Author: David Dahlem
Date: 16:43:26 11/20/00
Go up one level in this thread
On November 20, 2000 at 17:19:23, Dan Homan wrote:
>On November 20, 2000 at 17:04:37, David Dahlem wrote:
>
>>
>>>
>>>>also it seems if the program uses losses to tune it;s eval, then it can be
>>>>misinterpreted.
>>>>for instance, result from a game that was played in 5 min/game, will have lesser
>>>>values of importance then a game that was played at 40/40 or 40/60.
>>>>
>>>>
>>>>so it can be interpreted that an exchess version that plays only 5min blitz will
>>>>have one kind of tuned eval, while the other one that plays mainly 40/40 will
>>>>differant. Which one is best? :)
>>>>
>>>
>>>Yes, the learning results from 5 min blitz games might be quite different from
>>>40/40. One test I did was 200 games with crafty at 1 min bullet. EXchess lost
>>>a large percentage of the games in the first 100, but did much better in the
>>>second 100 (EXchess still clearly lost the match, but the results were
>>>significantly closer than the first 100 games). However, when I then matched
>>>this version of EXchess against GNUchess at longer time controls (5 min blitz, I
>>>think), my results were not better (and actually a bit worse) than before the 1
>>>min bullet games against crafty.
>>>
>>> - Dan
>>>
>>Along the same line, I have a question. Will losing to a much stronger engine
>>such as Fritz6 or losing to a program weaker than Fritz6, cause EXchess to
>>learn much faster or supposedly get stronger quicker?
>>
>>Dave
>
>I am not sure. Try it an see. I seem to remember something from the knightcap
>group saying that learning was most efficient from games with similar strength
>opponents.
>
> - Dan
Thanks for this info. I am very enthused about your program. By the way, in the
third game between EXchess and Fritz 6, EXchess managed to win. I thought, "Boy,
that is really learning fast. :-) But it lost 20 straight after this. Here is
the game just for your curiosity -
[Event "Blitz:10'"]
[Site "?"]
[Date "2000.11.19"]
[Round "2"]
[White "ExChess 4.01"]
[Black "Fritz 6"]
[Result "1-0"]
[ECO "C42"]
[PlyCount "169"]
{W=7.7 ply, B=8.7ply} 1. e4 {0.01/1 2} 1... e5 {0.09/10 16} 2. Nf3 { (Nc3)
0.01/1 1} 2... Nf6 {0.12/10 31} 3. Nxe5 {0.01/1 1} 3... Qe7 {0.00/9 21} 4. d4
{0.13/9 15} 4... d6 {0.19/9 28} 5. Nf3 {0.20/8 15} 5... Qxe4+ { (Nxe4) 0.28/8
14} 6. Be2 {0.15/8 15} 6... Be7 {(Bf5) 0.34/9 29} 7. Nc3 { 0.31/8 14} 7... Qf5
{(Qg4) 0.28/9 12} 8. O-O {0.28/8 14} 8... O-O {0.44/9 13} 9. Nb5 {(d5) 0.33/7
14} 9... Na6 {(Qd7) 0.00/9 12} 10. c4 {0.28/7 13} 10... Re8 {0.19/9 17} 11. Re1
{(d5) 0.22/8 13} 11... c6 {0.19/8 9} 12. Nc3 {0.26/7 13} 12... Nc7 {(d5) 0.00/8
9} 13. Be3 {(b4) 0.24/7 12} 13... d5 {0.09/8 8} 14. Ne5 {0.20/7 12} 14... dxc4
{(Qe6) 0.12/8 25} 15. Bxc4 {0.17/7 12} 15... Nfd5 { (Ncd5) 0.09/8 15} 16. Nxd5
{(Bd3) 0.16/7 11} 16... Nxd5 {0.12/8 4} 17. Bd2 { (Bxd5) 0.19/7 11} 17... Be6
{(Bh4) -0.09/8 10} 18. Bd3 {(Qb3) 0.19/8 11} 18... Qf6 {-0.16/8 10} 19. Qc2
{0.23/7 11} 19... h6 {(g6) -0.22/8 10} 20. Bh7+ { (Re2) 0.25/7 10} 20... Kh8
