Longest against Leela

Discussion of chess software programming and technical issues.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
smcracraft
Posts: 706
Joined: Wed Mar 08, 2006 7:08 pm
Location: Orange County California
Full name: Stuart Cracraft
Contact:

Longest against Leela

Post by smcracraft » Sat Jan 18, 2020 7:35 am

While I am quite fond of its slow positional style, I am not too happy with its ability to aggressively
attack after gaining both significant positional and material advantage.

It leads to a somewhat boring game generally speaking.

I continue to think that the descendent will re-emphasize the attacking spirit more forcefully.

--Stuart

[Event "banksia game"]
[Date "2020.01.17"]
[White "Lc0 v0.24.0-dev+git.108c11b"]
[Black "hm::Human"]
[Result "1-0"]
[TimeControl "40/3600+1"]
[Time "23:30:47"]
[Termination "mate"]
[ECO "D20"]
[Opening "QGA"]

1. d4 {+0.24/9 63.6 43688} d5 2. c4 {+0.28/9 62.0 47193} dxc4
3. e4 {+0.42/12 70.1 38977; D20: QGA, 3.e4} e6 4. Bxc4 {+0.62/14 108.3 72779} Bb4+
5. Nc3 {+0.87/12 66.6 33784} Bxc3+ 6. bxc3 {+1.84/10 75.8 45199} Ne7
7. Nf3 {+1.93/9 116.6 91041} O-O 8. O-O {+1.90/9 73.9 75284} Nbc6
9. h4 {+2.07/9 158.9 90375} a6 10. Re1 {+2.42/8 87.2 52916} b5
11. Bf1 {+2.44/8 95.0 55469} Bb7 12. a4 {+2.93/7 56.5 29828} bxa4
13. Qxa4 {+3.45/7 125.3 67788} Qc8 14. h5 {+4.54/6 113.0 68265} Kh8
15. h6 {+5.90/6 88.8 48281} g6 16. Qa3 {+6.83/6 93.8 49562} a5
17. Qc5 {+6.91/6 90.8 48569} f6 18. Bb5 {+7.11/6 90.6 52699} Qb8
19. d5 {+7.58/6 91.5 48287} Qa7 20. Qa3 {+5.51/6 86.1 43342} Rfe8
21. Bxc6 {+7.11/6 95.6 46799} Bxc6 22. dxc6 {+6.24/6 73.6 38697} Nxc6
23. Qa2 {+6.49/6 108.4 59455} Qc5 24. e5 {+6.66/6 91.1 49467} f5
25. Qa4 {+6.91/6 90.8 49674} Red8 26. Qh4 {+13.30/6 79.6 44846} Qe7
27. Qxe7 {+14.05/5 81.9 54310} Nxe7 28. Ng5 {+6.41/6 110.7 76936} Kg8
29. Nxe6 {+5.82/6 91.9 59168} Rd7 30. Nc5 {+6.02/6 90.7 51157} Rd5
31. Ba3 {+5.70/6 90.5 48974} Nc6 32. Ne6 {+6.01/6 91.0 51856} Rxe5
33. Ng5 {+6.31/6 91.4 57546} Rae8 34. Rxe5 {+7.41/6 67.9 42862} Rxe5
35. f4 {+6.21/6 113.0 79764} Re4 36. Rd1 {+7.20/6 90.3 69044} Re8
37. Rd7 {+9.73/5 76.2 71007} Re1+ 38. Kf2 {+128.00/2 106.7 185058857} Re7
39. Bxe7 {+128.00/2 45.5 217127} Nxe7 40. Rd8# {+128.00/4 126.2 315684} 1-0

User avatar
Laskos
Posts: 10198
Joined: Wed Jul 26, 2006 8:21 pm
Full name: Kai Laskos

Re: Longest against Leela

Post by Laskos » Sat Jan 18, 2020 10:52 am

smcracraft wrote:
Sat Jan 18, 2020 7:35 am
While I am quite fond of its slow positional style, I am not too happy with its ability to aggressively
attack after gaining both significant positional and material advantage.

It leads to a somewhat boring game generally speaking.

I continue to think that the descendent will re-emphasize the attacking spirit more forcefully.

--Stuart

[Event "banksia game"]
[Date "2020.01.17"]
[White "Lc0 v0.24.0-dev+git.108c11b"]
[Black "hm::Human"]
[Result "1-0"]
[TimeControl "40/3600+1"]
[Time "23:30:47"]
[Termination "mate"]
[ECO "D20"]
[Opening "QGA"]

1. d4 {+0.24/9 63.6 43688} d5 2. c4 {+0.28/9 62.0 47193} dxc4
3. e4 {+0.42/12 70.1 38977; D20: QGA, 3.e4} e6 4. Bxc4 {+0.62/14 108.3 72779} Bb4+
5. Nc3 {+0.87/12 66.6 33784} Bxc3+ 6. bxc3 {+1.84/10 75.8 45199} Ne7
7. Nf3 {+1.93/9 116.6 91041} O-O 8. O-O {+1.90/9 73.9 75284} Nbc6
9. h4 {+2.07/9 158.9 90375} a6 10. Re1 {+2.42/8 87.2 52916} b5
11. Bf1 {+2.44/8 95.0 55469} Bb7 12. a4 {+2.93/7 56.5 29828} bxa4
13. Qxa4 {+3.45/7 125.3 67788} Qc8 14. h5 {+4.54/6 113.0 68265} Kh8
15. h6 {+5.90/6 88.8 48281} g6 16. Qa3 {+6.83/6 93.8 49562} a5
17. Qc5 {+6.91/6 90.8 48569} f6 18. Bb5 {+7.11/6 90.6 52699} Qb8
19. d5 {+7.58/6 91.5 48287} Qa7 20. Qa3 {+5.51/6 86.1 43342} Rfe8
21. Bxc6 {+7.11/6 95.6 46799} Bxc6 22. dxc6 {+6.24/6 73.6 38697} Nxc6
23. Qa2 {+6.49/6 108.4 59455} Qc5 24. e5 {+6.66/6 91.1 49467} f5
25. Qa4 {+6.91/6 90.8 49674} Red8 26. Qh4 {+13.30/6 79.6 44846} Qe7
27. Qxe7 {+14.05/5 81.9 54310} Nxe7 28. Ng5 {+6.41/6 110.7 76936} Kg8
29. Nxe6 {+5.82/6 91.9 59168} Rd7 30. Nc5 {+6.02/6 90.7 51157} Rd5
31. Ba3 {+5.70/6 90.5 48974} Nc6 32. Ne6 {+6.01/6 91.0 51856} Rxe5
33. Ng5 {+6.31/6 91.4 57546} Rae8 34. Rxe5 {+7.41/6 67.9 42862} Rxe5
35. f4 {+6.21/6 113.0 79764} Re4 36. Rd1 {+7.20/6 90.3 69044} Re8
37. Rd7 {+9.73/5 76.2 71007} Re1+ 38. Kf2 {+128.00/2 106.7 185058857} Re7
39. Bxe7 {+128.00/2 45.5 217127} Nxe7 40. Rd8# {+128.00/4 126.2 315684} 1-0
That's because the goal in training is the score in many games. Maybe a combination of score and shorter games as a goal might give a more enjoyable engine. In Go it's similar and even simpler, just change the score in games as a goal to territory at the end of the games as a goal. I guess in Go, some hilarious nets can be built which regularly beat top humans by 20-50 points."Humiliators" :lol:.

Post Reply