Life In 19x19
http://lifein19x19.com/

New networks Leela Zero, a Sensible Artificial Intelligence
http://lifein19x19.com/viewtopic.php?f=18&t=17023
Page 3 of 3

Author:  bugsti [ Wed Jan 08, 2020 3:13 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

And wrote:
georgenan wrote:
SAI gives the only right answer to this

why is the only right answer? White can make any move and still win (maybe almost any :) )

here https://lifein19x19.com/viewtopic.php?p=244851#p244851 really the only right answer


Nope, you are probably forgetting the tiny differences between japanese and chinese rules :razz: :razz:

There is just one winning sequence for white under chinese rules.

Author:  Bill Spight [ Wed Jan 08, 2020 3:40 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

bugsti wrote:
And wrote:
georgenan wrote:
SAI gives the only right answer to this

why is the only right answer? White can make any move and still win (maybe almost any :) )

here https://lifein19x19.com/viewtopic.php?p=244851#p244851 really the only right answer


Nope, you are probably forgetting the tiny differences between japanese and chinese rules :razz: :razz:

There is just one winning sequence for white under chinese rules.


There are five winning moves for White under both Japanese and Chinese rules. (Edit: Corrected number of winning moves.) I haven't looked at the problem for a long time, but all problems in the book are White to play and win so that Japanese and Chinese rules agree. I.e., if White wins by 1 pt. of territory there will be an even number of stones on the board, the same number for each player.

Author:  And [ Thu Jan 09, 2020 3:22 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

Zenith Go 7 Analyze:
S14 - 52
Q10 - 51
H2 - 51
H16 - 49
G15 - 49

(fixed in SGF: Komi 0 and Chinese rules)

Author:  Bill Spight [ Fri Jan 10, 2020 7:59 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

And wrote:
Zenith Go 7 Analyze:
S14 - 52
Q10 - 51
H2 - 51
H16 - 49
G15 - 49

(fixed in SGF: Komi 0 and Chinese rules)


The five winning plays are S-14, Q-16, Q-07, Q-10, and R-09.

BTW, what is SAI's mainline solution?

Author:  And [ Fri Jan 10, 2020 10:10 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

SAI 0.17.5 (Sabaki, 10000 Playouts, komi 0)

#222 afe36b9d

R9 -> 8280 (V: 53.90%)
O2 -> 431 (V: 50.05%)
H2 -> 236 (V: 51.91%)
S14 -> 193 (V: 49.90%)
Q16 -> 104 (V: 45.57%)
A7 -> 101 (V: 48.08%)
A17 -> 95 (V: 46.22%)
C19 -> 94 (V: 46.42%)
Q10 -> 85 (V: 52.01%)

#223 2cc322d9

S14 -> 8423 (V: 54.28%)
O2 -> 369 (V: 50.57%)
Q7 -> 217 (V: 49.62%)
Q10 -> 212 (V: 50.41%)
A17 -> 127 (V: 45.19%)

#224 dd403c0b

S14 -> 7354 (V: 54.74%)
H2 -> 1628 (V: 51.89%)
O2 -> 241 (V: 48.18%)
Q16 -> 169 (V: 46.91%)
C19 -> 114 (V: 45.18%)

#225 ef877be3

H2 -> 8419 (V: 54.16%)
O2 -> 409 (V: 49.71%)
S14 -> 355 (V: 51.01%)
Q16 -> 174 (V: 48.24%)
A17 -> 104 (V: 42.99%)

#230 6ebb71b9

S14 -> 6576 (V: 51.31%)
Q7 -> 2199 (V: 50.75%)
O2 -> 170 (V: 44.59%)
B8 -> 115 (V: 44.15%)
A7 -> 113 (V: 45.65%)

Author:  Bill Spight [ Fri Jan 10, 2020 12:58 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

Thanks for SAI's variations. :)
Not solved, I am afraid. Here is an SGF file. :)


Author:  And [ Sat Jan 11, 2020 6:06 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

CS Zero probably sees some trick :) :
C19 - 57.6% 85178026
A17 - 56.5% 155301
H16 - 55% 122714
S14 - 56.1% 85306
A7 - 54.7% 65181

playing on dan 9 chooses a move H2

Author:  georgenan [ Sun Jan 12, 2020 8:32 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

Sorry for my unclear description.

It's is a test for professionals in endgame stage and it's designed by Berlekamp https://math.berkeley.edu/~berlek/index.html

It's Japanese Rule and its goal is to give the right sequence with no concession.

White first. When komi>2, it's easy to win for bots.

Since Katago is less-consolidation with territory-counting and SAI is less-consolidation with komi-estimation,
But SAI's 9b will give wrong answer in live-or-dead situation.
In this game, SAI always makes wrong judgement on top left 10 stones.

My SAI parameters:
Code:
-g -b 0  --noponder -p 0 -v 0 -r 3 -m 0 --komi 0 --nrsymm --lambda 0.1 --mu 0.5 -t 6 --batchsize 5 -j

SAI performs more accurate than KataGo when komi = 0.

I'm using weight file SAI238.

