Stockfish NN release (NNUE)

[Raphexon]

[pgn][Event "Computer chess game"]
[Site "DESKTOP-CQJF90G"]
[Date "2020.06.03"]
[Round "?"]
[White "Stockfish.nnue.halfkp_256x2-32-32"]
[Black "Houdini_15a_x64"]
[Result "1-0"]
[BlackElo "2000"]
[ECO "C50"]
[Opening "Giuoco Pianissimo"]
[Time "13:07:32"]
[Variation "Italian Four Knights, 5...d6"]
[WhiteElo "2000"]
[TimeControl "600+2"]
[Termination "normal"]
[PlyCount "125"]
[WhiteType "program"]
[BlackType "program"]

1. e4 {(e4 c5 c3 e6 d4 d5 exd5 exd5 Nf3 Bd6 Bd3 Nf6 0-0 0-0 dxc5 Bxc5 b4
Bb6 a4 Nc6 a5 Bc7 Nbd2 Bg4 h3 Ne5 Bxh7+ Nxh7 hxg4 Nxg4 Bb2 Re8) +0.12/25
13} e5 {(e5 Nf3 Nc6 Bb5 Nf6 0-0 Bd6 d4 Qe7 d5 Nd4 Nxd4 exd4 Qxd4 Be5 Qa4
Nxe4 Nd2 Nc5 Qa3 0-0 Re1 d6 Nf3 Bf5 Nxe5 dxe5 Bf4) -0.11/22 28} 2. Nf3
{(Nf3 Nc6 Bc4 Nf6 d3 Be7 0-0 d6 c3 Na5 Bb5+ c6 Ba4 b5 Bc2 c5 d4 cxd4 cxd4
0-0 h3 a6 dxe5 dxe5 Qxd8 Rxd8 Nxe5 Bb7 Re1 Nc6) +0.26/26 16} Nc6 {(Nc6 Bb5
Bc5 0-0 d6 c3 a6 Bxc6+ bxc6 d4 exd4 cxd4 Bb6 Nc3 Ne7 Be3 0-0 h3 a5 Re1 Ba6
Qc2 Qd7 d5 Bxe3 Rxe3 c5 Rae1 Rab8 b3 Ng6) -0.15/22 19} 3. Bc4 {(Bc4 Nf6 d3
Be7 c3 d6 0-0 Na5 Bb5+ c6 Ba4 0-0 b4 b5 Bc2 Nb7 a4 a5 d4 Qc7 Nbd2 bxa4 Rxa4
Bd7 dxe5 dxe5 Nc4 axb4 Rxa8 Rxa8) +0.22/26 17} Nf6 {(Nf6 Nc3 Bc5 d3 0-0 0-0
Rb8 Nd5 h6 Be3 d6 Qe2 b5 Nxf6+ Qxf6 Bd5 Ne7 Bxc5 dxc5 a4 bxa4 Rxa4 Rxb2
Rxa7 Nxd5 exd5) -0.03/21 14} 4. d3 {(d3 Be7) +0.27/24 5} Bc5 {(Bc5 Nc3 d6
0-0 h6 a4 Bg4 h3 Be6 Bxe6 fxe6 Ne2 0-0 c3 Bb6 b4 a6 Bd2 Qe7 b5 axb5 axb5
Rxa1 Qxa1 Na5 Qa4 Qd7) +0.01/22 22} 5. Nc3 {(Nc3 0-0) +0.16/24 13} d6 {(d6
0-0 0-0 Bg5 h6 Bxf6 Qxf6 Nd5 Qd8 c3 a6 Bb3 Ba7 Ne3 Na5 Bc2 Be6 b4 Bxe3 fxe3
Nc6 Bb3 Qd7 Bxe6 Qxe6 Nd2 Ne7 Qf3 f5 exf5 Nxf5 Qxb7 Nxe3 Rxf8+ Rxf8 Qxc7)
0.00/22 10} 6. Na4 {(Na4 Bb6 c3 0-0 b4 Ne7 h3 Ng6 0-0 h6 Re1 Be6 Bxe6 fxe6
d4 Kh7 Bd2 c6 Nxb6 axb6 a4 c5 a5 c4 Qe2 bxa5 Qxc4) +0.18/26 8} Bb6 {(Bb6 c3
0-0 0-0 h6 b4 Qe8 Bb3 Be6 h3 Qd7 Qc2 Ne7 Nxb6 axb6 Be3 Ng6 Rfd1 Qe7 c4 Rfd8
Qd2 Bd7 a4) +0.05/22 12} 7. c3 {(c3 h6 0-0 0-0 b4 Ne7 Nxb6 axb6 a4 Ng6 h3
Be6 Bxe6 fxe6 a5 bxa5 bxa5 Nd7 Be3 Kh8 Qb3 Nf4 Bxf4 Rxf4 Qxe6 Nc5 Qg6 Nxd3)
