https://lichess.org/rlAuPlFW#56
GoK (white https://turbowarp.org/148769358
GoK tweaked (black https://turbowarp.org/657300886

meonguy1 wrote:

https://lichess.org/rlAuPlFW#56
GoK (white https://turbowarp.org/148769358
GoK tweaked (black https://turbowarp.org/657300886

OK, some questions: So GoK (original) is winning? What did you tweak in your version? GoK (original) seems to make one mistake at move 9, however I cannot reproduce that at all, Na3 ist not even considered for a second.

ArnoHu wrote:

meonguy1 wrote:

https://lichess.org/rlAuPlFW#56
GoK (white https://turbowarp.org/148769358
GoK tweaked (black https://turbowarp.org/657300886

OK, some questions: So GoK (original) is winning? What did you tweak in your version? GoK (original) seems to make one mistake at move 9, however I cannot reproduce that at all, Na3 ist not even considered for a second.

The opening that is played is the A03 Birdgame. After the sixth move of both sides, White plays on the 7th move c3. Not in TurboWarp this is played by GoK. but at level Easy Ne5, level medium Be3 and Difficult Nc3. So why is c3 played here and Na3 later ?? is unclear.

i think its because of my bad computer performance

meonguy1 wrote:

i think its because of my bad computer performance

I don't think so, as mentioned it does not even consider those moves once on my system, not even during the first milliseconds. Are you sure you didn't have junior mode on in between by accident?

ArnoHu wrote:

meonguy1 wrote:

i think its because of my bad computer performance

I don't think so, as mentioned it does not even consider those moves once on my system, not even during the first milliseconds. Are you sure you didn't have junior mode on in between by accident?

its not on…
heres the retested vers… i think its because of my bad device performance that these tests are bad
https://lichess.org/qZZbKz7O#93
same colours again

meonguy1 wrote:

ArnoHu wrote:

meonguy1 wrote:

i think its because of my bad computer performance

I don't think so, as mentioned it does not even consider those moves once on my system, not even during the first milliseconds. Are you sure you didn't have junior mode on in between by accident?

its not on…
heres the retested vers… i think its because of my bad device performance that these tests are bad
https://lichess.org/qZZbKz7O#93
same colours again

Could you please attach a GoK logfile (press ‘l’)? Thank you!

meonguy1 wrote:

ArnoHu wrote:

meonguy1 wrote:

i think its because of my bad computer performance

I don't think so, as mentioned it does not even consider those moves once on my system, not even during the first milliseconds. Are you sure you didn't have junior mode on in between by accident?

its not on…
heres the retested vers… i think its because of my bad device performance that these tests are bad
https://lichess.org/qZZbKz7O#93
same colours again

Supposing GoK is playing white on TurboWarp, I tried to reproduce this. 4 out of those 5 mistakes where discarded after 3 seconds or less on my system (an average notebook), the 5th one was not even considered for a millisecond.

GoK Medium vs. Bonsai Blue Belt Testgames

An improved GoK version (5-10% speedup) on Scratch 3 / Medium level (white) wins two test games against Bonsai Blue Belt. GoK Medium usually runs at 1/5th of Bonsai Blue Belt's think time during midgame. Before the improvement, GoK Medium at times was struggling when facing Bonsai Blue Belt and also lost some games.

Game #1: https://lichess.org/uRec2WDT
Game #2: https://lichess.org/7Z7AHV8N

ArnoHu wrote:

GoK Medium vs. Bonsai Blue Belt Testgames

An improved GoK version (5-10% speedup) on Scratch 3 / Medium level (white) wins two test games against Bonsai Blue Belt. GoK Medium usually runs at 1/5th of Bonsai Blue Belt's think time during midgame. Before the improvement, GoK Medium at times was struggling when facing Bonsai Blue Belt and also lost some games.

Game #1: https://lichess.org/uRec2WDT
Game #2: https://lichess.org/7Z7AHV8N

Good job Arno. A chess program is never finished.

