Chess2u
Would you like to react to this message? Create an account in a few clicks or log in to continue.

Chess2uLog in

Alpha subZero - NN for CPU

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
if there is some vituses show what viruses it is because i dont put in the code any viruses.if there was some viruses i should know about it.This files client.exe libopenblas is official Lc0 release and this engine file asubzero.exe is clean and its to small to inlude some viruses.If you have doubts just scan this with antivitus program.

Cheers.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
@bookmaker wrote:
if there is some vituses show what viruses it is because i dont put in the code any viruses.if there was some viruses i should know about it.This files client.exe libopenblas is official Lc0 release and this engine file asubzero.exe is clean and its to small to inlude some viruses.If you have doubts just scan this with antivitus program.

Cheers.


what do you mean by "small to include viruses"?
how large viruses are you think? gigabytes?
they are usually in order of hundreds of bytes, your subzero has space for 10000 viruses.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
anyway, this thread already has couple of independent arguments that this file looks like malware, is suspicious, and there is also no doubt that author explicitly took actions to hide something there.

the only agrument against it is bookmaker's words "there's no virus, you are all wrong".

i think there has to be stronger evidence if it's clean. and also explanation why bookmaker encrypted the file.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
Look  if i put there some viruses it will be stupid because in chess community is many policamans and  some guys like this Alexa who reverse programs,so it will be very stupid to put there viruses because this will be find.So i am not that stupid to put there viruses LOL.I test this now and as i said subzero is about 20 elo weaker than Stockfish
Alpha subZero - NN for CPU - Page 3 Zrzut_11

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
It is well known that here in this forum one is constantly confronted with viruses. So nothing new. The forum operator consistently bans chess enthusiasts, partly without any reason, and lets such criminals go. And now, please for the sake of the truth, I should be banned. After that I can clean my computer from all these viruses, trojans.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
the truth = trhere is no viruses i changed few bits in header file and thats it , file is clean but modified.which dosnt mean protected and few (4 of 73) AV make false alerts.Here test after 134 games 3+2 10 core  PB ON ,test will be 200 games overall Alpha subZero - NN for CPU - Page 3 Zrzut_12

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
Thank you for compilation.Can you answer me why network for engine, if he Is Stockfish??

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
just read earlier posts everything is explained including motivation and idea also technical things and some options.I answer on all of question.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
@bookmaker wrote:
just read earlier posts everything is explained including motivation and idea also technical things and some options.I answer on all of question.


Is it lc0 or SF ?

the network file is a renamed sergio network. "subzero.gz" is a sergio network
256x20-mod-swa-1467000.pb (Dec 29 2019).

why the exe depends on the client? no logic.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
because sergio network is good and i try to unified it with my engine ,and is of course not ideal but look what you better expect ,it play awesome maybe 10-20 elo behind SF but very strong.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyLet’s see the source

more_horiz
Since you are distributing the program and it’s derived from one or more GPL’d programs, I am formally requesting that you make the code available, as required by the terms of the license.

As a note by someone more than a little familiar with leela:

1. You won’t get more than single digit nps on cpu for a big net. Combining that with SF seems at best primitive.

2. The client executable is a golang program used exclusively for training. It controls lc0 for selfplay, not the other way around. If you run the engine and it launches the client, something suspicious is going on.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
ok i release that code soon ,lthere is three versions available and first one and maybe second one is buggy freezing after some games ,you should use last version,in last version client not runing but it just need to be in the folder.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
don't make the source public, you're not selling it. So you don't have to do it if you don't want. See Eman and all the Stockfish clones! It plays very strong, here in the forum people are just jealous that you made the cpu Leela so strong. Please keep it to yourself. Think it over carefully.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
test 200 games
alpha subzero vs Stockfish dev 070120
3min + 2 sec
permament brain on
Opening suite http://www.rybkachess.com/free/NoomenShortLinesTestsuite.pgn 100 openings playing reversed
3 GB HASH (cache)
10 threads per engine
full 6-man syzygy TB