+0.19/24 5} O-O {(0-0 0-0 h6 b4 Qe8 Bb3 Be6 h3 Qd7 Qc2 Rfd8 Rd1 Ne7 d4 Bxb3
axb3 exd4 cxd4 Qe6 Nc3 Nc6 Ra4 a6 d5) +0.02/23 15} 8. O-O {(0-0 h6)
+0.12/26 21} h6 {(h6 b4 Ne7 Nxb6 axb6 a4 Ng6 Be3 d5 exd5 Nxd5 Qb3 Be6 d4 e4
Ne5 Nxe5 dxe5 Nxe3 fxe3 Qd2 Kh1 Kh7 Rad1 Qxe3 Rde1 Bxc4 Qxc4 Qd3 Rxe4 Rfd8
a5) +0.06/23 14} 9. Re1 {(Re1) +0.16/26 14} Na5 {(Na5 Bb3 Nxb3 axb3 Bd7 h3
Qe7 Qe2 Qe6 b4 Rfe8 Qc2 Nh5 Nxb6 axb6 Be3 Nf6 c4 b5 b3 Bc6 Nd2 bxc4 bxc4 b5
cxb5 Bxb5 Qxc7 Bxd3) +0.11/22 10} 10. Nxb6 {(Nxb6 axb6 Bb5 Bg4 Ba4 c5 h3
Bh5 a3 d5 b4 dxe4 dxe4 Nc4 Bb3 Qxd1 Rxd1 b5 g4 Bg6 Bxc4 bxc4 bxc5 Nxe4 Nxe5
Nxc3 Nxg6 Nxd1) +0.09/25 5} axb6 {(axb6 Bb5 Bd7 Bxd7 Qxd7 a4 Nc6 h3 Rfe8
Nh4 b5 Qf3 Qe6 Nf5 Kh8 b3 bxa4 bxa4 Ne7 c4 Kg8 Bd2 Nxf5 exf5 Qd7 a5 d5)
+0.10/23 8} 11. Bb5 {(Bb5 Bg4 Ba4 Bh5 h3 c5 Bd2 c4 Bb5 cxd3 g4 Bg6 Bxd3 Qc7
Nh4 Nc4 b3 Nb2 Qc2 Nxd3 Qxd3 Kh7 c4 Nd7 Nxg6 fxg6) +0.14/25 8} Bd7 {(Bd7
Bxd7 Qxd7 a4 Nc6 h3 Rfe8 c4 Qe6 Qd2 Ne7 Qe2 Ng6 b3 Rf8 Bd2 Nh5 Ng5 hxg5
Qxh5 Nf4 Bxf4 gxf4 Qg4) +0.13/23 12} 12. Bxd7 {(Bxd7 Qxd7 Bd2 c5 b4 Nc6 a4
Rfe8 Qc2 Rec8 Qb3 Qe6 Qd1 Ne7 h3 Ng6 c4 cxb4 Rb1 Nd7 Rxb4 Nc5 Qc2 f5 Bc3
Nf4 exf5 Qxf5 Rxb6) +0.24/25 8} Qxd7 {(Qxd7 a4 Nc6 c4 Rfe8 Bd2 Qe6 b3 Nd7
Bc3 Qg6 b4 Qf6 Qb3 Nd8 d4 Ne6 d5 Ng5 Re3 Qf4 Nxg5 hxg5 g3) +0.11/23 17} 13.
a4 {(a4 Nc6) +0.11/29 52} Nc6 {(Nc6 c4 Rfe8 Bd2 Qe6 b3 Nd7 Bc3 Qg6 b4 Qf6
Re3 Qe6 Re1 Ne7 h3 Qf6 Qb3 Ng6 Bd2 Nh4 Nxh4 Qxh4 Rac1 Qf6 Re2 Qg6 Be3)
+0.11/23 8} 14. c4 {(c4 Nb4 b3 c5 Nh4 Kh7 Nf5 Rae8 h3 Qd8 Rf1 Ng8 Bb2 g6
Ng3 Nc6 Bc3 f5 exf5 gxf5 Qf3 f4 Ne4 Nf6 Bd2 Nd4 Nxf6+ Rxf6 Qxb7+ Re7 Qe4+
Rg6) +0.34/26 16} Rfe8 {(Rfe8 Bd2 Qe6 b3 Nd7 Bc3 Qg6 b4 Qf6 Re3 Qe6 Re1 Ne7
h3 Qf6 Qb3 Ng6 Bd2 Nh4 Nxh4 Qxh4 Rac1 Qf6 Re2 Qg6) +0.11/22 12} 15. Bd2
{(Bd2 Nd8) +0.18/26 26} Qe6 {(Qe6 b3 Nd7 Bc3 Qg4 b4 Qh5 h3 Nf8 b5 Na5 Qc2
Ne6 Bd2 Qg6 Be3 Qf6 Rad1 Rad8 d4 Rf8 dxe5 dxe5 Rd5) +0.08/21 8} 16. Nh4
{(Nh4 Nd4 Nf3 Nc6 h3 Nd7 Bc3 Ne7 Nh2 f5 exf5 Qxf5 d4 Ng6 Qf3 Qf7 Qxb7 exd4
Bxd4 Nde5 Bxe5 Nxe5 Ng4 Nxg4 hxg4 Rxe1+ Rxe1 Rxa4 Qb8+ Kh7) +0.22/26 20}
Ne7 {(Ne7 b4 Nd7 Re3 Kh8 h3 Nf6 Re1 g5 Nf3 Rg8 Nh2 Ng6 Qc2 Kh7 f3 Nh5 Ng4
