Page 1 of 3

semi-monumental achievement

Posted: Tue Jul 26, 2022 10:26 am
by deeds
hardware : xeon e5-1650v4
software : cutechess-gui 1.2.0
engines :
- Eman 7.80 played with nn-ac07bd334b62.nnue net and experience data from 2 learning sessions
- StockFish 14.1 played with nn-13406b1dcbe0.nnue net
settings : TC 60s+1s, 11 threads, hash 2048 MB, partial syzygy 7men

Code: Select all

   # PLAYER                :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 11T Eman experience   :     197     65    68.0      92  73.9   44   48    0  52.2       0     1
   2 11T stockfish 14.1    :       0   ----    24.0      92  26.1    0   48   44  52.2     197     1

White advantage = 103.67 +/- 32.47
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : cFLrIihXlFymGtl1dPrNfg

Re: semi-monumental achievement

Posted: Sun Sep 11, 2022 3:41 pm
by deeds
hardware : xeon e3-1275v2
software : cutechess-gui 1.2.0
engines :
- Eman 8.xx played with last green nets and experience data from 6 learning sessions
- StockFish 15.0 played with nn-6877cd24400e.nnue net
settings : TC 60s+1s, 7 threads, hash 512 MB, partial syzygy 7men

Code: Select all

   # PLAYER                :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 7T Eman experience    :     193     44   127.5     170  75.0   85   85    0  50.0       0     1
   2 7T stockfish 15.0     :       0   ----    42.5     170  25.0    0   85   85  50.0     193     1

White advantage = -19.28 +/- 22.03
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : cFLrIihXlFymGtl1dPrNfg

Re: semi-monumental achievement

Posted: Sun Sep 11, 2022 3:49 pm
by deeds
hardware : dual xeon e5-2660v3
software : cutechess-gui 1.2.0
settings : TC 60s+1s, 40 threads, hash 4096 MB, partial syzygy 7men

Code: Select all

   # PLAYER                                :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 40T Depth4_050522.exp (Eman 8.10)     :      39     27   177.5     320  55.5   42  271    7  84.7       0     1
   2 40T Solista-010722.bin (Eman 8.10)    :       0   ----   142.5     320  44.5    7  271   42  84.7      39     1

White advantage = 23.31 +/- 13.36
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : 59Ef9tOGT1RCBGYgpMTuzw

Re: semi-monumental achievement

Posted: Sun Sep 11, 2022 4:03 pm
by deeds
hardware : dual xeon e5-2660v3
software : arena 3.5.1
engines :
- Eman 8.20 played with last green net and experience data from 6 learning sessions
- Stockfish 15.0 played with nn-6877cd24400e.nnue net
settings : TC 1m+1s vs TC 4m+4s, 40 threads, hash 4096 MB, partial syzygy 7men

opening list : 100 openings (50 openings with the most experience data + 50 openings from the learned openings list)

Code: Select all

   # PLAYER                                 :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 40T 1m+1s Eman 8.20 nn-ad9b42354671    :       0     35   100.0     200  50.0   27  146   27  73.0       0     1
   2 40T 4m+4s Stockfish 15.0 x64 BMI2      :       0   ----   100.0     200  50.0   27  146   27  73.0       0     1

White advantage = -38.71 +/- 17.67
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : 555Obk6h321cD00Z-vrtFw

Re: semi-monumental achievement

Posted: Sun Sep 11, 2022 4:08 pm
by deeds
hardware : dual xeon e5-2660v3
software : cutechess-gui 1.2.0
settings : TC 60s+1s, 40 threads, hash 4096 MB, partial syzygy 7men

Code: Select all

Depth4_050522.exp    : experience data trained only with Eman
ProteusSF_030622.exp : experience data merged from Sarona's engine and Igor's data

Code: Select all

   # PLAYER                                  :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 40T Depth4_050522.exp (Eman 8.10)       :      50     25   217.0     380  57.1   70  294   16  77.4       0     1
   2 40T ProteusSF_030622.exp (Eman 8.10)    :       0   ----   163.0     380  42.9   16  294   70  77.4      50     1

White advantage = -11.30 +/- 12.83
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : ilUjtD1BLyiiL3gGfA9ihA

Re: semi-monumental achievement

Posted: Sun Sep 11, 2022 4:24 pm
by deeds
hardware : dual xeon e5-2660v3
software : arena 3.5.1
settings : TC 3m+2s, 40 threads, hash 4096 MB, partial syzygy 7men

