Huh, learning, what is it good for? Absolutly ...

[Michael Sherwin]

... you tell me!

In George Lyapko's 30 minute tournament this report starts with 'Leauge M', the leauge just before RomiChess joins in.

SpiderChess, BBChess, Deuterium and LittleThought promote while NanoSzachy and Anechka are left behind 1/2 point back.

Leauge L saw the addition of TwistedLogic, RomiChess and Homer to the fray. The only survivor of the preveous promotees was SpiderChess in the third spot:

Code: Select all

Leauge L
TwistedLogic 23.5/30 (TL lost 4 games on time)
RomiChess... 23.5/30
SpiderChess. 23.0/30
Homer....... 21.5/30

Only BBChess challanged the top four with 21.0/30 and most likely would have promoted over RomiChess if Romi would not have been extreamly lucky against Spidie and Twisted.

The strongest program to join the battle in Leauge K was DanaSah but was left behind. However, RomiChess finished fourth this time 1.5 points behind Homer and I thought that Romi's ride was about over.

Code: Select all

Leauge K
TwistedLogic 25.5/30
SpiderChess. 24.5/30
Homer....... 24.0/30
RomiChess... 22.5/30

A lot of talent was present for Leauge J. RomiChess after 22 games was in fith place behind Xpdnt, however, I was confident that RomiChess would once again promote as Romi had played the more strong opponents. RomiChess did finish in the fourth spot, only 1/2 point behind Merlin which replaced Homer in the list of promotees.

Code: Select all

Leauge J
TwistedLogic 25.5/30
SpiderChess. 21.0/30
Merlin...... 21.0/30
RomiChess... 20.5/30

Poor Homer had a couple too many beers and only finished with 15.5/30.

RomiChess started in Leauge L without an opening book and has been creating one of her own ever since as well as remembering scores of positions, that she places in the hash file prior to searching. I believe that the learning by RomiChess is what is keeping her in this compitition.

Even more talent shows up for Leauge I. It did not look like RomiChess would continue on. Merlin was replaced by Typhoon and finished in 5th place, two points behind RomiChess!

Code: Select all

Leauge I
TwistedLogic 25.0/30
SpiderChess. 21.0/30
Typhoon..... 19.5/30
RomiChess... 19.5/30

Leauge H was the biggest surprise yet! This time SpiderChess was left behind(?!), TwistedLogic came down below the clouds, Diablo joined the promotee list and RomiChess again held down fourth place, but tied for second with 20 points. Is Romi's learning having an effect?

Code: Select all

Leauge H
TwistedLogic 21/30
Typhoon..... 20/30
Diablo...... 20/30
RomiChess... 20/30

Leauge G has started and it does not look very good for Romi, 3/6, with a lot of tough opponents to play. I wonder if she can pull another one out of her hat?

If anyone is interested enough to want to examine Romi's games and then make a report about how Romi's learning has affected the outcome of George's tournament it could be most interesting and very appretiated! The games ofcourse are downloadable from George's site.

Thanks

[Michael Sherwin]

Seems that I overlooked the fact that George preserved Romi's learn file from his 0608 Leauge K event and that Romi started that event with a few test games. So, if there is anyone that will be making a report (and I really hope that there will be) on the effectiveness of Romi's learning then they should start with the 0608k event.

[Michael Sherwin]

George has updated Leauge G with results from the end of round 14. Romi is still in touch with 4th place at only 1 point back. Romi has the 5th highest S-B score so her chances of another 4th place finish do not look hopeless with 16 games left to play. The good news is that Romi managed to split with Typhoon and take 1.5/2 from Twisted! Can Romi make it 5, 4th place finishes in a row?

[Michael Sherwin]

Well Romi's ride is over.

73.BugChess2 v1.5.2 - 25.5(350.50)/30
74.Typhoon 1.00r343 - 22(310.50)/30
75.Delphil 1.8 - 22(289.50)/30
76.Diablo v0.5.1 - 18.5(243.25)/30
77.PostModernist 1016 - 17.5(236.25)/30
78.Snitch 1.6.2 - 16.5(223.50)/30
79.Bruja v1.9 - 16(212.75)/30
80.RomiChess_NG4 - 15.5(199.25)/30
81.Matacz_13 - 15.5(186.50)/30
82.TwistedLogic 20070915 - 13(158.50)/30
83.Esc_116 - 12(161.00)/30
84.GNUChess_507DC - 10.5(152.50)/30
85.Patzer 3.80 - 10.5(137.75)/30
86.Butcher 1.58 - 9.5(128.50)/30
87.Beowulf_24 - 8.5(100.50)/30
88.GNUChess_4TM - 7(88.75)/30

Romi did finish ahead of Matacz (Tie Break) and 2.5 points ahead of Twisted!

Not a bad little run by Romi, I would say!

The End.

[Tony Thomas]

Looking at the version number, NG4 was 6 versions or so ago.

[Dr.Wael Deeb]

Hi Tony,
I am back pal and will stay for a long time now....
At a slower time control it seems that NG4 is the strongest version

[swami]

Hi Tony,
I am back pal and will stay for a long time now....
At a slower time control it seems that NG4 is the strongest version

Welcome back,Wael

How was your studies?

[Dr.Wael Deeb]

Hi Tony,
I am back pal and will stay for a long time now....
At a slower time control it seems that NG4 is the strongest version

Welcome back,Wael

How was your studies?

Hi Swami,
Thanks pal
It was a nightmare which ended with full success

[Tony Thomas]

Hi Tony,
I am back pal and will stay for a long time now....
At a slower time control it seems that NG4 is the strongest version

Welcome back,Wael

How was your studies?

Hi Swami,
Thanks pal
It was a nightmare which ended with full success

Time to get back to posthwhoring.

[BubbaTough]

Howdy,

I will give you some unscientific feedback. I downloaded RomiP3jx64 to test against LearningLemming using the Manhattan book some time ago. I don't have the numbers, but Romi very clearly improved its play dramatically from its learning, much more than I would have expected given what I know of your technique. It went from losing almost all its games to being quite competitive pretty quickly...maybe 20 games or so. This was true even when I had my learning on also. The improvement did not seem to transfer over that well if I switched time controls, however. I am not sure if that was only true when switching to a slower time control, or was bidirectional. I also never tried Romi against other opponents, so I am not sure how well its learning generalizes.

One thing I did note is learning seemed more effective in openings that were prone to positional maneuvering with lots of transpositions (like c4/g3 systems) than anything else. If there is a more recent version / opening book let me know and I will try to do a little testing and post results if you want.

Your nodes per second are amazing by the way. Much much better than LearningLemming.

By the way x 2, Romi's presence is missed in the ACCA events. Any reason why you haven't been entering?

-Sam