Alpha subZero - NN for CPU - Page 3 Subpgn10

PGN download: http://www.filedropper.com/subzero-test
Engine download https://www.filemail.com/d/kcpyunivplfeqws

example wins:
[pgn][Event "asubzer-sf dev, Blitz 3m+2s"]
[Site "komp"]
[Date "2020.01.15"]
[Round "185"]
[White "Stockfish 070120 64 POPCNT"]
[Black "alpha subZero 64"]
[Result "1-0"]
[ECO "E04"]
[Annotator "0.35;0.76"]
[PlyCount "135"]
[TimeControl "180+2"]

{Intel(R) Xeon(R) CPU      3.3GHz W=30.1 plies; 10 291kN/s;
3 603 983 TBAs B=14.5 plies; 38kN/s; 2 417 TBAs} 1. d4 Nf6 2. c4 e6 3. g3 d5 4.
Bg2 dxc4 5. Nf3 a6 6. Ne5 c5 {Both last book move} 7. Na3 {[%eval 35,29] [%emt
0:00:37]} Qxd4 {[%eval 76,14] [%emt 0:00:33] (cxd4)} 8. Qa4+ {[%eval 92,24]
[%emt 0:00:04] (Qxd4)} Bd7 {[%eval 79,13] [%emt 0:00:08]} 9. Nxd7 {[%eval 94,
28] [%emt 0:00:00]} Qxd7 {[%eval 90,14] [%emt 0:00:11]} 10. Qxd7+ {[%eval 107,
28] [%emt 0:00:00]} Nbxd7 {[%eval 85,13] [%emt 0:00:03]} 11. Bxb7 {[%eval 98,
25] [%emt 0:00:02]} Ra7 {[%eval 81,13] [%emt 0:00:00]} 12. Bg2 {[%eval 99,26]
[%emt 0:00:06]} c3 {[%eval 109,13] [%emt 0:00:03]} 13. bxc3 {[%eval 107,25]
[%emt 0:00:03]} Nd5 {[%eval 110,14] [%emt 0:00:10] (Be7)} 14. Bd2 {[%eval 97,
23] [%emt 0:00:02]} Be7 {[%eval 100,14] [%emt 0:00:07] (g6)} 15. O-O {[%eval
98,25] [%emt 0:00:08] (Rb1)} O-O {[%eval 88,13] [%emt 0:00:03] (N5b6)} 16. Rab1
{[%eval 83,26] [%emt 0:00:05]} Ne5 {[%eval 95,14] [%emt 0:00:05] (N5b6)} 17. f4
{[%eval 114,27] [%emt 0:00:08] (e4)} c4 {[%eval 141,15] [%emt 0:00:08]} 18.
Bxd5 {[%eval 113,30] [%emt 0:00:00]} Bc5+ {[%eval 119,14] [%emt 0:00:04]} 19.
Kg2 {[%eval 123,31] [%emt 0:00:00]} exd5 {[%eval 114,15] [%emt 0:00:07]} 20.
fxe5 {[%eval 131,32] [%emt 0:00:00]} Bxa3 {[%eval 148,16] [%emt 0:00:11]} 21.
Be3 {[%eval 133,32] [%emt 0:00:00]} Re7 {[%eval 129,15] [%emt 0:00:03]} 22. Bd4
{[%eval 138,31] [%emt 0:00:02]} Re6 {[%eval 137,15] [%emt 0:00:08]} 23. Rb7 {
[%eval 138,33] [%emt 0:00:02]} Be7 {[%eval 129,15] [%emt 0:00:01]} 24. Rd7 {
[%eval 139,33] [%emt 0:00:09]} Rd8 {[%eval 103,15] [%emt 0:00:00]} 25. Rxd8+ {
[%eval 139,31] [%emt 0:00:04]} Bxd8 {[%eval 103,15] [%emt 0:00:02]} 26. Rb1 {
[%eval 145,31] [%emt 0:00:03]} Kf8 {[%eval 153,15] [%emt 0:00:07]} 27. Rb8 {
[%eval 118,34] [%emt 0:00:06]} Ke7 {[%eval 135,16] [%emt 0:00:00]} 28. Rb7+ {
[%eval 139,32] [%emt 0:00:05]} Ke8 {[%eval 113,15] [%emt 0:00:00]} 29. Kf3 {
[%eval 140,34] [%emt 0:00:10]} h5 {[%eval 171,16] [%emt 0:00:23]} 30. e4 {
[%eval 143,37] [%emt 0:00:00]} dxe4+ {[%eval 154,16] [%emt 0:00:03]} 31. Kxe4 {
[%eval 147,34] [%emt 0:00:05]} h4 {[%eval 158,17] [%emt 0:00:00]} 32. g4 {
[%eval 153,33] [%emt 0:00:05]} h3 {[%eval 166,16] [%emt 0:00:00]} 33. a4 {
[%eval 153,34] [%emt 0:00:08]} Rh6 {[%eval 153,17] [%emt 0:00:31]} 34. Kf4 {
[%eval 160,36] [%emt 0:00:00]} Re6 {[%eval 186,16] [%emt 0:00:08]} 35. Ra7 {
[%eval 164,35] [%emt 0:00:01]} Rc6 {[%eval 186,16] [%emt 0:00:07]} 36. Kg3 {