Code: Select all

JudaS experience : JudaS 190722 played with last green net and experience data from 4 learning sessions
JudaS only       : JudaS 190722 played with last green net without experience data

Code: Select all

   # PLAYER                  :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 40T JudaS experience    :      10     36   113.0     220  51.4   37  152   31  69.1       0     1
   2 40T JudaS only          :       0   ----   107.0     220  48.6   31  152   37  69.1      10     1

White advantage = -77.78 +/- 17.19
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : cFLrIihXlFymGtl1dPrNfg

Re: semi-monumental achievement

Posted: Sun Sep 11, 2022 4:28 pm
by deeds
hardware : dual xeon e5-2660v3
software : arena 3.5.1
settings : TC 3m+2s, 40 threads, hash 4096 MB, partial syzygy 7men

Code: Select all

HypnoS experience : HypnoS 220822 played with last green net and experience data from 5 learning sessions
HypnoS only       : HypnoS 220822 played with last green net without experience data

Code: Select all

   # PLAYER                   :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 40T HypnoS experience    :      20     30   142.5     270  52.8   55  175   40  64.8       0     1
   2 40T HypnoS only          :       0   ----   127.5     270  47.2   40  175   55  64.8      20     1

White advantage = -56.48 +/- 15.46
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : cFLrIihXlFymGtl1dPrNfg

Re: semi-monumental achievement

Posted: Mon Sep 19, 2022 6:46 am
by deeds
hardware : dual xeon e5-2660v3
software : arena 3.5.1
settings : TC 3m+2s, 40 threads, hash 4096 MB, partial syzygy 7men

Code: Select all

BrainLearn experience : BrainLearn 18.0 played with last green net and experience data from 5 learning sessions
BrainLearn only       : BrainLearn 18.0 played with last green net without experience data

Code: Select all

   # PLAYER                       :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 40T BrainLearn experience    :      19     30   141.0     268  52.6   64  154   50  57.5       0     1
   2 40T BrainLearn only          :       0   ----   127.0     268  47.4   50  154   64  57.5      19     1

White advantage = -66.35 +/- 15.72
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : cFLrIihXlFymGtl1dPrNfg

Re: semi-monumental achievement

Posted: Sat Oct 29, 2022 7:01 am
by deeds
hardware : dual xeon e5-2660v3
software : arena 3.5.1
settings : TC 3m+2s, 40 threads, hash 4096 MB, partial syzygy 7men

Code: Select all

Eman experience : Eman 7.30 played with last green net and experience data from 7 learning sessions
Eman only       : Eman 7.30 played with last green net without experience data

Code: Select all

   # PLAYER                 :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 40T Eman experience    :      16     25   192.5     368  52.3   77  231   60  62.8       0     1
   2 40T Eman only          :       0   ----   175.5     368  47.7   60  231   77  62.8      16     1

White advantage = -43.17 +/- 12.97
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games
pass : cFLrIihXlFymGtl1dPrNfg

Re: semi-monumental achievement

Posted: Thu Nov 03, 2022 6:39 pm
by deeds
hardware : dual xeon e5-2660v3
software : arena 3.5.1
engines :
- Eman 8.30 played with last green net and experience data from 7 learning sessions
- Stockfish 15.0 played with nn-6877cd24400e net
settings : TC 1m+1s vs TC 5m+5s, 40 threads, hash 4096 MB, partial syzygy 7men

Here's what happens when we reinforce experience data :

[Event "reinforced experience data"]
[Site "WORKSTATION"]
[Date "2022.11.03"]
[Round "85"]
[White "1m+1s Eman 8.30 nn-ad9b42354671"]
[Black "5m+5s Stockfish 15.0 x64 BMI2"]
[Result "1-0"]
[BlackElo "2200"]
[ECO "B06"]
[Opening "Moderne"]
[Time "17:38:08"]
[Variation "1.e4 g6 2.d4 Fg7"]
[WhiteElo "2200"]
[TimeControl "300+5"]
[Termination "normal"]
[PlyCount "225"]
[WhiteType "program"]
[BlackType "program"]