Nhf4 Bxf4 Nxf4 Ne3 c5 Rab1) +0.06/21 14} 17. b4 {(b4 g5 Nf3 Ng6 Be3 Qg4 Nd2
Qxd1 Rexd1 Ne7 Rdb1 Ra7 a5 Rea8 Nb3 Ng4 Bd2 f5 h3 Nf6 exf5 Nxf5 g4 Nh4 Kf1
d5 cxd5 Nxd5) +0.28/25 13} Ra7 {(Ra7 Re3 Rea8 Rg3 Kh8 h3 c6 Re3 Rd8 Re1 Kg8
Be3 c5 Qb3 Nd7 b5 Raa8 Bd2 Re8 Qc2 Nf6 Rf1 Nh5 a5 bxa5 Bxa5 Nf4) +0.04/21
7} 18. h3 {(h3 g5 Nf3 Ng6 Nh2 Nf4 Bxf4 gxf4 Nf3 Kh7 Kh2 c5 b5 Rg8 Nh4 Raa8
Qf3 Rg5 Nf5 Rag8 a5 Rxg2+ Qxg2 Rxg2+ Kxg2 bxa5) +0.45/21 4} Rea8 {(Rea8 Bc3
Nh7 Re3 Ng5 Rg3 Qf6 Nf3 Ng6 Nxg5 hxg5 Bd2 Nf4 Ra2 Re8 Rg4 Raa8 Be3 Kh8 d4
Qg6 Qf3 Ne6 dxe5 dxe5) +0.07/21 11} 19. Be3 {(Be3 c5) +0.25/28 38} Nd7
{(Nd7 Nf3 Qf6 Re2 c5 b5 Ng6 Rea2 Ndf8 g3 Ne6 Bd2 Qd8 Qe2 Qc7 Bc3 Nd4 Bxd4
exd4 Kg2 Ne5 Nxe5 dxe5 Qh5 Qd6) +0.13/23 11} 20. Bd2 {(Bd2 c5) +0.04/25 14}
Nf8 {(Nf8 Be3 Qf6 Nf3 c5 b5 Ne6 g3 Rd8 Kg2 Rf8 Qc2 Re8 Qd1 Rd8 Qd2 Rda8 Qc2
Rf8 Qb3 Raa8 Qc3 Rfe8 a5 bxa5 Rxa5 b6) +0.18/23 12} 21. Be3 {(Be3 c5 b5 f5
exf5 Nxf5 Nxf5 Qxf5 Bc1 Ne6 Re3 Nd4 Rg3 Qf6 Bb2 Ne6 Bc1) 0.00/26 15} Qf6
{(Qf6 Nf3 Ne6 d4 c5 d5 Nd4 Bxd4 cxd4 g3 Ng6 Kg2 Qe7 Re2 Rc8 Qb3 b5 cxb5 Rc3
Qd1 Qc7 a5 Qc4 b6 Ra8 Rb1 Rc8) +0.19/24 27} 22. Nf3 {(Nf3 c5 b5 Ne6 Nd2 Kh7
Nb1 g5 Nc3 Qg6 g3 Nd4 Rf1 Rf8 Rc1 Rfa8 Kg2 f5 Bxd4 exd4 exf5 Nxf5 Nd5 Rxa4
Nxb6) +0.15/25 4} Ne6 {(Ne6 d4 c5 d5 Nd4 Bxd4 cxd4 g3 Ng6 Kg2 Qe7 Re2 Rc8
Qb3 b5 cxb5 Rc3 Qd1 Qc7 a5 Qc4 b6 Ra8 Rb1 Rc8 Nd2 Qb5 Rb2) +0.15/22 8} 23.
d4 {(d4) +0.18/26 6} c5 {(c5 d5 Nd4 Bxd4 cxd4 g3 Ng6 Kg2 Qe7 Re2 Qc7 Qb3 h5
Rea2 h4 Qd3 Qe7 Ra3 Qd7 Qc2 Rc8 Qd3 hxg3 fxg3 Qc7 Nd2 Qd7 a5) +0.18/23 8}
24. d5 {(d5 Nd4) +0.18/29 8} Nd4 {(Nd4 Bxd4 cxd4 g3 Ng6 Kg2 Qe7 Re2 Nf8
Rea2 Nd7 Qd3 Nf6 Nd2 Qd7 Qb3 Qc7 Qf3 Qb8 Qd3 Qe8 Qb3 Qd7 a5 bxa5 Rxa5 Rxa5
Rxa5 Rxa5 bxa5) +0.18/25 6} 25. Nxd4 {(Nxd4 exd4 Bd2 g5 Qb3 Ng6 Ra3 Kh7 Rb1
Rg8 bxc5 bxc5 a5 Ne5 f3 h5 Rf1 Qg6 Ra4 d3 Qb6 Raa8 Qc7 b6 Qxb6 Rab8 Qa7)
+0.15/26 8} exd4 {(exd4 Bd2 Ng6 Qb3 Ra6 Ra2 Qe7 f4 h5 Qd1 h4 Qb3 R6a7 bxc5
bxc5 a5 Qf6 Rf1 Nf8 Rb2 Qe7 Qd3 Nd7 Re1 Re8) +0.23/23 19} 26. Bd2 {(Bd2 g5)
+0.15/26 6} Ng6 {(Ng6 Qb3 h5 bxc5 bxc5 a5 Re8 Rab1 Re7 Qd3 Ne5 Qb3 Re8 f4