[%eval 142,36] [%emt 0:00:05]} Rh6 {[%eval 186,17] [%emt 0:00:00]} 37. Rb7 {
[%eval 167,32] [%emt 0:00:11]} Ba5 {[%eval 172,16] [%emt 0:00:00] (Bg5)} 38.
Rb8+ {[%eval 168,30] [%emt 0:00:05]} Bd8 {[%eval 180,17] [%emt 0:00:03]} 39. a5
{[%eval 170,33] [%emt 0:00:03] (Rb7)} Kd7 {[%eval 195,14] [%emt 0:00:05]} 40.
Rb7+ {[%eval 170,34] [%emt 0:00:00]} Ke8 {[%eval 210,15] [%emt 0:00:05]} 41.
Bb6 {[%eval 170,37] [%emt 0:00:02]} Bh4+ {[%eval 225,15] [%emt 0:00:04]} 42.
Kf4 {[%eval 171,36] [%emt 0:00:03]} Be1 {[%eval 237,15] [%emt 0:00:10]} 43. Bd4
{[%eval 184,37] [%emt 0:00:00] (Rb8+)} Rg6 {[%eval 276,15] [%emt 0:00:08]} 44.
Rb8+ {[%eval 195,33] [%emt 0:00:00]} Ke7 {[%eval 275,14] [%emt 0:00:06] (Kd7)}
45. Kf5 {[%eval 216,34] [%emt 0:00:24] (Rh8)} Bd2 {[%eval 296,15] [%emt 0:00:
03]} 46. Ke4 {[%eval 226,32] [%emt 0:00:00]} Ke6 {[%eval 270,15] [%emt 0:00:06]
} 47. Kf3 {[%eval 241,31] [%emt 0:00:00]} Ke7 {[%eval 293,15] [%emt 0:00:09]
(Kd7)} 48. Rb7+ {[%eval 247,30] [%emt 0:00:07] (Ke4)} Ke8 {[%eval 289,15]
[%emt 0:00:02]} 49. Rc7 {[%eval 270,31] [%emt 0:00:08]} Be1 {[%eval 269,15]
[%emt 0:00:00] (Bg5)} 50. Rxc4 {[%eval 279,29] [%emt 0:00:07]} Kd7 {[%eval 346,
14] [%emt 0:00:00]} 51. Rb4 {[%eval 284,32] [%emt 0:00:12]} Kc8 {[%eval 351,14]
[%emt 0:00:00]} 52. Rb6 {[%eval 292,32] [%emt 0:00:22] (Rb3)} Re6 {[%eval 318,
14] [%emt 0:00:02]} 53. Rb3 {[%eval 307,27] [%emt 0:00:04]} Rg6 {[%eval 407,14]
[%emt 0:00:05]} 54. Kf4 {[%eval 304,28] [%emt 0:00:07] (Rb6)} Bd2+ {[%eval 301,
13] [%emt 0:00:03] (Kd8)} 55. Kg3 {[%eval 299,27] [%emt 0:00:06] (Kf3)} Rc6 {
[%eval 288,14] [%emt 0:00:02] (Rh6)} 56. Kxh3 {[%eval 354,25] [%emt 0:00:03]}
Bg5 {[%eval 402,13] [%emt 0:00:01]} 57. Kg3 {[%eval 368,25] [%emt 0:00:01]} Bd8
{[%eval 401,13] [%emt 0:00:02]} 58. h4 {[%eval 372,27] [%emt 0:00:03]} Bxa5 {
[%eval 450,14] [%emt 0:00:07]} 59. Kf3 {[%eval 399,28] [%emt 0:00:05] (Kf4)}
Bd8 {[%eval 458,12] [%emt 0:00:02]} 60. h5 {[%eval 397,25] [%emt 0:00:00]} a5 {
[%eval 497,13] [%emt 0:00:06] (Be7)} 61. Rb1 {[%eval 439,26] [%emt 0:00:05]
(Rb2)} Bg5 {[%eval 564,13] [%emt 0:00:05]} 62. Rb5 {[%eval 451,29] [%emt 0:00:
00]} Bd2 {[%eval 583,13] [%emt 0:00:02]} 63. e6 {[%eval 487,28] [%emt 0:00:01]
(Ke4)} Rxe6 {[%eval 522,12] [%emt 0:00:01]} 64. Bxg7 {[%eval 513,30] [%emt 0:
00:06]} a4 {[%eval 591,13] [%emt 0:00:00]} 65. Ra5 {[%eval 549,27] [%emt 0:00:
05]} Re3+ {[%eval 671,13] [%emt 0:00:00]} 66. Kf2 {[%eval 561,25] [%emt 0:00:
02]} Rh3 {[%eval 705,15] [%emt 0:00:05] (Re4)} 67. Kg2 {[%eval 590,23] [%emt 0:
00:04]} Rh4 {[%eval 740,13] [%emt 0:00:00]} 68. Kg3 {[%eval 602,25] [%emt 0:00:
04]} 1-0
[/pgn]
[pgn]
[Event "asubzer-sf dev, Blitz 3m+2s"]
[Site "komp"]
[Date "2020.01.15"]
[Round "186"]
[White "alpha subZero 64"]
[Black "Stockfish 070120 64 POPCNT"]
[Result "1-0"]
[ECO "E04"]
[Annotator "0.65;0.00"]
[PlyCount "109"]
[TimeControl "180+2"]