Grijfland wrote:

ArnoHu wrote:

GoK Medium vs. Bonsai Blue Belt Testgames

An improved GoK version (5-10% speedup) on Scratch 3 / Medium level (white) wins two test games against Bonsai Blue Belt. GoK Medium usually runs at 1/5th of Bonsai Blue Belt's think time during midgame. Before the improvement, GoK Medium at times was struggling when facing Bonsai Blue Belt and also lost some games.

Game #1: https://lichess.org/uRec2WDT
Game #2: https://lichess.org/7Z7AHV8N

Good job Arno. A chess program is never finished.

As I fixed another quiescence-evasions issue yesterday, I now ran a full 10 game comparison, with the following result:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

7.5 : 2.5

Game #1: Draw (29 moves): https://lichess.org/RM7aP5hx (phew, good luck GoK has draw-by-repetition detection built in for full search - Bonsai only made that one mistake)
Game #2: GoK wins in 69 moves: https://lichess.org/E0zJGnzh
Game #3: GoK wins in 28 moves: https://lichess.org/R2cmrD9f
Game #4: Bonsai wins in 56 moves: https://lichess.org/laNv6FVZ
Game #5: GoK wins in 59 moves: https://lichess.org/Fszttjq7
Game #6: GoK wins in 67 moves: https://lichess.org/DAMhcQdS
Game #7: GoK wins in 75 moves: https://lichess.org/OxMs3Uzd (talk about a comeback!)
Game #8: Bonsai wins in 66 moves: https://lichess.org/lBrCsXEz
Game #9: GoK wins in 37 moves: https://lichess.org/2YWVT2Um
Game #10: GoK wins in 19 moves: https://lichess.org/oi3xIGDM (short game with nice checkmate sequence)

I replayed all 11 blunders GoK did in its 2 losses. Not a single one of them happened on TurboWarp. So clearly those were all search-depth-related, due to the slowness of Scratch 3 compared to TurboWarp.

ArnoHu wrote:

Kochy-Richter Checkmate

With the recent quiescent-checks implementation and several tunings, GoK Chess now finds the Kochy-Richter checkmate sequence in 0.2 seconds on TurboWarp and in 2.8 seconds on Scratch 3. You can try by playing white, and then importing the following FEN data: 1k6/p5pp/1p1q1p2/2p5/3P1nPP/Q1P2PK1/P3r3/6RR b - - 0 1

And…. we are down to 0.08 seconds on TurboWarp and 0.8 seconds on Scratch 3.

ArnoHu wrote:

Grijfland wrote:

ArnoHu wrote:

GoK Medium vs. Bonsai Blue Belt Testgames

An improved GoK version (5-10% speedup) on Scratch 3 / Medium level (white) wins two test games against Bonsai Blue Belt. GoK Medium usually runs at 1/5th of Bonsai Blue Belt's think time during midgame. Before the improvement, GoK Medium at times was struggling when facing Bonsai Blue Belt and also lost some games.

Game #1: https://lichess.org/uRec2WDT
Game #2: https://lichess.org/7Z7AHV8N

Good job Arno. A chess program is never finished.

As I fixed another quiescence-evasions issue yesterday, I now ran a full 10 game comparison, with the following result:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

7.5 : 2.5

Game #1: Draw (29 moves): https://lichess.org/RM7aP5hx (phew, good luck GoK has draw-by-repetition detection built in for full search - Bonsai only made that one mistake)
Game #2: GoK wins in 69 moves: https://lichess.org/E0zJGnzh
Game #3: GoK wins in 28 moves: https://lichess.org/R2cmrD9f
Game #4: Bonsai wins in 56 moves: https://lichess.org/laNv6FVZ
Game #5: GoK wins in 59 moves: https://lichess.org/Fszttjq7
Game #6: GoK wins in 67 moves: https://lichess.org/DAMhcQdS
Game #7: GoK wins in 75 moves: https://lichess.org/OxMs3Uzd (talk about a comeback!)
Game #8: Bonsai wins in 66 moves: https://lichess.org/lBrCsXEz
Game #9: GoK wins in 37 moves: https://lichess.org/2YWVT2Um
Game #10: GoK wins in 19 moves: https://lichess.org/oi3xIGDM (short game with nice checkmate sequence)

I replayed all 11 blunders GoK did in its 2 losses. Not a single one of them happened on TurboWarp. So clearly those were all search-depth-related, due to the slowness of Scratch 3 compared to TurboWarp.