Nd7 Qf3 h4 Qg4 Qe7 Qf3 Nf6 e5) +0.24/23 23} 27. Qb3 {(Qb3 cxb4 Bxb4 Rc8 Bd2
Raa8 Rac1 Ne5 Bb4 d3 Bd2 Qg6 Re3 Qf6) 0.00/27 13} Ra6 {(Ra6 Ra2 h5 Rb1 Qe5
Qc2 R6a7 Ra3 Nf4 Bxf4 Qxf4 Rf3 Qh6 bxc5 bxc5 Ra1 Qg6 Rf4 b6 Qb3 Qg5 Rf5 Qe7
Rxh5 Qxe4) +0.23/23 27} 28. Ra3 {(Ra3 Qh4 Ra2 Kh7 Raa1 Kg8 Ra2) 0.00/30 12}
h5 {(h5 Qc2 Qd8 Rb1 Qd7 Rba1 Ne5 b5 R6a7 f4 Ng6 Qb3 Qd8 Qf3 h4 Qb3 Qe8 Qf3
Qe7 a5 bxa5 Rxa5 Rxa5 Rxa5 Rxa5 Bxa5 Nf8 Kf2 Nd7 Ke2) +0.21/22 10} 29. b5
{(b5 R6a7 Qd1 d3 Bc3 Qg5 Bd2 Qh4 Qb3 Ne5 Bc3 Ng6) 0.00/28 9} R6a7 {(R6a7
Qc2 Ne5 Qb3 h4 Rea1 Qg6 f4 Nd7 Qc2 Re8 Re1 Qf6 Qd3 Raa8 Ra2 Qh6 Qf3 Qg6 a5
bxa5 Rxa5 Rxa5 Bxa5 b6 Bd2 Ra8 e5) +0.35/22 6} 30. Qd1 {(Qd1 Ne5 Qc2 Qh4 f4
d3 Qb3 Rxa4 Rxa4 Nf3+ gxf3 Qg3+ Kf1 Qxf3+ Kg1 Qg3+ Kf1 Qxh3+ Kg1 Qg3+ Kh1
Qh3+) 0.00/32 4} Ne5 {(Ne5 Qb3 h4 Qc2 Qg6 Kh1 Nd7 Kg1 Qh5 Bf4 Ne5 Ra2 Re8
Qb3 Ra5 Rae2 Qg6 Ra2 Raa8 Bh2 Re7 Rae2 Ra5 Rb2 Qg5 Bxe5 Qxe5) +0.33/24 5}
31. Qc2 {(Qc2 Qh4 f4 d3 Qb3 Nf3+ gxf3 Qg3+ Kf1 Qxf3+ Kg1 Qg3+ Kf1) 0.00/36
6} h4 {(h4 Qa2 Nd7 Qb3 Qe5 Qc2 Re8 Qb3 Qe7 Bf4 Qf6 Bh2 Ne5 Ra2 Ra5 Rd1 Raa8
Rad2 Qe7 Re1 Qg5 Ra2 Qh6 Rae2 Qg6 Bf4 Re7 Ra2 Qf6 Bh2 Qh6 Bxe5 Rxe5 Qf3 Qg5
Rd1 Qe7) +0.33/23 8} 32. f4 {(f4 Ng6 Rf1 Qe7 Qd1 Qxe4 f5 Nf8 f6 gxf6 Rxf6
d3 Qh5 Qd4+ Rf2 Ng6 Bc3 Qe3 Qf5 Ne5 Bxe5 Qxe5 Qxf7+ Kh8 Rf5 Qe3+ Kh2 Qg3+
Kh1 Qe1+ Rf1 d2 Qf6+ Kg8 Qg6+ Kh8) +0.52/28 5} Nd7 {(Nd7 Qa2 Qe7 Qa1 Re8 a5
Rea8 a6 bxa6 Rxa6 Nb8 Rxa7 Rxa7 Qb1 f6 Rf1 Nd7 Qd3 Ra8 Bc1 Ra2 Rf2 Ra1 Rf1
Nf8 Bb2 Ra2) +0.25/23 7} 33. Kh1 {(Kh1 Qg6) +0.01/27 16} Re8 {(Re8 Kg1 Qd8
Rf3 Qa8 Ra3 Qb8 f5 Qd8 Qc1 f6 Bf4 Qe7 Qc2 Ne5 Bxe5 Qxe5 Qe2 Rea8 Rea1 Qf4
Rf3 Qg5 Qc2 Ra5 Rd3 Re8) +0.31/22 6} 34. g4 {(g4 Qg6 Kh2 Qf6 Kg1 Qg6 Qd3
Qh7 Kg2 Raa8 Qc2 Qg6 Rf3 Ra7 Bc1 Rea8 Ra3 Qf6 Bd2 Qg6 Re2 Re8 Qd3 Qh7 Ra1
g5 fxg5 Qg6 Qb3 Rxe4) +0.46/26 12} Qe7 {(Qe7 Kg2 f6 Qd3 Qf7 Rd1 Raa8 Rda1
Qe7 Re1 Nf8 Qf3 Ng6 a5 bxa5 Bxa5 b6 Bd2 Rxa3 Qxa3 Kh7 Qd3 Ra8 e5 fxe5 fxe5
dxe5 Qf5 Ra2) +0.26/21 6} 35. g5 {(g5 Qf8 Kh2 Raa8 Rf3 Ra7 Rff1 g6 Ra1 Qg7
Rfe1 Qh8 Rg1 Qg7 Qd3 Rea8 Rg4 Rxa4 Rxa4 Rxa4 Rxh4) +0.70/22 5} Qf8 {(Qf8