{Intel(R) Xeon(R) CPU      3.3GHz W=14.0 plies; 38kN/s;
1 532 TBAs B=27.2 plies; 9 522kN/s; 1 996 522 TBAs} 1. d4 Nf6 2. c4 e6 3. g3 d5
4. Bg2 dxc4 5. Nf3 a6 6. Ne5 c5 {Both last book move} 7. Na3 {[%eval 65,14]
[%emt 0:00:21]} Qxd4 {[%eval 0,29] [%emt 0:00:23] (cxd4)} 8. Qa4+ {[%eval 84,
13] [%emt 0:00:03] (Qxd4)} Bd7 {[%eval 68,27] [%emt 0:00:14]} 9. Nxd7 {[%eval
116,14] [%emt 0:00:04]} Qxd7 {[%eval 55,28] [%emt 0:00:08]} 10. Qxd7+ {[%eval
134,14] [%emt 0:00:00]} Nbxd7 {[%eval 64,28] [%emt 0:00:04]} 11. Bxb7 {[%eval
131,14] [%emt 0:00:02]} Ra7 {[%eval 61,25] [%emt 0:00:02]} 12. Bg2 {[%eval 146,
14] [%emt 0:00:02]} c3 {[%eval 53,27] [%emt 0:00:17]} 13. bxc3 {[%eval 114,14]
[%emt 0:00:00]} Be7 {[%eval 58,28] [%emt 0:00:16]} 14. Rb1 {[%eval 133,14]
[%emt 0:00:00] (Bd2)} O-O {[%eval 73,27] [%emt 0:00:26] (Nd5)} 15. Nc4 {
[%eval 164,13] [%emt 0:00:03] (Bd2)} Nd5 {[%eval 131,26] [%emt 0:00:11]} 16.
Kd2 {[%eval 211,14] [%emt 0:00:00]} Bf6 {[%eval 118,26] [%emt 0:00:02] (f5)}
17. Rb3 {[%eval 214,13] [%emt 0:00:05]} Ne5 {[%eval 128,28] [%emt 0:00:00]} 18.
Nxe5 {[%eval 219,14] [%emt 0:00:05]} Bxe5 {[%eval 129,28] [%emt 0:00:00]} 19.
Kc2 {[%eval 220,14] [%emt 0:00:06]} f5 {[%eval 156,31] [%emt 0:00:10]} 20. Rd1
{[%eval 211,15] [%emt 0:00:04]} Rd8 {[%eval 162,27] [%emt 0:00:00]} 21. Ba3 {
[%eval 207,14] [%emt 0:00:10]} c4 {[%eval 147,31] [%emt 0:00:12]} 22. Rb6 {
[%eval 207,15] [%emt 0:00:03]} Kf7 {[%eval 144,29] [%emt 0:00:01]} 23. Rc6 {
[%eval 244,15] [%emt 0:00:04]} g6 {[%eval 158,30] [%emt 0:00:03]} 24. Rxc4 {
[%eval 245,15] [%emt 0:00:01] (Bxd5)} Rdd7 {[%eval 155,26] [%emt 0:00:04]} 25.
Bb2 {[%eval 258,15] [%emt 0:00:00]} Rac7 {[%eval 153,25] [%emt 0:00:04]} 26.
Rxc7 {[%eval 231,15] [%emt 0:00:07]} Bxc7 {[%eval 160,27] [%emt 0:00:00]} 27.
Ba3 {[%eval 234,14] [%emt 0:00:05] (e4)} Be5 {[%eval 146,26] [%emt 0:00:03]}
28. Bxd5 {[%eval 245,14] [%emt 0:00:06]} Rxd5 {[%eval 157,30] [%emt 0:00:06]}
29. Rb1 {[%eval 257,15] [%emt 0:00:03]} Ra5 {[%eval 164,27] [%emt 0:00:00]} 30.
