101=
10==
==11
1111
111=
1011
1111
1111
1111
=111
1101
1111
1111
1111
1111
4317,50
38
Zappa Mexico II x64 [t6]
73,5/212
0000
0000
0000
0000
0000
0000
0000
000=
000=
=000
000=
0000
0000
0000
0000
0010
0=00
0=00
0000
=00=
0000
0000
00=0
0000
=00=
0001
0001
0100
0010
0=00
=000
=00=
00=1
==00
0101
=1=0
=10=
· ·· ·
01=1
1=11
=101
1110
1=01
11=1
1=10
1111
1111
0111
11=1
1=11
1111
1111
11=1
0111
4169,50
39
Crafty-25-64-uci-ja [t6]
66,0/212
0000
0=00
0000
0000
=000
0000
0000
=000
=000
0000
0=00
0000
0000
0000
0000
0000
=000
0000
0000
0000
=000
00=0
=000
000=
0=00
0000
0000
0000
1000
100=
1===
=000
0001
=0=0
0=10
11=0
0010
10=0
· ·· ·
110=
1010
1111
==1=
101=
=111
1100
=111
11=0
1=01
1101
1111
1111
1111
1111
3610,75
40
Prodeo3.1
62,5/212
0=00
000=
0000
0000
00=0
0000
0==0
0=00
0000
0000
0=0=
0000
0000
0000
0100
000=
=000
0==0
000=
0=00
0000
0000
0000
0010
0000
=001
=000
00=0
=000
0=00
0=00
0000
1001
0000
=000
0==0
010=
0=00
001=
· ·· ·
==0=
==1=
==0=
1==1
=10=
1111
1111
=101
1===
1111
1111
0111
1111
1111
3675,00
41
Benjamin 1.1 [pb]
59,5/212
0000
0000
0000
0000
0=00
0000
0000
0=00
0000
0000
0000
0000
00=0
0000
0000
0100
0000
00==
0010
0000
0000
0000
0000
=000
000=
0000
00=0
0000
0000
0000
0000
=000
0011
1=00
=000
===0
01==
=010
0101
==1=
· ·· ·
10=0
===0
=1=1
1=0=
=111
1==1
1011
1011
1111
1111
1101
1111
1111
3064,00
42
RodentII_x64
57,5/212
0000
0000
0000
0000
0000
0000
0000
0000
0=00
0000
00=0
0000
0000
0000
0000
0000
0000
0000
0000
000=
0000
0000
00=0
0000
000=
0000
=00=
0=0=
0000
0000
0000
0000
0100
=00=
00=1
==00
==00
0001
0000
==0=
01=1
· ·· ·
1111
===1
0111
11=1
1111
11=1
11=1
1=1=
1111
1111
1111
1=11
2573,75
43
Prodeo YAT
47,5/212
0000
00=0
0000
0010
0000
0000
0000
00=0
0000
00=0
0010
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
00=0
0000
0000
0000
0000
0=00
0000
0010
0010
0000
0000
0000
0=10
==0=
==1=
===1
0000
· ·· ·
==10
1010
1=1=
1110
1=11
=101
1110
1111
1111
1111
0=1=
2260,25
44
Orion64-v0.7
44,5/212
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0=00
0000
=000
0100
000=
0000
=000
0000
=0==
0000
0000
000=
1=00
00==
000=
00=0
010=
0==0
=0=0
===0
==01
· ·· ·
1100
=00=
=11=
==1=
1===
1=1=
=1=1
1111
1111
1111
1891,75
45
Colossus2021b [pb]
44,5/212
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0001
0000
0000
0000
0000
0000
0000
0000
0000
0000
00=0
0000
0000
000=
0000
0000
0000
0000
0100
0=01
=000
=01=
0=1=
1000
0101
0011
· ·· ·
1=01
===1
1011
1==1
1111
1111
1=11
1111
1111
1645,25
46
DeepGandalf7-64 [t2]
37,5/212
0000
0000
0000
0000
0000
0000
0000
0=00
0000
0000
1000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
00=0
00=0
0000
0000
0000
0000
1111
0=00
00=0
0000
0000
0000
0011
0000
=000
00=0
0=0=
=11=
0=10
· ·· ·
1000
00=0
1111
0=01
1111
11==
1111
11=1
1652,75
47
Colossus2021a [pb]
36,5/212
0000
0000
0000
0001
0000
0000
0000
0000
0000
0000
0010
0000
0000
0000
0000
0000
0000
0000
0000
00=0
0000
000=
0000
0000
0000
0000
=000
0000
0000
==00
=000
0000
0110
0000
0000
000=
0000
0000
=000
0000
0==0
0000
0001
=00=
===0
0111
· ·· ·
=1==
0100
11==
11=1
1111
1011
1111
1651,25
48
Colossus2008b
34,5/212
0000
0000
0000
0000
0000
0000
=00=
0000
0=0=
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
=000
0000
0000
0=00
000=
0000
0000
0000
0=00
00=0
0000
1000
00=1
=010
0100
00=0
0=00
==0=
0100
11=1
=0==
· ·· ·
=1=0
=010
=111
111=
0=11
=111
1524,25
49
Delfi5.0
32,5/212
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
000=
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
00=0
0000
0000
0=00
0000
0=00
1100
000=
0000
0000
=000
00=0
0=10
0===
0100
00=0
=010
0===
0==0
0000
1011
=0=1
· ·· ·
111=
1011
=1=0
0001
1111
1391,00
50
K2_091 [pb]
24,5/212
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0001
0000
0010
0=00
0010
0000
0000
0=0=
0001
0=0=
0000
1=10
00==
=101
000=
· ·· ·
0000
1111
1111
111=
794,00
51
DeepPatzer 3.80 [t2]
15,5/212
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
=00=
0000
0000
0000
0000
0000
0000
0000
0=00
0000
0000
0000
0000
0000
0000
0000
0000
=0=0
0000
0000
00=0
=000
0100
1111
· ·· ·
010=
0==1
=111
444,00
52
RomiChess64P3n
14,5/212
0000
0000
0000
0000
0000
0000
0000
0000
0=00
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
=000
0000
0000
0000
0000
1000
0010
0000
0000
0000
0=00
00==
0000
000=
=0=1
0000
101=
· ·· ·
=011
=011
489,00
53
DeepPatzer3.61 [t2]
14,0/212
0000
0000
0010
0000
0000
0000
0000
0000
0000
0000
0000
0001
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0010
0=00
0000