Some more tuning and bugfixing for GoK 3.75, with the following result:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

4 : 0

Game #1: GoK wins in 47 moves: https://lichess.org/53jxc5D5
Game #2: GoK wins in 59 moves: https://lichess.org/OOk2HKIm
Game #3: GoK wins in 101 moves: https://lichess.org/XTtSubij
Game #4: GoK wins in 51 moves: https://lichess.org/9EaYul15

ArnoHu wrote:

ArnoHu wrote:

Grijfland wrote:

ArnoHu wrote:

GoK Medium vs. Bonsai Blue Belt Testgames

An improved GoK version (5-10% speedup) on Scratch 3 / Medium level (white) wins two test games against Bonsai Blue Belt. GoK Medium usually runs at 1/5th of Bonsai Blue Belt's think time during midgame. Before the improvement, GoK Medium at times was struggling when facing Bonsai Blue Belt and also lost some games.

Game #1: https://lichess.org/uRec2WDT
Game #2: https://lichess.org/7Z7AHV8N

Good job Arno. A chess program is never finished.

As I fixed another quiescence-evasions issue yesterday, I now ran a full 10 game comparison, with the following result:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

7.5 : 2.5

Game #1: Draw (29 moves): https://lichess.org/RM7aP5hx (phew, good luck GoK has draw-by-repetition detection built in for full search - Bonsai only made that one mistake)
Game #2: GoK wins in 69 moves: https://lichess.org/E0zJGnzh
Game #3: GoK wins in 28 moves: https://lichess.org/R2cmrD9f
Game #4: Bonsai wins in 56 moves: https://lichess.org/laNv6FVZ
Game #5: GoK wins in 59 moves: https://lichess.org/Fszttjq7
Game #6: GoK wins in 67 moves: https://lichess.org/DAMhcQdS
Game #7: GoK wins in 75 moves: https://lichess.org/OxMs3Uzd (talk about a comeback!)
Game #8: Bonsai wins in 66 moves: https://lichess.org/lBrCsXEz
Game #9: GoK wins in 37 moves: https://lichess.org/2YWVT2Um
Game #10: GoK wins in 19 moves: https://lichess.org/oi3xIGDM (short game with nice checkmate sequence)

I replayed all 11 blunders GoK did in its 2 losses. Not a single one of them happened on TurboWarp. So clearly those were all search-depth-related, due to the slowness of Scratch 3 compared to TurboWarp.

Some more tuning and bugfixing for GoK 3.75, with the following result:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

4 : 0

Game #1: GoK wins in 47 moves: https://lichess.org/53jxc5D5
Game #2: GoK wins in 59 moves: https://lichess.org/OOk2HKIm
Game #3: GoK wins in 101 moves: https://lichess.org/XTtSubij
Game #4: GoK wins in 51 moves: https://lichess.org/9EaYul15

The series continues:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

7 :1

Game #5: GoK wins in 36 moves: https://lichess.org/3f0Xh16M
Game #6: Draw after 80 moves: https://lichess.org/hUUU6m01
Game #7: GoK wins in 34 moves: https://lichess.org/9MW6QPvt
Game #8: Draw after 46 moves: https://lichess.org/tDx43VQQ

It goes wrong on move 21 … Qc5b6. GoK does not see the attack on the King's position.
TurboWarp / Difficult 40sec.
GRY = Grijfland

PGN (Game Protocol):

Event “Test GoK”
Date “2022.03.30”
Round “05”
White “GRY”
Black “Chess GoK”
Result “1-0”
ECO “C30 King's Gambit, General”