Rb7+ {[%eval 271,13] [%emt 0:00:05]} Kg8 {[%eval 166,29] [%emt 0:00:04] (Ke8)}
31. Rb3 {[%eval 276,14] [%emt 0:00:05]} g5 {[%eval 192,26] [%emt 0:00:00]} 32.
c4 {[%eval 273,14] [%emt 0:00:05]} Ra4 {[%eval 170,28] [%emt 0:00:01]} 33. Kd3
{[%eval 280,14] [%emt 0:00:05]} Ba1 {[%eval 207,30] [%emt 0:00:27] (Kf7)} 34.
Bc1 {[%eval 289,14] [%emt 0:00:12]} e5 {[%eval 203,29] [%emt 0:00:03]} 35. Bxg5
{[%eval 289,14] [%emt 0:00:02]} e4+ {[%eval 211,30] [%emt 0:00:16]} 36. Ke3 {
[%eval 306,16] [%emt 0:00:00]} Rxc4 {[%eval 219,30] [%emt 0:00:06]} 37. Kf4 {
[%eval 310,15] [%emt 0:00:00]} Rc5 {[%eval 232,28] [%emt 0:00:04]} 38. Be7 {
[%eval 313,14] [%emt 0:00:02]} Rc2 {[%eval 232,29] [%emt 0:00:01]} 39. Kxf5 {
[%eval 328,15] [%emt 0:00:07]} Rxe2 {[%eval 238,30] [%emt 0:00:00]} 40. Rb8+ {
[%eval 351,16] [%emt 0:00:19] (Bc5)} Kf7 {[%eval 240,30] [%emt 0:00:03]} 41.
Bc5 {[%eval 353,14] [%emt 0:00:00]} Rxa2 {[%eval 249,27] [%emt 0:00:05]} 42.
Rb7+ {[%eval 360,14] [%emt 0:00:02]} Kg8 {[%eval 251,29] [%emt 0:00:01]} 43.
Kxe4 {[%eval 362,15] [%emt 0:00:13]} Bc3 {[%eval 247,30] [%emt 0:00:00] (Ra5)}
44. h3 {[%eval 361,14] [%emt 0:00:06] (Ra7)} Be1 {[%eval 269,27] [%emt 0:00:19]
(Rb2)} 45. f4 {[%eval 420,13] [%emt 0:00:06]} Ra4+ {[%eval 291,25] [%emt 0:00:
02]} 46. Kf3 {[%eval 479,13] [%emt 0:00:02]} Rc4 {[%eval 332,25] [%emt 0:00:04]
(a5)} 47. Bd6 {[%eval 495,13] [%emt 0:00:04]} a5 {[%eval 377,26] [%emt 0:00:06]
(h6)} 48. Be5 {[%eval 667,13] [%emt 0:00:06]} h6 {[%eval 468,23] [%emt 0:00:02]
(h5)} 49. Rg7+ {[%eval 727,13] [%emt 0:00:04]} Kf8 {[%eval 529,24] [%emt 0:00:
01]} 50. Rg6 {[%eval 761,13] [%emt 0:00:03]} Ra4 {[%eval 577,24] [%emt 0:00:06]
(h5)} 51. g4 {[%eval 882,12] [%emt 0:00:04] (Kg4)} Ra3+ {[%eval 454,17] [%emt
0:00:01]} 52. Kg2 {[%eval 998,13] [%emt 0:00:04]} Rg3+ {[%eval 700,22] [%emt 0:
00:03] (Ra2+)} 53. Kh2 {[%eval 1072,13] [%emt 0:00:05]} Rf3 {[%eval 723,23]
[%emt 0:00:02]} 54. Rxh6 {[%eval 1073,13] [%emt 0:00:02]} Rf2+ {[%eval 791,23]
[%emt 0:00:01]} 55. Kg3 {[%eval 1272,14] [%emt 0:00:26] (Kg1)} 1-0[/pgn]