Using --lambda 0.5 SAI may try aggressive moves. Since --lambda 0.1 is used in selfplays I think it could be a proper choice.
Using --mu less than 0.2 will give H2 as other strong bots.

Author:  georgenan [ Sun Jan 12, 2020 8:33 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

S14 is the best move by designer Prof. Berlekamp and Chang Hao gave the right answer in news report.
But the mathematical analysis method hadn't accepted by go players.
I read this on a Chinese articlehttps://zhuanlan.zhihu.com/p/33860825 without the sequence after S14.

Author:  Bill Spight [ Mon Jan 13, 2020 1:31 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

georgenan wrote:
It's is a test for professionals in endgame stage and it's designed by Berlekamp https://math.berkeley.edu/~berlek/index.html


This problem is based upon Theorem 8 in the book, which David Wolfe solved for his Ph.D. thesis. :)

Quote:
It's Japanese Rule and its goal is to give the right sequence with no concession.


The problems in the book lead to the same answer under both Japanese and Chinese rules, as with correct play White wins by 1 pt. with 0 komi. Generally there is no single correct sequence, and not always only one correct first play. Amateurs who have understood the book can solve these problems in a minute or two.

Quote:
In this game, SAI always makes wrong judgement on top left 10 stones.


I take it that you are referring to the 10 White stones that eventually will be put in atari. One important aspect of this position is that with correct play that atari will not happen before the dame stage. It is unlikely that a bot will have learned those conditions through self play. That fact makes plays gain less than 1 pt. by territory scoring and less than 2 pts. by area scoring. IMHO the book does not explain the conditions under which that is the case clearly enough for practical go players who are not mathematicians. Oven 10 years ago I wrote up a small article which I hope does a better job of explaining Wolfe's results. I would like to get it published in a top weiqi magazine.

Author:  Bill Spight [ Mon Jan 13, 2020 1:50 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

georgenan wrote:
S14 is the best move by designer Prof. Berlekamp and Chang Hao gave the right answer in news report.


David Wolfe may be the composer of that problem, as it is based upon his theorem. :) Anyway, you can show that White S-14 is the best firs move, but four other first moves produce a White win with correct play.


Quote:
But the mathematical analysis method hadn't accepted by go players.


This problem may be analyzed using difference games, which Berlekamp and Wolfe explain in their book, and by the method of multiples, which I deveolped to illustrate the mean value theorem. I have written about both methods online for many years. Instead of the method of multiples it may be analyzed by thermography, which also appears in the book, or by other methods which I discovered on my own, and I am sure other people have, as well. See Robert Jasiek's books.

But, yes, sadly, the methods in the book have not been adopted by pros, even though the book came out in 1994, and other papers on these topics have been published since. I ran into unexpected opposition a few years ago in getting my article about getting the last play published in the magazine Myosu. The opposition came not from a pro, but from a strong amateur on their editorial staff. {shrug}

Author:  And [ Wed Jan 15, 2020 8:06 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

networks of 12x256 appeared in test matches, soon there will probably be a change http://sai.unich.it

Author:  Javaness2 [ Wed Jan 15, 2020 8:43 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

I am rather late to the party, so forgive me if my question sounds a bit rude, but...

What is different about this HNG variant of LeelaZero, why should we be interested in it?

Author:  And [ Wed Jan 15, 2020 9:16 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

https://github.com/sai-dev/sai/wiki/Features

Author:  And [ Fri Feb 14, 2020 4:23 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

Probably today there will be a transition to 12x256

Author:  And [ Fri Feb 14, 2020 12:58 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

"Hello! We are moving to 12x256 with the next promotion. Stay tuned!"
https://github.com/sai-dev/sai/issues/3 ... -586396400

Author:  lichigo [ Fri Feb 14, 2020 2:20 pm ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

Hello, I'm wondering what is the difference between 40x256 and 12x256?

Author:  counting [ Sat Feb 15, 2020 11:43 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

lichigo wrote:
Hello, I'm wondering what is the difference between 40x256 and 12x256?


Obviously 40x256 is 10240 and 12x256 is 3072, so 40x256 has more power :rambo:

Kidding aside, how seriously do you want your answer?

If you only want a graphic representation of the structure, LZ's structure is similar to AlphaGo Zero
https://i.imgur.com/2e4iyAy.jpg

If you want to know what each technical term means, there is a technical explanation for Leela Chess Zero, and the terminologies and technologies used are the same.
https://github.com/LeelaChessZero/lc0/w ... Chess-Zero

If you really want to know the details you can go to their github page and read all the AlphaGo papers by Deepmind
https://github.com/leela-zero/leela-zer ... /README.md

And as the number suggested, more blocks it has, more complex patterns it can remember and has a better instinct to find the next move.

Author:  splee99 [ Mon Feb 17, 2020 10:39 am ]
Post subject:  Re: New networks Leela Zero, a Sensible Artificial Intellige

Don't forget the capability to play handicap games. Although the 40 block network is powerful, it basically has not much idea how to play handicap games.

Page 3 of 3 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/