Raa1 Raa8 Re2 f5 Rae1 fxe4 Rxe4 Rxe4 Rxe4 Ne5 Kg2 Ng6 Bc1 Qf7 Kg1 Rf8 Qd3
Qf5 Kh2 Ne7 Qf3 Qf7 Re6 Nf5) +0.17/21 6} 36. Kh2 {(Kh2 g6 Ra2 Qg7 Qd3 Qh8
Kg2 Qh5 Raa1 Kg7 Kh2 Raa8 Rg1 Re7 Rg4 Rae8 Re1 Qh7 Rg2 Qh5 Ree2 Ra8 a5 bxa5
Be1 Kg8 Rg4 Qh7) +0.51/23 8} Re7 {(Re7 Raa1 Ra8 Rg1 g6 Rae1 Qe8 Re2 Qb8 Kg2
Kg7 Rf1 Re8 Ra1 Qd8 Be1 Rh8 Qd3 Qe7 a5 bxa5 Rxa5 Rxa5 Bxa5 f6 gxf6+ Qxf6
Qf3 b6 Qg4) +0.19/21 8} 37. f5 {(f5 Re8 Bf4 Qe7 Raa1 Ne5 Rf1 Rea8 Kh1 Qe8
Ra3 Qe7 Kg1 d3 Qb3 Nd7 Re1 Ne5 Kg2 Re8 Raa1 Rea8 Ra2 Rf8 Rf1 Rfa8 Bd2 Nd7
Qxd3 Rxa4 f6 gxf6) +0.36/30 37} Ne5 {(Ne5 f6 Re8 Rf1 Rea8 Bf4 Ng6 Kg2 gxf6
gxf6 Qb8 Qd1 Qc7 Bh2 Qd8 Qc2 Ra5 Qd1 Kf8 Qg4 R5a7 Qd1 Qd7 Rf5 Re8 Qd3 Kg8
Ra1 Raa8) +0.05/21 5} 38. Rf1 {(Rf1 Qe8 Rf4 Nxc4 Qxc4 Rxe4 Rf2 Qe5+ Kg1 Ra8
Raf3 Re8 g6 Rf8 Qf1 f6 Qc4 Re8 Kg2 Qe7 Kf1 Qe5 Kg1 d3 Qxd3 Rd4 Qc2 Qxd5 Bc3
Rd1+ Kh2 Re3) +1.81/24 5} Re8 {(Re8 f6 Rea8 Bf4 Ng6 Kg2 gxf6 gxf6 Qb8 Bh2
Qd8 Rf5 Ra5 Ra1 Qd7 Rh5 Re8 Rg5 Kf8 Rg4 Rea8 Qb3 Kg8 e5) 0.00/20 11} 39.
Rf4 {(Rf4 d3 Qd1 g6 Rxh4 Qe7 Rh6 Nxc4 Bc3 Ne5 h4 c4 Kg2 Raa8 Qf3 Rac8 Ra1
Rf8 Rf1 Rce8 Qf4 f6 Bxe5 Qxe5 Rxg6+ Kh8 Rh6+ Kg8 Qxe5 Rxe5 Rg6+ Kh8)
+2.95/24 6} Qe7 {(Qe7 f6 Qd7 Rxh4 Ng6 Rg4 Raa8 Bf4 Nxf4 Rxf4 g6 Kg2 Re5 Rg4
Ra5 Qd3 Re8 Ra2 Raa8 Ra1 Re5 Ra2 Ra7 Ra1 Ra8 Ra2) -0.13/19 6} 40. Rxh4
{(Rxh4 Rea8 Kg2 Qd7 Ra1 Qe8 Qb3 Rc8 Qd1 g6 Qb3 Nd7 f6 Qe5 Ra3 d3 Qxd3 Qd4
Bf4 Qxd3 Rxd3 Rxa4 Bxd6 Rxc4 Kg3) +3.66/24 5} Qc7 {(Qc7 Kg2 Rea8 Kg3 Qe7
Bf4 Re8 Qd1 g6 Bxe5 Qxe5+ Rf4 gxf5 exf5 Qe1+ Qxe1 Rxe1 Rh4 Rg1+ Kf4 Rf1+
Rf3 Rxf3+ Kxf3 Rxa4 Rh6 Ra3+ Kf4) -0.22/19 12} 41. Bf4 {(Bf4 Qe7 Qe2 Ng6
fxg6 fxg6 Rg4 Qd7 Ra1 Rea8 Qc2 Qe7 Ra3 Rf8 Bg3 Rfa8 Qb3 Ra5 Qd1 Qd7 e5 dxe5
Bxe5 Re8 Bf4) +5.72/24 4} Qe7 {(Qe7 Kg3 Rea8 Qd1 g6 Bxe5 Qxe5+ Rf4 gxf5
exf5 Re8 h4 Qe1+ Qxe1 Rxe1 g6 fxg6 fxg6 Rg1+ Kf2 Rxg6 h5 Rg7 Rf6 Rd7 Ke2
Kh7 Kd2 Ra5 Ra1 Ra8 Kd3 Rg8 Raf1 Rg3+ R1f3 Rdg7) -2.15/20 30} 42. Qe2
{(Qe2) +6.73/26 5} Ng6 {(Ng6 fxg6 fxg6 Bg3 Raa8 Kg2 Rf8 Qg4 Rae8 Rf3 Rxf3
Kxf3 Qc7 Ke2 Qd8 Kd3 Re7 Qf4 Re5 Qf3 Re8 Qf2 Re7 Rf4 Qa8 Qc2 Qc8 h4 Qh3