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
how will you make the engine more powerful? Tell us a little about your next steps. Thank you

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
I also don't care if you release the source code or not, I would just like what I asked you in private, for the sole purpose of understanding the behavior of your interesting experiment.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
@Fire_Fox wrote:
how will you make the engine more powerful? Tell us a little about your next steps. Thank you


In good NN network few nps as dkappe wrote is enough to proceed a good move and because NN and AB variants  is in shared cache then all the variants are searching as wider tree in SMP way,that mean threads overall proceeds  more option to search. To understand this u have to know how SMP working ,it just calculate everything what was not calculated already it means that NN calculation can be extended by A/B way.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyGPL requires publishing source

more_horiz
@Fire_Fox wrote:
don't make the source public, you're not selling it. So you don't have to do it if you don't want. See Eman and all the Stockfish clones!  It plays very strong, here in the forum people are just jealous that you made the cpu Leela so strong. Please keep it to yourself. Think it over carefully.


You are mistaken. The GPLv3 requires you to make source available for any derived work that you distribute, regardless of whether you charge for it.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
dkappe so why fat fritz chessbase not share any source code ,and it is well known that it is based on lc0 ... hmm

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
maybe im stupid but im searching on github fat fritz and i dont find any results

descriptionAlpha subZero - NN for CPU - Page 3 EmptyFat Fritz source

more_horiz
@bookmaker wrote:
dkappe so why fat fritz chessbase not share any source code ,and it is well  known that it is based on lc0 ... hmm


github.com/DanielUranga/lc0/tree/uci-wdl-reporting

descriptionAlpha subZero - NN for CPU - Page 3 EmptyFat fritz source

more_horiz
@bookmaker wrote:
maybe im stupid but im searching on github fat fritz and i dont find  any results


github.com/DanielUranga/lc0/tree/uci-wdl-reporting

If you ask chessbase for the source, that’s where they’ll point you.

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
but GPL v3 licence not does not specify when in terms of time source code must be published.So i can do this now or in few months because in GPL licence there is no regulation about terms of time.
if there is show where.
i can point you to some theory  modifications idea https://github.com/nodchip/Stockfish

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
[quote="bookmaker"]dkappe so why fat fritz chessbase not share any source code ,and it is well  known that it is based on lc0 ... hmm [/quote] that's right, they just try to encourage you with such statements to publish the source. I wouldn't do it!

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
You don't sell it, so no need to publish, even if you do, I'm not sure if you have to do.
And you can do the same as chessbase. Point anyone who wants the source to the Stockfish and LC0 page

descriptionAlpha subZero - NN for CPU - Page 3 EmptyRe: Alpha subZero - NN for CPU

more_horiz
Permissions in this forum:
You cannot reply to topics in this forum