1. e2e4 e7e5 2. f2f4 e5xf4 3. Nb1c3 Qd8h4 4. Ke1e2 Nb8c6 5. Ng1f3 Qh4g4 6. d2d4 Bf8b4 7. Nc3d5 Bb4d6 8. e4e5 Ng8e7 9. c2c4 Ne7xd5 10. c4xd5 Nc6xe5 11. d4xe5 Bd6xe5 12. Ke2f2 O-O 13. Bf1d3 Be5d6 14. Qd1c2 Qg4h5 15. Qc2c4 b7b6 16. Bc1xf4 b6b5 17. Qc4d4 Bd6xf4 18. Qd4xf4 Qh5xd5 19. Rh1d1 Qd5c5 20. Kf2f1 c7c6 21. Ra1c1 Qc5b6 22. Rd1e1 d7d5 23. Re1e7 Bc8e6 24. Re7c7 Ra8c8 25. Rc1xc6 Qb6xc6 26. Rc7xc6 Rc8xc6 27. Nf3d4 Rc6a6 28. a2a3 Be6d7 29. Qf4g5 Rf8e8 30. Kf1g1 Re8e1 31. Kg1f2 Re1e8 32. Qg5xd5 Ra6f6 33. Kf2g3 Re8e3 34. Nd4f3 Bd7c6 35. Qd5d4 Rf6e6 36. Kg3f2 h7h6 37. Qd4xe3 Re6xe3 38. Kf2xe3 a7a6 39. Ke3d4 g7g5 40. Kd4c5 Bc6b7 41. Bd3f5 Kg8g7 42. Kc5b6 Bb7d5 43. Kb6xa6 Kg7f6 44. Bf5d7 b5b4 45. a3xb4 h6h5 46. Ka6b6 g5g4 47. Nf3e1 Bd5e6 48. Bd7xe6 f7xe6 49. b4b5 h5h4 50. Kb6c6 Kf6f5 51. b5b6 g4g3 52. h2xg3 Kf5g4 53. b6b7 Kg4xg3 54. b7b8 Kg3f2 55. Qb8e5 Kf2g1 56. Qe5e3 Kg1f1 57. Ne1d3 Kf1xg2 58. Nd3f4 Kg2f1 59. Qe3e2 Kf1g1 60. Qe2g2# 1-0

On move 9, GoK doubts between e5 or Na5, but Ne7 plays. The endgame with pawns is played badly by GoK.
Why on move 38 …. Rxa7 ? After that, things go really bad for GoK.
You can see from the evaluation column whether GoK correctly assesses the situation on the chessboard.

TurboWarp / Difficult / 40 sec.

PGN (Game Protocol):

Event “Test GoK”
Date “2022.03.31”
Round “06”
White “GRY”
Black “Chess GoK”
Result “1-0”
ECO “A90 Dutch Defense, Classical Variation”

1. d2d4 f7f5 2. c2c4 Ng8f6 3. g2g3 e7e6 4. Bf1g2 Nf6e4 5. Nb1d2 Bf8b4 6. a2a3 Bb4xd2 7. Bc1xd2 O-O 8. Bd2f4 Nb8c6 9. d4d5 Nc6e7 10. f2f3 Ne4f6 11. d5d6 c7xd6 12. Bf4xd6 Nf6e8 13. f3f4 Ne8xd6 14. Qd1xd6 Ne7c6 15. Ra1d1 Qd8b6 16. b2b4 Qb6e3 17. Qd6d3 Qe3xd3 18. Rd1xd3 a7a6 19. Ng1f3 b7b5 20. c4xb5 a6xb5 21. Nf3d4 Bc8b7 22. Ke1f2 Rf8c8 23. Nd4xb5 Bb7a6 24. Nb5d6 Ba6xd3 25. Nd6xc8 Bd3e4 26. Bg2xe4 f5xe4 27. b4b5 Ra8xc8 28. b5xc6 Rc8xc6 29. a3a4 d7d5 30. a4a5 Kg8f7 31. Rh1a1 e6e5 32. a5a6 Rc6c8 33. a6a7 Rc8a8 34. Kf2e3 Kf7e6 35. Ra1a6 Ke6d7 36. f4xe5 Kd7c7 37. Ke3d4 Kc7b7 38. Ra6a3 Ra8xa7 39. Ra3xa7 Kb7c8 40. Kd4xd5 h7h5 41. e5e6 Kc8b8 42. Ra7d7 Kb8c8 43. Kd5d6 g7g6 44. e6e7 e4e3 45. e8=Q+# 1-0