00=0
0000
0000
0000
0000
0000
0000
0000
0000
0100
1=00
1110
0000
1==0
=100
· ·· ·
0001
737,25
54
SOS 5.1 for Arena
11,0/212
0000
0000
0000
0000
0000
0000
0000
0=00
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
0000
=000
0000
1000
0000
0000
0000
0=00
1=0=
0000
0000
00=0
0000
=000
0000
000=
=000
=100
1110
· ·· ·
440,50
5724 Partien gespielt / Turnier beendet
Beginn des Turniers: 2021.01.04, 14:39:31 Letzte Aktualisierung: 2022.01.15, 17:33:01 Ort/ Land: ORION8, Deutschland Spielstufe: Turnier 40/5 Hardware: AMD Ryzen 5 3600 mit AMD RX570 8 GB Betriebssystem: Windows 10 Home Home Edition (Build 9200) PGN-Datei:Ryzen 5 3600.pgn Tabelle erstellt mit:Arena 2.0.1
Admin likes this post
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Sat Jan 15, 2022 9:16 pm
Not bad at all.
Mclane likes this post
matejst
Posts : 612 Join date : 2020-11-26
Subject: Re: Rebel 14 Sat Jan 15, 2022 9:20 pm
Ed, Toga III has a net trained on Toga's data. Dark Toga is Toga's search + Dietrich's net The White Rose, trained on Bad Gyal data. Dietriech created several interesting nets before he started working for the Komodo team.
To add to your point, I asked in the thread about analysing with an engine how we should call these "hybrids". The point is that if you call Fruit with Rebel's net "Rebel", then you have to rename Sf 13 and 14 "Leela cpu", since the net is based on Lc0 data.
matejst
Posts : 612 Join date : 2020-11-26
Subject: Re: Rebel 14 Sat Jan 15, 2022 9:28 pm
The reaction had little to nothing to do with the commercial brand "Rebel", but with SF and its cult.
Mclane
Posts : 3009 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: Rebel 14 Sun Jan 16, 2022 9:08 am
Admin wrote:
Not bad at all.
Especially since the others use 6 threads while rebel14 only 1.
Admin and matejst like this post
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Sun Jan 16, 2022 5:12 pm
Dio wrote:
I am [J.B.] currently testing Rebel 14 for the 40/4 CEGT ratinglist together with Wolfgang Battig, the first results are looking very good, the Elo rating in CEGT is about 100 Elo lower than in the GRL-list. Thanks for this strong engine. https://cegt.forumieren.com/t1654-testing-rebel-14nn
The 1700 games show an almost 400 elo increase over its HCE version Benjamin 1.1
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Sun Jan 16, 2022 5:13 pm
Mclane wrote:
Admin wrote:
Not bad at all.
Especially since the others use 6 threads while rebel14 only 1.
Ah, hence all these [t6].
And as a sort of compensation you gave Rebel 14 permanent brain access ?
Chris Whittington
Posts : 1254 Join date : 2020-11-17 Location : France
Subject: Re: Rebel 14 Sun Jan 16, 2022 5:23 pm
Admin wrote:
Dio wrote:
I am [J.B.] currently testing Rebel 14 for the 40/4 CEGT ratinglist together with Wolfgang Battig, the first results are looking very good, the Elo rating in CEGT is about 100 Elo lower than in the GRL-list. Thanks for this strong engine. https://cegt.forumieren.com/t1654-testing-rebel-14nn
The 1700 games show an almost 400 elo increase over its HCE version Benjamin 1.1
I remember your reaction when you first got this to work. Shock and awe.
Mclane and TheSelfImprover like this post
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands
Subject: Re: Rebel 14 Sun Jan 16, 2022 6:25 pm
Chris Whittington wrote:
Admin wrote:
Dio wrote:
I am [J.B.] currently testing Rebel 14 for the 40/4 CEGT ratinglist together with Wolfgang Battig, the first results are looking very good, the Elo rating in CEGT is about 100 Elo lower than in the GRL-list. Thanks for this strong engine. https://cegt.forumieren.com/t1654-testing-rebel-14nn
The 1700 games show an almost 400 elo increase over its HCE version Benjamin 1.1
I remember your reaction when you first got this to work. Shock and awe.
Indeed.
And ~400 is a nice after shock.
Thought these kind of things only happen in Holywood movies.
Mclane
Posts : 3009 Join date : 2020-11-17 Age : 57 Location : United States of Europe, Germany, Ruhr area
Subject: Re: Rebel 14 Sun Jan 16, 2022 6:36 pm
Admin wrote:
Mclane wrote:
Admin wrote:
Not bad at all.
Especially since the others use 6 threads while rebel14 only 1.
Ah, hence all these [t6].
And as a sort of compensation you gave Rebel 14 permanent brain access ?
No all engines use pb. I had to switch pb on in the style, otherwise it was off by default.
Admin Admin
Posts : 2536 Join date : 2020-11-17 Location : Netherlands