Re: [Computer-go] Detlef's DCNN data

[Hiroshi Yamashita]

Congratulations. Very promising results. I just noticed Aya is now much


Thanks!
AyaMC6 (1 playouts/move + DCNN) got weak 4d after 67 rated games.
Time setting is 10 minutes + 30 sec x5.
Without DCNN, AyaMC2 (1 playouts/move, same time setting)
is 2.7d (from rank graph). So DCNN changes from 2.7d to 4.0d,
about +1.3d. 
KGS 1 rank is +230 Elo for 2d+.(maybe) So increase is +299 Elo.

I wonder if there is faster DCNN library without GPU.

1 thread, Xeon W3680 3.3GHz, 1 playout/move
 KGS rankGPU 
withDCNN,   1160 playout/sec,4.0d,GTS 450

without DCNN,   2206 playout/sec,2.7d ---

KGS Rating Math
http://senseis.xmp.net/?KGSRatingMath

Reagrds,
Hiroshi Yamashita

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] How to only play games with correct handicap on KGS?

[Nick Wedd]

Hi Martin,

I don't think so.  Hideki Kato had the same problem with Zen, and as far as
I know, he never found a satisfactory solution.

Cheers,
Nick

On 1 October 2015 at 16:58, Martin Mueller  wrote:

> I have a new version of Fuego playing on KGS as ranked robot fuego19. I
> would like people to play with correct handicap to get a more reliable
> rating. Is there a way to do this with kgsgtp?
>
> Thanks
>
> Martin
>
> ___
> Computer-go mailing list
> Computer-go@computer-go.org
> http://computer-go.org/mailman/listinfo/computer-go




-- 
Nick Wedd  mapr...@gmail.com
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

[Computer-go] How to only play games with correct handicap on KGS?

[Martin Mueller]

I have a new version of Fuego playing on KGS as ranked robot fuego19. I would 
like people to play with correct handicap to get a more reliable rating. Is 
there a way to do this with kgsgtp?

Thanks

Martin

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] How to only play games with correct handicap on KGS?

[Gonçalo Mendes Ferreira]

I'm not sure but try removing the free handicap placement commands from 
the list of supported commands, returned by the command list_commands. 
The commands should be set_free_handicap and place_free_handicap.


The documentation on kgsGTP reads "If your engine does not support these 
commands then handicap games will be refused.".


Gonçalo F.

On 01/10/2015 16:58, Martin Mueller wrote:

I have a new version of Fuego playing on KGS as ranked robot fuego19. I would 
like people to play with correct handicap to get a more reliable rating. Is 
there a way to do this with kgsgtp?

Thanks

Martin

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go


___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] How to only play games with correct handicap on KGS?

[Hideki Kato]

I've asked wms several times to add such option to the config file but 
his answer was always "it's human players' right to play with any 
handicap they like."

Technically, it's possible; peeping the communication between the 
server and client, pick-up the account of the opponent, and get his/her 
rank using web interface of KGS, though I haven't tried because I'd like 
to respect wms's policy upon using KGS.  Instead, I've asked those 
players to respect "correct" handicap with some comments. ex. "I'm 
running bots as an academic research in order to evaluate their 
performance against human players."

Related issue: To prevent free placement of the handicap stone(s), 
use Japanese rules.

Hideki

Martin Mueller: :
>I have a new version of Fuego playing on KGS as ranked robot fuego19. I would 
>like people to 
>play with correct handicap to get a more reliable rating. Is there a way to do 
>this with 
>kgsgtp?
>
>Thanks
>
>   Martin
>
>___
>Computer-go mailing list
>Computer-go@computer-go.org
>http://computer-go.org/mailman/listinfo/computer-go
-- 
Hideki Kato 
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Detlef's DCNN data

[Hiroshi Yamashita]

Hm, AyaMC4 is 4d since last year.


AyaMC4 is 10sec/move on 6 cores, about 200,000 playouts/move.
And time setting is blitz, 1 minutes + 15 sec x10.
AyaMC uses same setting, except time is 10 mimutes + 30sec x5.
Its rank is 3d. Aya never get 4d in normal time setting.

Hiroshi Yamashita

___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] Detlef's DCNN data

[Hideki Kato]

Aja Huang: 

Re: [Computer-go] How to only play games with correct handicap on KGS?

[Xavier Combelle]

If you refuse the handicap, the games will be *without* handicap not *with
correct* handicap

2015-10-01 19:13 GMT+02:00 Gonçalo Mendes Ferreira :

> I'm not sure but try removing the free handicap placement commands from
> the list of supported commands, returned by the command list_commands. The
> commands should be set_free_handicap and place_free_handicap.
>
> The documentation on kgsGTP reads "If your engine does not support these
> commands then handicap games will be refused.".
>
> Gonçalo F.
>
>
> On 01/10/2015 16:58, Martin Mueller wrote:
>
>> I have a new version of Fuego playing on KGS as ranked robot fuego19. I
>> would like people to play with correct handicap to get a more reliable
>> rating. Is there a way to do this with kgsgtp?
>>
>> Thanks
>>
>> Martin
>>
>> ___
>> Computer-go mailing list
>> Computer-go@computer-go.org
>> http://computer-go.org/mailman/listinfo/computer-go
>>
>
> ___
> Computer-go mailing list
> Computer-go@computer-go.org
> http://computer-go.org/mailman/listinfo/computer-go
>
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] I would like people to play with correct handicap to get a more reliable rating

[djhbrown .]

.
whereas the majority of teenage kgsers i encounter suffer from
hyperhubris amongst a plethora of other sociopathological handicaps,
mathematically, it ought not to matter a whit to a bot's own
self-esteem when the witless up themselves against it because the kgs
ranking algorithm surely ought to either take rank aberrations into
account or ignore delusionally-handicapped-game results because they
don't just do it against you, they do it against everybody.

but in any case, how reliable is poking a stick around in the dark anyway?

PS  Whereas it is true that a camel is a horse that was designed by a
committee, why on earth don't you guys stop trying to get one over
each other in the kgs playground and start working as a team?  For
example, DCNNigo plays a very respectable opening but falls apart in
the yose, so a different technique is required for that phase of the
game.  If you put your heads together, you might come up with a
multibot that arguably could be considered smart instead of just
lucky.  You should be able to hook up across internet so the many
heads don't all have to be in the same room.  Yonks ago Oliver
Selfridge [1] proposed just such an architecture for an intracranial
artificial intelligence - but you could go intercranial.  Now that
would be an IT worth checking out.
[1] http://sites.sinauer.com/wolfe4e/wa04.02.html
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go