Rf3) -3.88/19 4} 43. fxg6 {(fxg6) +6.97/26 4} fxg6 {(fxg6 Bg3 Raa8 Kg2 Rf8
Qg4 Rae8 Rf3 Rxf3 Kxf3 Qc7 Ke2 Qd8 Kd3 Re7 Qf4 Re5 Qf3 Re8 Qf2 Re7 Rf4 Qa8
Qc2 Qc8 h4 Qh3 Rf3) -3.82/18 0} 44. Rg4 {(Rg4) +7.26/26 5} Qd7 {(Qd7 Bg3
Raa8 e5 dxe5 Re4 Re7 h4 Rf8 Ra1 Ra8 Kg2 Ree8 Bxe5 Re7 Kh2 Qe8 d6 Re6 Kg3
Ra5 Ra3 Ra8 Ra2 Ra5 Ra1 d3 Qe3) -3.88/20 5} 45. Bg3 {(Bg3) +7.73/25 6} Raa8
{(Raa8 e5 dxe5 Re4 Kh7 Kg2 Qd8 Rxe5 Rxe5 Qxe5 Qd7 Qe2 Re8 Qg4 Qe7 h4 Rd8
Qd1 Qe4+ Rf3 Re8 Qd3 Qg4 d6 Ra8 Rf7 Rxa4 d7 Ra2+ Kg1 Ra1+ Kh2 Ra2+ Bf2
Rxf2+ Rxf2 Qxd7 Kg3 Qd6+ Kg2) -4.51/20 7} 46. Qd3 {(Qd3 Rf8 e5 dxe5 Qxg6
Rae8 Re4 Qf5 Qxf5 Rxf5 g6 Rg5 Rg4 Rf5 Bh4 Rff8 Kg2 e4 d6 d3 d7 Re6 d8Q Rxd8
Bxd8 d2 Ra1 Rd6 Bh4 d1Q) +8.21/24 5} Kf7 {(Kf7 Qc2 Kg8 Kg2 Rac8 e5 Qf5 Qe4
dxe5 Rf3 Qxe4 Rxe4 Ra8 Ra3 Kf7 Bxe5 Re7 Rf3+ Kg8 d6 Ree8 Rf2 Rad8 Kf3 Rd7
Bg3 Rf7+ Bf4) -5.00/18 7} 47. e5 {(e5) +9.27/25 4} dxe5 {(dxe5 Qe4 Kg8 Qxg6
Rf8 Bxe5 Qf7 Qxf7+ Rxf7 Kg3 Re7 Re4 Rae8 d6 Rd7 Kf4 g6 Rf3 Re6 h4 Re8 Rf2
Ra8 Ra2 Rf8+ Kg4 Rf1 Rd2) -5.81/19 4} 48. Qe4 {(Qe4) +9.69/24 3} Kg8 {(Kg8
Qxg6 Rf8 Bxe5 Qf7 Qxf7+ Rxf7 Kg3 Re7 Re4 Rae8 d6 Rd7 Kf3 Rf8+ Ke2 Re8 Kd3
Rf8 Kd2 Re8 Re1 Kf7 Rf3+ Kg6 h4 Ra8 Ra1 Re8) -5.88/20 4} 49. Qxg6 {(Qxg6
Rf8 Bxe5 Qe7 Re4 Qf7 Qxf7+ Kxf7 Bc7 Rae8 Rxe8 Kxe8 Kg2 Rf5 Bxb6 Rxg5+ Kf2
Rh5 Bxc5 Rh4 Rd3 Kd8 Bb6+ Kd7 Kg3 Re4 Rxd4 Re3+ Kg2 Rc3 Rg4 Rc2+ Kf3)
+10.21/25 4} Rf8 {(Rf8 Bxe5 Qf7 Qxf7+ Rxf7 g6 Rd7 Kg3 Re7 Bf4 Rc8 d6 Rd7
Be5 Re8 Rg5 Ra8 Kf4 Re8 h4 Rdd8 Rf3 Re6 h5 Rd7 Rf5 Re8 Kg5 Rdd8 h6 gxh6+
Kxh6 Rd7 Rf7) -6.95/21 22} 50. Bxe5 {(Bxe5 Qe7 Re4 Rae8 d6 Qf7 Qxf7+ Rxf7
g6 Rd7 a5 bxa5 Rxa5 b6 Ra3 Rf8 Kg3 Rf1 Ra8+ Rf8 Rxf8+ Kxf8 Rh4 Kg8 Rf4 d3
Rf3 d2 Rd3) +10.62/24 3} Qf7 {(Qf7 Qxf7+ Rxf7 g6 Rd7 Kg3 Re7 Bf4 Rc8 d6 Rd7
Be5 Re8 Rg5 Ra8 Kf4 Re8 h4 Ra8 Rf5 Rad8 Ke4 Re8 Kd3 Ra8 h5 Rad8 Ra1 Re8
Raf1 Rdd8 R1f2) -6.99/20 3} 51. Qxf7+ {(Qxf7+) +11.42/24 6} Rxf7 {(Rxf7 g6
Rd7 Kg3 Re7 Rg5 Rd7 d6 Rad8 Rf3 Re8 Rgf5 Rdd8 Kf4 Re6 Rf7 Rxg6 Rxb7 Rg1 Rg3
Rxg3 Kxg3 d3 Kf3 d2 Ke2 Re8 Rxg7+ Kf8 Rg5 Kf7 Kxd2 Ke6 h4) -7.54/21 5} 52.
g6 {(g6) +11.61/21 3} Re7 {(Re7 Rg5 Rf8 d6 Rd7 Kg3 Ra8 Kf4 Rc8 Rf5 Rcd8 Ke4