1. e4 g6 2. d4 {+1.01/22 1} Bg7 {-0.70/32 13} 3. h3 {+1.09/24 0} d6
{-0.62/30 5} 4. Nf3 {+0.98/26 2} Nf6 {-0.50/30 5} 5. Bd3 {+0.98/27 1} e5
{-0.52/34 22} 6. c3 {+1.07/23 1} O-O {-0.46/36 18} 7. O-O {+1.01/25 1} Re8
{-0.46/34 12} 8. Re1 {+0.77/22 2} Nbd7 {-0.42/34 4} 9. d5 {+1.09/19 1} a5
{-0.48/35 47} 10. Bc2 {+0.97/26 3} Rf8 {-0.42/27 4} 11. Na3 {+0.76/27 7}
Kh8 {-0.47/31 13} 12. Bg5 {+0.95/23 2} h6 {-0.46/32 13} 13. Be3 {+0.94/28
2} Nh7 {-0.53/32 16} 14. Qd2 {+0.97/23 2} f5 {-0.50/34 18} 15. exf5
{+1.09/25 2} gxf5 {-0.51/33 4} 16. Bxh6 {+1.19/25 2} Rg8 {-0.51/37 6} 17.
Bxf5 {+1.42/27 6} Qf6 {-0.76/37 16} 18. Bxg7+ {+1.35/22 1} Rxg7 {-0.68/36
16} 19. Bg4 {+1.80/27 4} Ndf8 {-0.76/38 6} 20. Nh2 {+2.05/24 2} Ng6
{-0.92/36 12} 21. Nc4 {+1.94/29 6} Nf4 {-0.92/36 8} 22. Ne3 {+2.11/21 1}
Qg5 {-0.93/29 15} 23. f3 {+1.88/33 10} Nf6 {-1.10/30 9} 24. c4 {+2.27/20 0}
b6 {-1.09/26 11} 25. Rac1 {+2.45/23 3} Bd7 {-1.43/35 52} 26. Rc2 {+2.51/23
0} Rh7 {-1.44/33 52} 27. Nhf1 {+2.42/27 4} Rg8 {-1.33/26 3} 28. Bxd7
{+2.43/27 0} Nxd7 {-1.42/29 3} 29. Ng4 {+2.64/24 0} Nf8 {-1.80/31 10} 30.
Qc3 {+2.63/23 0} N8g6 {-1.33/28 2} 31. Ng3 {+2.36/27 3} Nh4 {-1.34/27 5}
32. Ne4 {+2.24/21 0} Qg7 {-0.97/24 2} 33. Kh1 {+2.37/23 1} Rf8 {-1.31/29 7}
34. Rf2 {+2.25/20 0} Nf5 {-1.11/27 2} 35. Qd2 {+2.52/23 2} Qe7 {-1.12/24 3}
36. Rg1 {+2.05/22 1} Rff7 {-1.21/30 10} 37. b3 {+2.16/20 0} Rh5 {-1.28/28
10} 38. a3 {+2.51/16 0} Rg7 {-1.79/24 4} 39. Nef6 {+2.60/21 1} Ng3+
{-2.37/25 9} 40. Kh2 {+2.59/16 0} Rgh7 {-2.34/25 5} 41. Nxh5 {+2.59/21 0}
Rxh5 {-2.27/24 2} 42. Rh1 {+2.72/18 0} Qh4 {-2.41/24 2} 43. Kg1 {+2.81/18
0} Kg7 {-2.14/26 4} 44. Qd1 {+3.49/17 0} Rh8 {-2.57/27 5} 45. Ra2 {+3.71/17
0} Kf7 {-2.43/29 9} 46. b4 {+3.51/21 3} a4 {-2.46/26 4} 47. Rd2 {+3.56/20
1} Rh7 {-2.35/25 4} 48. Qe1 {+3.67/18 0} Rh5 {-2.51/23 1} 49. Ne3 {+3.36/21
1} Rg5 {-2.69/28 2} 50. c5 {+3.61/21 0} b5 {-2.77/24 2} 51. Ng4 {+3.66/20
0} Rh5 {-2.75/23 2} 52. Rh2 {+3.71/21 0} Rg5 {-3.18/29 15} 53. Qe3
{+3.90/22 1} Rg7 {-2.79/27 6} 54. Rh1 {+3.93/23 1} Rh7 {-2.85/24 0} 55. Qe1
{+3.91/24 1} Rh8 {-2.65/26 1} 56. Ne3 {+3.77/21 1} Rg8 {-2.64/26 2} 57. Ng4
{+3.93/21 0} Rh8 {-2.65/28 2} 58. Qf2 {+3.91/22 0} Kg7 {-3.13/31 16} 59.
