Something new ?

[Rebel]

Only to Plentychess to show that learning matters.

I'd be curious to see what happens when both learn. My guess is that the learning impact would shrink, but not quite vanish, as the weaker side has more to gain from papering over difficult positions. Can you please test this?

I am currently running the first match on the "AI Rating List", Dog 4.1.0 playing 9000 games, so no time yet.

Code: Select all

No. Name                Win Draw Loss Unf.  Score Games       %
---------------------------------------------------------------
  1 Dog-4.10.1         +215 =262 -134   *0  346.0   611   56.6%
  2 Jet-1.2             +44 =129  -31   *0  108.5   204   53.2%
  3 Simbelmyne-v1.10.0  +58  =62  -83   *0   89.0   203   43.8%
  4 Pingu-5.0.0         +32  =71 -101   *0   67.5   204   33.1%

[royb]

Only to Plentychess to show that learning matters.

I'd be curious to see what happens when both learn. My guess is that the learning impact would shrink, but not quite vanish, as the weaker side has more to gain from papering over difficult positions. Can you please test this?

I am currently running the first match on the "AI Rating List", Dog 4.1.0 playing 9000 games, so no time yet.

Code: Select all

No. Name                Win Draw Loss Unf.  Score Games       %
---------------------------------------------------------------
  1 Dog-4.10.1         +215 =262 -134   *0  346.0   611   56.6%
  2 Jet-1.2             +44 =129  -31   *0  108.5   204   53.2%
  3 Simbelmyne-v1.10.0  +58  =62  -83   *0   89.0   203   43.8%
  4 Pingu-5.0.0         +32  =71 -101   *0   67.5   204   33.1%

What time control are you using for these 9000 games?

[Leto]

Can you explain what the learning file does? Is it the book learning function in the Fritz GUI?

[Rebel]

First match of the AI (all inclusive) Rating List

Code: Select all

No. Name                 Win  Draw  Loss Score  Games   %   book
----------------------------------------------------------------
  1 Dog-4.10.1         +3183 =3756 -2061 5061.0  9000 56.2% dog-book.bin
  2 Jet-1.2            +690  =1725 -585  1552.5  3000 51.8% none
  3 Simbelmyne-v1.10.0 +897  =980  -1123 1387.0  3000 46.2% none
  4 Pingu-5.0.0        +474  =1051 -1475  999.5  3000 33.3% none

Next Komodo Dragon 2.5 with book.

~~~~~~~~~~

IMPORTANT

From now on results will be posted on :

https://prodeo.actieforum.com/t1818-ai- ... ults#18607

[Rebel]

What time control are you using for these 9000 games?

Currently 40/10, when the list has some body (100+ engines) I will consider a second list at a slower time control.

But one thing counts -> volume is all, more important than time control.

[royb]

What time control are you using for these 9000 games?

Currently 40/10, when the list has some body (100+ engines) I will consider a second list at a slower time control.

But one thing counts -> volume is all, more important than time control.

40 moves in 10 minutes? That will take quite some time to complete 9000 games! I know you realize that being the man who wrote chess programs for so many years - I'm surprised to see such a long time control in an era where many tests seems to be done at very fast time controls. I'm eager to see the results when this is done!

[Rebel]

What time control are you using for these 9000 games?

Currently 40/10, when the list has some body (100+ engines) I will consider a second list at a slower time control.

But one thing counts -> volume is all, more important than time control.

40 moves in 10 minutes? That will take quite some time to complete 9000 games! I know you realize that being the man who wrote chess programs for so many years - I'm surprised to see such a long time control in an era where many tests seems to be done at very fast time controls. I'm eager to see the results when this is done!

40 moves in 10 seconds

[royb]

What time control are you using for these 9000 games?

Currently 40/10, when the list has some body (100+ engines) I will consider a second list at a slower time control.

But one thing counts -> volume is all, more important than time control.

40 moves in 10 minutes? That will take quite some time to complete 9000 games! I know you realize that being the man who wrote chess programs for so many years - I'm surprised to see such a long time control in an era where many tests seems to be done at very fast time controls. I'm eager to see the results when this is done!

40 moves in 10 seconds

Aha! Thanks for clarifying! I will you well with this concept of yours - it's is intriguing!