Rxd6 Bxd6 Rxd6 Rf7 Rxg6 Rxb7 Re6+ Kd5 Rh6 Rg3 Kf8 Rbxg7 d3 Ke4 d2 Rd7 Re6+
Kf3 Re1 Rxd2 Ra1 Rg6 Rxa4 Rxb6 Rxc4 Rb7 Ke8 Rdd7 Rc3+ Kg4 c4 h4) -7.69/20
4} 53. Rg5 {(Rg5 Rf8 Kg2 Rd7 Rf3 Re8 d6 Re6 Kg3 Rexd6 Bxd6 Rxd6 Rff5 Rf6
Re5 Kf8 Rgf5 Rxf5 Rxf5+ Ke7 Rf7+ Kd6 Kf2 Ke5 Kf3 d3 Ke3 d2 Kxd2 Kd4 Rxg7)
+11.93/24 3} Rf8 {(Rf8 d6 Rd7 Kg3 Ra8 Kf4 Rc8 Rf5 Rcd8 Ke4 Rxd6 Bxd6 Rxd6
Rf7 Rxg6 Rxb7 Re6+ Kd5 Rh6 Rg3 Kf8 Rbxg7 d3 Ke4 d2 Rd7 Re6+ Kf3 Ke8 Rxd2
Rh6 Ke2 Kf7 Re3 Rg6 Rd7+ Kf8 a5 bxa5 Rd5 a4 Rxc5 Rg2+ Kf3 Rb2 Rf5+ Kg7 Ke4)
-8.65/21 8} 54. Kg2 {(Kg2 Rd8 Rf3 d3 Rxd3 Kf8 Rf3+ Ke8 d6 Re6 Kg3 Kd7 Rf7+
Kc8 Rxg7 Ree8 Rc7+ Kb8 Rf7 Ka7 g7 Re6 d7 Rg8 Kf2 Rd8 g8R Rf6+ Bxf6 Rxd7
Ra8+ Kxa8 Rxd7 Ka7 Be5) +12.20/27 5} Rd8 {(Rd8 Rf3 Rde8 d6 Re6 Kg3 Rd8 Rf7
Rxg6 Rxg7+ Rxg7 Rxg7+ Kf8 Rh7 d3 Kf2 Ke8 Rxb7 Kf8 Rxb6 Re8 Bf6 d2 Be7+ Kf7
Ke2 Ra8 Ra6 Rg8 Kxd2 Ke6 Rc6 Rg2+ Ke3 Ra2 Rxc5 Rxa4 Kd4 Ra1 b6 Rd1+ Kc3
Rc1+ Kb4 Rb1+ Ka5 Rb3 h4 Kd7 Rc7+ Ke6) -10.22/21 9} 55. Rf3 {(Rf3)
+12.55/29 5} Rdd7 {(Rdd7 Rgf5 Rd8 Kh2 Ree8 Bc7 Ra8 Bxb6 d3 Rxd3 Rab8 d6
Re2+ Kg3 Re6 Bxc5 Rxg6+ Kf4 Rd8 d7 Re6 Rf8+ Rxf8+ Bxf8 Rf6+ Kg5 Rxf8 d8Q
Rxd8 Rxd8+ Kf7 Rd7+ Ke8 Rxg7 b6 Rb7 Kd8 Rxb6 Ke7 h4 Kd7 a5 Kc7 h5 Kd7 a6)
-13.11/21 18} 56. Rh5 {(Rh5 Rf7 Rxf7 Rxf7 Rh8+ Kxh8 gxf7 Kh7 f8Q Kg6 Qxg7+
Kf5 Kf3 d3 Qf6+) +M8/57 3} Rf7 {(Rf7 Rxf7 Rxf7 Rh8+ Kxh8 gxf7 Kh7 f8Q Kg6
Qxg7+ Kf5 Kf3 d3 Qf6+) -M7/23 2} 57. Rxf7 {(Rxf7 Rxf7 Rh8+ Kxh8 gxf7 Kh7
f8Q Kg6 Qxg7+ Kf5 Kf3 d3 Qf6+) +M7/76 3} Rxf7 {(Rxf7 Rh8+ Kxh8 gxf7 Kh7 f8Q
Kg6 Qxg7+ Kf5 Kf3 d3 Qf6+) -M6/22 0} 58. Rh8+ {(Rh8+ Kxh8 gxf7 Kh7 f8Q Kg6
Qxg7+ Kf5 Kf3 d3 Qf6+) +M6/158 3} Kxh8 {(Kxh8 gxf7 Kh7 f8Q Kg6 Qxg7+ Kf5
Kf3 d3 Qf6+) -M5/1 0} 59. gxf7 {(gxf7 Kh7 f8Q Kg6 Qxg7+ Kf5 Kf3 d3 Qf6+)
+M5/245 1} Kh7 {(Kh7 f8Q Kg6 Qxg7+ Kf5 Kf3 d3 Qf6+) -M4/18 0} 60. f8=Q
{(f8Q Kg6 Qxg7+ Kf5 Kf3 d3 Qf6+) +M4/245 1} Kh6 {(Kh6 Qxg7+ Kh5 Bf4 d3)
-M3/16 0} 61. Qxg7+ {(Qxg7+ Kh5 Bf6 d3 Qg5+) +M3/245 1} Kh5 {(Kh5 Bf4 d3)
-M2/1 0} 62. Bf6 {(Bf6 d3 Qg5+) +M2/245 1} d3 {(d3) -M1/1} 63. Qg5# {(Qg5+)
+M1/245 1} 1-0
[/pgn]