Ne3 {+3.94/19 0} Kf6 {-3.04/28 4} 60. Rh2 {+3.51/18 0} Nfh5 {-3.23/26 1}
61. Ng4+ {+4.11/20 1} Kg6 {-3.36/28 4} 62. Qe1 {+4.08/18 0} Ng7 {-3.48/24
2} 63. Nf2 {+4.33/20 0} N7f5 {-3.29/27 2} 64. Ne4 {+4.30/20 0} Rg8
{-3.79/30 9} 65. Rd1 {+4.31/20 0} Kf7 {-4.09/27 10} 66. Qf2 {+4.53/21 0}
Rg6 {-4.20/25 4} 67. Re1 {+4.86/20 0} Kg8 {-4.44/24 4} 68. Rh1 {+4.58/25 3}
Qf4 {-4.49/24 2} 69. Rh2 {+4.88/15 0} Qh4 {-4.91/26 8} 70. Qd2 {+5.21/19 0}
Kf7 {-3.58/19 1} 71. Nxg3 {+6.30/21 2} Nxg3 {-5.76/24 8} 72. f4 {+6.50/18
1} exf4 {-5.96/25 5} 73. Qd3 {+6.39/17 0} Kg7 {-6.33/23 5} 74. Qc3+
{+7.00/15 0} Kh6 {-6.33/24 5} 75. Qh8+ {+8.41/16 0} Kg5 {-6.75/19 0} 76.
Qd8+ {+9.13/17 0} Kh5 {-7.06/26 6} 77. Qxc7 {+9.36/18 1} Kh6 {-8.45/21 7}
78. cxd6 {+9.43/16 0} Qf6 {-8.60/27 3} 79. Qc5 {+10.45/16 0} Kh7 {-9.48/24
6} 80. d7 {+11.37/16 0} Rg7 {-10.67/22 5} 81. Qc2+ {+11.82/17 0} Kh8
{-11.21/24 2} 82. Re8+ {+12.39/19 0} Rg8 {-11.24/21 3} 83. Qf2 {+12.64/21
1} Qa1+ {-12.90/23 7} 84. Re1 {+12.72/20 0} Qg7 {-13.63/24 5} 85. Qxf4
{+12.97/24 0} Qxd7 {-13.99/20 5} 86. Qh6+ {+13.94/22 3} Qh7 {-15.61/18 0}
87. Qf6+ {+14.18/18 0} Qg7 {-15.69/26 8} 88. Qh4+ {+14.28/17 0} Qh7
{-18.45/24 3} 89. Qf6+ {+14.39/20 1} Qg7 {-18.47/22 1} 90. Qh4+ {+14.50/19
0} Qh7 {-20.12/23 8} 91. Qxh7+ {+14.59/19 0} Kxh7 {-20.49/15 0} 92. h4
{+14.74/22 0} Nh5 {-21.46/23 10} 93. d6 {+15.18/20 0} Rd8 {-22.34/23 6} 94.
Rd1 {+15.59/23 1} Nf6 {-24.69/19 5} 95. Rh3 {+16.34/19 0} Nd7 {-36.47/23 5}
96. Re1 {+17.90/18 0} Nb6 {-21.37/14 1} 97. Re7+ {+20.02/17 0} Kg6
{-53.34/23 8} 98. Rg3+ {+100.00/37 0} Kh6 {-152.51/245 3} 99. Re6+
{+100.00/39 0} Kh7 {-M20/245 5} 100. h5 {+M18/47 1} Nd7 {-151.41/23 1} 101.
Re7+ {+M15/49 0} Kh8 {-M12/80 4} 102. Rgg7 {+M13/56 0} Nf8 {-M13/29 1} 103.
d7 {+M12/61 0} Ne6 {-M10/83 1} 104. Rg6 {+M10/65 0} Nf8 {-M9/89 2} 105. Rd6
{+M9/76 0} Nxd7 {-M8/91 2} 106. Rdxd7 {+M8/80 0} Rxd7 {-M7/175 2} 107. Rxd7
{+M7/97 1} Kg8 {-M6/75 0} 108. h6 {+M6/101 1} Kh8 {-M5/245 2} 109. g4
{+M5/108 1} Kg8 {-M4/86 0} 110. g5 {+M4/117 1} Kf8 {-M3/245 2} 111. h7
{+M3/95 1} Ke8 {-M2/95 1} 112. Rg7 {+M2/103 1} Kf8 {-M1/245 2} 113. h8=Q#
{+M1/112 1} 1-0