ArnoHu wrote:

ArnoHu wrote:

ArnoHu wrote:

Grijfland wrote:

ArnoHu wrote:

GoK Medium vs. Bonsai Blue Belt Testgames

An improved GoK version (5-10% speedup) on Scratch 3 / Medium level (white) wins two test games against Bonsai Blue Belt. GoK Medium usually runs at 1/5th of Bonsai Blue Belt's think time during midgame. Before the improvement, GoK Medium at times was struggling when facing Bonsai Blue Belt and also lost some games.

Game #1: https://lichess.org/uRec2WDT
Game #2: https://lichess.org/7Z7AHV8N

Good job Arno. A chess program is never finished.

As I fixed another quiescence-evasions issue yesterday, I now ran a full 10 game comparison, with the following result:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

7.5 : 2.5

Game #1: Draw (29 moves): https://lichess.org/RM7aP5hx (phew, good luck GoK has draw-by-repetition detection built in for full search - Bonsai only made that one mistake)
Game #2: GoK wins in 69 moves: https://lichess.org/E0zJGnzh
Game #3: GoK wins in 28 moves: https://lichess.org/R2cmrD9f
Game #4: Bonsai wins in 56 moves: https://lichess.org/laNv6FVZ
Game #5: GoK wins in 59 moves: https://lichess.org/Fszttjq7
Game #6: GoK wins in 67 moves: https://lichess.org/DAMhcQdS
Game #7: GoK wins in 75 moves: https://lichess.org/OxMs3Uzd (talk about a comeback!)
Game #8: Bonsai wins in 66 moves: https://lichess.org/lBrCsXEz
Game #9: GoK wins in 37 moves: https://lichess.org/2YWVT2Um
Game #10: GoK wins in 19 moves: https://lichess.org/oi3xIGDM (short game with nice checkmate sequence)

I replayed all 11 blunders GoK did in its 2 losses. Not a single one of them happened on TurboWarp. So clearly those were all search-depth-related, due to the slowness of Scratch 3 compared to TurboWarp.

Some more tuning and bugfixing for GoK 3.75, with the following result:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

4 : 0

Game #1: GoK wins in 47 moves: https://lichess.org/53jxc5D5
Game #2: GoK wins in 59 moves: https://lichess.org/OOk2HKIm
Game #3: GoK wins in 101 moves: https://lichess.org/XTtSubij
Game #4: GoK wins in 51 moves: https://lichess.org/9EaYul15

The series continues:

GoK Medium (Scratch 3, white) vs. Bonsai Blue Belt (TurboWarp, black):

7 :1

Game #5: GoK wins in 36 moves: https://lichess.org/3f0Xh16M
Game #6: Draw after 80 moves: https://lichess.org/hUUU6m01
Game #7: GoK wins in 34 moves: https://lichess.org/9MW6QPvt
Game #8: Draw after 46 moves: https://lichess.org/tDx43VQQ

GoK Medium (Scratch 3) vs. Bonsai Blue Belt (TurboWarp) ended 8 : 2
GoK Difficult (Scratch 3) vs. Bonsai Blue Belt (TurboWarp) ended 9,5 : 0,5

Grijfland wrote:

It goes wrong on move 21 … Qc5b6. GoK does not see the attack on the King's position.
TurboWarp / Difficult 40sec.
GRY = Grijfland