This is SFNN halfkp (the bigger net architecture) after 300 million trained positions (which are SF eval at depth = 4 MultPV=4, validation data is generated at D=6) against Houdini 1.5a. Both use 8 threads (Houdini 1.5 can't use more) and 4 GB of hash. The pawn game of SF with this very infantile net is already quite nice. This game somehow reminds of Leela getting her first wins against SF with excellent pawn play. This net seems to be a good candidate for self-play. But we'll see. (Note: SF had access to full 6-men and some 7-men tbs because I forgot to turn them off. However, the first time it showed a winning eval it didn't report a tb hit.)

EDIT: Forgot to mention this is a 10 minutes + 2 seconds increment game.

First thing after testing this net more thoroughly I will generate 300 million depth=6 positions with SF's original eval and train with the same command as before and play a match between this and the resulting net to see any potential gain.

Could you share the net that beat H1.5 on google drive or mediafire?

[ChickenLogic]

This is the net that played H 1.5: https://drive.google.com/drive/folders/ ... sp=sharing

[Dann Corbit]

I can recommend pgn-extract for converting .png s to other formats.

It isn't about that. I know those tools, I simply don't know how to convert the data into something the engine can read. When I tell the engine to produce fens they aren't simply a list of fens. They are somehow encoded. The file is a .bin file but it isn't a polyglot book.

Can you post a sample of the data that you are trying to convert?

[ChickenLogic]

I can recommend pgn-extract for converting .png s to other formats.

It isn't about that. I know those tools, I simply don't know how to convert the data into something the engine can read. When I tell the engine to produce fens they aren't simply a list of fens. They are somehow encoded. The file is a .bin file but it isn't a polyglot book.

Can you post a sample of the data that you are trying to convert?

I've got your sample:


It needs to be unzipped for training.

[Dann Corbit]

I do not think I will be able to help you.
I have no idea what format that data is.

[cucumber]

I've got the first 300,000,000 (s)fens based on SF's original eval. For now these are all D=4. The file is roughly 11GB large. If anyone wants it I'll gladly share. I'll try to figure out good settings. I've got the big net to 26 - 27% move accuracy pretty easily with only 150 mil. games. The small one after one iteration is at 31.7%

I'm kinda starting all over again with NN stuff. Any recommended settings for the full run?

Btw, D=4 on 12 threads results in ~ 2 million (s)fens per minute while for D=8 it needs some minutes for 200k. I think we need to distribute training data generation for the big net to work properly. Although I don't know how much more depth will affect the net.

Also, is there a way to split the training data files into smaller ones and then telling the engine to use multiple files?

You can get hundreds of millions of d20+ evals from Fishtest's training data. I have some 92 million.

[Dann Corbit]

Did the fishtest training data come from Bojun Guo's database or were they generated by Fishtest?
Where is the data, and what format is it in?

[cucumber]

Did the fishtest training data come from Bojun Guo's database or were they generated by Fishtest?
Where is the data, and what format is it in?

I did some processing on my end to get into a format that Ethereal can read for some testing I did. It looks like this:

4k2r/rpBnbppp/p1p2n2/3p1b2/3P4/2N1PN1P/PP2BPP1/3RK2R b Kk - 0 12 [0.5] 080
(FEN [Game outcome] Centipawns)

Not sure how to upload a 2.1GB zip file

They were generated by Fishtest. Fishtest's Tom Vijlbrief uploads backups of PGNs here: https://drive.google.com/drive/folders/ ... sp=sharing

There are so many hundreds of millions of positions that it's almost overwhelming to process them. Google Drive's servers don't seem to be able to compress the folder, so you need a utility like rclone to be able to download them. It's kind of funny, but very cool.

[cucumber]

I made a Dropbox for the ~100,000,000 FEN + game result + evals that I have. Here, for anyone who wants them.
https://www.dropbox.com/s/lri6ujmnrtm8o ... D.zip?dl=0

They've been shuffled already, so the FENs aren't in any particular order. Only problem is that the neural network will never see mate, since Fishtest adjudicates games. It might not see the fifty move rule either, but I'm not sure. SF's search is good enough that maybe that isn't an issue. I hope they can still be of some use.

[Dann Corbit]

Is the score relative to the side to move or is it relative to white.
Looks like no depth data.