[Event "reinforced experience data"]
[Site "WORKSTATION"]
[Date "2022.11.03"]
[Round "86"]
[White "5m+5s Stockfish 15.0 x64 BMI2"]
[Black "1m+1s Eman 8.30 nn-ad9b42354671"]
[Result "1/2-1/2"]
[BlackElo "2200"]
[ECO "B07"]
[Opening "Pirc"]
[Time "17:55:55"]
[Variation "Attaque 150, 4.Fe3 Fg7 5.Dd2 c6"]
[WhiteElo "2200"]
[TimeControl "300+5"]
[Termination "adjudication"]
[PlyCount "88"]
[WhiteType "program"]
[BlackType "program"]

1. e4 g6 2. d4 {+0.71/32 11} Bg7 {-0.88/24 4} 3. Nc3 {+0.70/33 20} d6
{-0.73/21 3} 4. Be3 {+0.88/26 4} Nf6 {-0.87/27 1} 5. Qd2 {+0.86/28 8} c6
{-1.13/21 1} 6. Nf3 {+0.82/31 18} O-O {-0.78/27 5} 7. Bh6 {+0.75/31 6} d5
{-0.76/24 9} 8. Bxg7 {+0.53/40 49} Kxg7 {-0.93/28 3} 9. e5 {+0.56/36 21}
Ne4 {-0.76/26 2} 10. Nxe4 {+0.49/32 5} dxe4 {-0.51/29 5} 11. Ng5 {+0.65/33
6} Bf5 {-0.82/32 6} 12. f3 {+0.68/34 6} exf3 {-0.90/30 1} 13. Nxf3
{+0.61/33 6} Qd5 {-0.94/31 2} 14. Bd3 {+0.62/33 9} Bxd3 {-0.98/32 3} 15.
Qxd3 {+0.58/33 19} Qb5 {-0.94/27 2} 16. Qxb5 {+0.64/34 11} cxb5 {-0.86/26
1} 17. c3 {+0.62/35 10} Nd7 {-0.90/23 1} 18. Ke2 {+0.48/32 14} Nb6
{-0.76/26 2} 19. Kd3 {+0.50/40 21} Rac8 {-0.87/24 1} 20. Rhe1 {+0.45/33 22}
f5 {-0.83/29 5} 21. exf6+ {+0.44/33 4} exf6 {-0.85/30 0} 22. g3 {+0.42/34
9} Rf7 {-0.86/28 1} 23. Re2 {+0.33/38 16} Rd7 {-0.79/28 1} 24. Rae1
{+0.33/32 16} Kf7 {-0.85/29 4} 25. Rf2 {+0.42/39 20} Kg7 {-0.82/31 0} 26.
Rfe2 {+0.40/40 13} Kf7 {-0.75/29 2} 27. Rf2 {+0.40/37 18} Kg7 {-0.71/27 0}
28. a3 {+0.28/36 6} h5 {-0.74/28 0} 29. Nh4 {+0.31/36 19} Kf7 {-0.80/24 0}
30. h3 {+0.41/39 16} Nd5 {-0.76/26 1} 31. Ng2 {+0.36/35 2} Re7 {-0.66/24 1}
32. Rxe7+ {+0.50/29 3} Kxe7 {-0.56/23 0} 33. Nf4 {+0.13/42 23} Nxf4+
{-0.65/25 0} 34. Rxf4 {+0.12/38 3} Ke6 {-0.73/25 1} 35. Re4+ {+0.23/33 12}
Kd7 {-0.48/27 2} 36. Re2 {+0.30/32 1} b6 {-0.34/25 0} 37. Rf2 {+0.07/34 3}
Ke6 {-0.28/24 0} 38. Re2+ {+0.09/33 5} Kd6 {-0.35/26 0} 39. b3 {+0.11/38 1}
h4 {-0.36/32 0} 40. Rf2 {+0.24/39 2} Ke6 {-0.27/31 0} 41. g4 {0.00/38 4} f5
{-0.22/29 0} 42. gxf5+ {0.00/48 3} gxf5 {0.00/31 0} 43. Re2+ {0.00/49 3}
Kd7 {0.00/35 1} 44. Rf2 {0.00/75 4} Rg8 {0.00/37 0 Adjugé par Arena}
1/2-1/2

As we saw previously, with 4x longer TC, SF15 could not win against experience data learned at TC 1m+1s (avg. D27).

Now, against reinforced experience data (avg. D32), SF15 is in great difficulty even with 5x longer TC.

Since Eman is now training at TC 2m+2s (avg. D37), SF16 will fall into a butchery...