PGN (Game Protocol):

Event “Test GoK”
Date “2022.03.30”
Round “05”
White “GRY”
Black “Chess GoK”
Result “1-0”
ECO “C30 King's Gambit, General”

1. e2e4 e7e5 2. f2f4 e5xf4 3. Nb1c3 Qd8h4 4. Ke1e2 Nb8c6 5. Ng1f3 Qh4g4 6. d2d4 Bf8b4 7. Nc3d5 Bb4d6 8. e4e5 Ng8e7 9. c2c4 Ne7xd5 10. c4xd5 Nc6xe5 11. d4xe5 Bd6xe5 12. Ke2f2 O-O 13. Bf1d3 Be5d6 14. Qd1c2 Qg4h5 15. Qc2c4 b7b6 16. Bc1xf4 b6b5 17. Qc4d4 Bd6xf4 18. Qd4xf4 Qh5xd5 19. Rh1d1 Qd5c5 20. Kf2f1 c7c6 21. Ra1c1 Qc5b6 22. Rd1e1 d7d5 23. Re1e7 Bc8e6 24. Re7c7 Ra8c8 25. Rc1xc6 Qb6xc6 26. Rc7xc6 Rc8xc6 27. Nf3d4 Rc6a6 28. a2a3 Be6d7 29. Qf4g5 Rf8e8 30. Kf1g1 Re8e1 31. Kg1f2 Re1e8 32. Qg5xd5 Ra6f6 33. Kf2g3 Re8e3 34. Nd4f3 Bd7c6 35. Qd5d4 Rf6e6 36. Kg3f2 h7h6 37. Qd4xe3 Re6xe3 38. Kf2xe3 a7a6 39. Ke3d4 g7g5 40. Kd4c5 Bc6b7 41. Bd3f5 Kg8g7 42. Kc5b6 Bb7d5 43. Kb6xa6 Kg7f6 44. Bf5d7 b5b4 45. a3xb4 h6h5 46. Ka6b6 g5g4 47. Nf3e1 Bd5e6 48. Bd7xe6 f7xe6 49. b4b5 h5h4 50. Kb6c6 Kf6f5 51. b5b6 g4g3 52. h2xg3 Kf5g4 53. b6b7 Kg4xg3 54. b7b8 Kg3f2 55. Qb8e5 Kf2g1 56. Qe5e3 Kg1f1 57. Ne1d3 Kf1xg2 58. Nd3f4 Kg2f1 59. Qe3e2 Kf1g1 60. Qe2g2# 1-0

Thank you! Today I uploaded a new version, with which the two blunders at move 7 and move 15 won't happen any more. Not sure whether there is a direct causation though, I tweaked several parts of evaluation. I think at move 21 the game is lost, although I admit that move is suboptimal too.

hi i am inviting a CCC Tournament
in this Tournament all engines are play against each other and find out who is the best one among them
it was 5 years and lets see how many engines have become stronger and who is going to be winner
the winner
1st to 3rd winner's engines will be translate to java and these engines will be going to challenge other engines like Lc0 , stockfish, and chess.com
if your engine beat chess.com engine 25 (finale level) your engine will be announced to whole over the world and you will be part of the strongest engines or neutral engines
so be ready this is your chance to announce your chess engine to the world

MEGATRON132358 wrote:

hi i am inviting a CCC Tournament
in this Tournament all engines are play against each other and find out who is the best one among them
it was 5 years and lets see how many engines have become stronger and who is going to be winner
the winner
1st to 3rd winner's engines will be translate to java and these engines will be going to challenge other engines like Lc0 , stockfish, and chess.com
if your engine beat chess.com engine 25 (finale level) your engine will be announced to whole over the world and you will be part of the strongest engines or neutral engines
so be ready this is your chance to announce your chess engine to the world

a) We already know the strongest Scratch engine is GoK
b) GoK will not beat Lc0 or Stockfish
c) The chess.com engine is the same as Stockfish
d) We can announce our engines to the world just as well as you can