{-0.34/9 8} 21. Be4 {0.27/7 10} 21... Bd6 { -0.31/8 22} 22. Nc4 {(Bf3) 0.09/7
11} 22... Bc7 {-0.44/8 14} 23. Qd3 { (Ba5) 0.00/7 10} 23... Rad8 {(Nf4) -0.66/7
6} 24. Na5 {-0.24/7 9} 24... Bc8 { (Nf4) -0.63/7 12} 25. Nb3 {(Bxd5) 0.06/7 9}
25... Qd6 {-0.66/8 12} 26. Qg3 { (g3) 0.11/6 9} 26... Qxg3 {(Qd7) -0.59/8 12}
27. hxg3 {0.10/7 9} 27... Nf6 { (Bg4) -0.63/8 10} 28. Bf3 {0.09/7 8} 28... g5
{(Rxe1+) -0.69/8 16} 29. Bb4 { 0.22/6 8} 29... Be6 {-0.56/8 8} 30. Nc5 {0.31/7
8} 30... Nd5 {(Bc8) -0.25/7 6} 31. Nxe6 {(Bd2) 0.55/8 8} 31... Rxe6 {-0.31/7 1}
32. Bc5 {0.40/8 8} 32... b6 { (Bb6) 0.00/8 5} 33. Ba3 {0.62/8 7} 33... Bd6
{(Rf6) 0.09/8 5} 34. Bxd5 { 0.37/9 7} 34... cxd5 {0.22/8 1} 35. Rxe6 {0.05/11 7}
35... fxe6 {0.00/10 1} 36. Bxd6 {-0.06/10 7} 36... Rxd6 {0.12/11 3} 37. Rc1
{-0.08/10 7} 37... Kg7 { (Rd7) 0.12/10 5} 38. Rc7+ {0.61/9 6} 38... Kf6 {0.09/10
4} 39. Rh7 {0.52/9 6} 39... Rc6 {0.03/9 7} 40. Rxh6+ {(f3) 0.47/8 6} 40... Kf5
{(Kg7) -0.13/9 3} 41. f3 {0.26/8 6} 41... Rc2 {-0.22/9 4} 42. Rh7 {0.09/9 6}
42... a5 {-0.22/9 4} 43. b3 {0.15/9 6} 43... Rxa2 {-0.47/8 3} 44. Rf7+ {-0.38/9
6} 44... Kg6 {-0.47/7 0} 45. Rb7 {-0.44/10 5} 45... a4 {-0.38/9 3} 46. Rxb6
{-0.39/9 5} 46... Kf6 { -0.50/9 4} 47. bxa4 {(f4) -0.53/9 6} 47... Rxa4
{-0.81/10 2} 48. f4 { (Kf2) -0.76/10 5} 48... Rxd4 {-1.34/9 3} 49. fxg5+
{-0.79/10 5} 49... Kf5 { -1.34/9 2} 50. Kf2 {-0.93/10 5} 50... e5 {(Rg4) -1.41/8
1} 51. Rb8 { (Rd6) -1.17/8 9} 51... Rg4 {(Kxg5) -1.56/8 2} 52. Rd8 {(Rb5)
-0.96/8 4} 52... d4 {-1.78/9 2} 53. Kf3 {(Re8) -1.20/9 4} 53... e4+ {-1.53/9 2}
54. Kf2 { -1.23/9 4} 54... Ke5 {(d3) -1.53/9 1} 55. Re8+ {-1.23/7 4} 55... Kd5 {
(Kf5) -1.72/10 3} 56. Ra8 {-0.50/7 4} 56... Kc4 {-1.72/9 2} 57. Ra5 {-2.00/9 14
} 57... Kd3 {(d3) -1.72/9 1} 58. Re5 {(Ra3+) -0.88/11 12} 58... Kd2 {-1.50/9 1}
59. Ra5 {-0.88/10 3} 59... Kc2 {(d3) -1.31/9 2} 60. Rd5 {-0.49/7 3} 60... e3+ {
(d3) -1.00/9 2} 61. Kf3 {-0.15/9 3} 61... Re4 {-0.81/10 2} 62. Ke2 {-0.48/9 3}
62... Kb3 {(Kc3) -0.78/10 4} 63. g6 {-0.01/7 3} 63... Kc4 {(Rg4) -0.53/10 2} 64.
Rd6 {(Rd8) 0.00/8 3} 64... Re7 {(Kc5) -1.22/9 2} 65. Rd8 {(Rc6+) 0.00/7 5} 65...
Ra7 {(Rg7) -1.19/8 1} 66. g4 {(Rc8+) 0.22/7 5} 66... Rb7 {(Kc5) -1.16/7 1 } 67.
g5 {(Rc8+) 0.65/6 2} 67... Rc7 {(Rg7) -0.25/8 4} 68. g4 {1.19/7 2} 68... Re7
{(Ra7) -0.16/8 1} 69. Rc8+ {0.97/7 2} 69... Kd5 {-0.53/9 2} 70. Rf8 { (Rd8+)
0.05/8 4} 70... Ke4 {-0.81/9 2} 71. Rd8 {0.05/9 2} 71... Rc7 { (Rb7) -0.41/9 2}
72. Re8+ {0.05/8 2} 72... Kd5 {(Kf4) -0.44/10 5} 73. Rd8+ { (Ra8) 0.00/8 7}
73... Kc5 {(Ke4) 0.09/9 2} 74. Rg8 {(Kd3) 1.67/6 2} 74... Rb7 { (Kd6) -0.44/7 1}
75. g7 {2.55/5 2} 75... Rc7 {1.06/9 3} 76. Kd3 {3.16/7 3} 76... e2 {(Kc6) 1.06/9
2} 77. Kxe2 {4.83/6 2} 77... Kc4 {1.47/9 2} 78. Kd2 { 5.75/8 2} 78... d3 {2.78/9
2} 79. g6 {5.53/7 2} 79... Kc5 {(Rc6) 2.81/9 1} 80. Kxd3 {6.77/7 2} 80... Rd7+
{2.81/9 2} 81. Ke4 {6.77/8 2} 81... Rc7 { (Kd6) 3.03/9 2} 82. g5 {7.31/7 2}
82... Re7+ {(Kc4) 3.37/8 1} 83. Kf5 {7.29/7 1 } 83... Rc7 {6.69/9 1} 84. Rd8
{8.96/8 2} 84... Rxg7 {7.44/11 2} 85. Kf6 { 10.02/8 1} 1-0
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.