Re: [Computer-go] Our Silicon Overlord

[Robert Jasiek]

On 06.01.2017 23:37, Jim O'Flaherty wrote:

into a position with superko [...] how do you even get AlphaGo into a the arcane
state in the first place,


I can't in practice.

I have not provided a way to beat AlphaGo from the game start at the 
empty board.


All I have shown is that there are positions beyond AlphaGo's 
capabilities to refute your claim that AlphaGo would handle all 
positions well.


Hui and Lee constructed positions with such aspects: Hui with long-term 
aji, Lee with complex reduction aji. Some versions of AlphaGo mishandled 
the situations locally or locally + globally.



The professional players will be
open to all sorts of creative ideas on how to find weaknesses with AlphaGo.


Or the amateur players or theoreticians.


Perhaps you can persuade one of the 9p-s to explore your idea
of pushing the AlphaGo AI in this direction.


Rather I'd need playing time against AlphaGo.


IOW, we are now well outside of provable spaces


For certain given positions, proofs of difficulty exist. Since Go is a 
complete-information game, there can never be a proof that AlphaGo could 
never do it. There can only ever be proofs of difficulty.



mathematical proof around a full game


From the empty board? Of course not (today).


We cannot formally prove much simpler models,


Formal proofs for certain types of positions (such as with round_up(n/2) 
n-tuple kos) exist.


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

Re: [Computer-go] Our Silicon Overlord

[Jim O'Flaherty]

Okay. So I will play along. How do you think you would coax AlphaGo into a
position with superko without AlphaGo having already simulated that pathway
as a less probable win space for itself when compared to other playing
trees which avoid it? IOW, how do you even get AlphaGo into a the arcane
state in the first place, especially since uncertainty of outcome is
weighted against wins for itself?

And since I know you cannot definitively answer that, it looks like we'll
just have to wait and see what happens. The professional players will be
open to all sorts of creative ideas on how to find weaknesses with AlphaGo.
And until they get free reign to play as many games as they like against it
so they can begin to get a feel for strategies that do expose probable
weaknesses (we won't know with certainty as it appears AlphaGo is now
generating its own theories where a situation is rated a weakness by a
human turns out to be incorrect and AlphaGo ends up leveraging it to its
advantage). Perhaps you can persuade one of the 9p-s to explore your idea
of pushing the AlphaGo AI in this direction.

IOW, we are now well outside of provable spaces and into probabilistic
spaces. At the scales we are discussing, it is improbable we will ever
directly experience seeing anything approaching a mathematical proof around
a full game of Go between two experts, even if those experts are two
competing AIs. We cannot formally prove much simpler models, much less ones
with the complexity of a game of Go.


On Fri, Jan 6, 2017 at 12:55 AM, Robert Jasiek  wrote:

> On 05.01.2017 17:32, Jim O'Flaherty wrote:
>
>> I don't follow.
>>
>
> 1) "For each arcane position reached, there would now be ample data for
> AlphaGo to train on that particular pathway." is false. See below.
>
> 2) "two strategies. The first would be to avoid the state in the first
> place." Does AlphaGo have any strategy ever? If it does, does it have
> strategies of avoiding certain types of positions?
>
> 3) "the second would be to optimize play in that particular state." If you
> mean optimise play = maximise winning probability.
>
> But... optimising this is hard when (under positional superko) optimal
> play can be ca. 13,500,000 moves long and the tree to that is huge. Even
> TPU sampling can be lost then.
>
> Afterwards, there is still only one position from which to train. For NN
> learning, one position is not enough and cannot replace analysis by
> mathematical proofs ALA the NN does not emulate mathematical proving.
>
>
> --
> robert jasiek
> ___
> 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] it's alphago

[Xavier Combelle]

To my knowledge, fishtest is also a major part of stockfish engine. It
is essential because there is lot of possible improvement and most of
them win only 2 or 3 elo points, but added, it lead to 60-70 elo points
between each release (every one year or something like that)


Le 06/01/2017 à 17:22, daniel rich a écrit :
> A closer example than the mersenne prime search is fishtest from the
> chess engine world. My understanding is that it is a key part of why
> stockfish is such a strong chessengine.
>
> https://github.com/glinscott/fishtest
>
> A large group of volunteers that essentially donate compute power to
> test changes and improve the bot. That would be a fairly cool way
> compute time to be made available to the community. The plus is that
> eventually big corporate players may lose interest to devote the same
> level of spending and compute that we have seen so far.
>
>
>
> On Fri, Jan 6, 2017 at 8:01 AM, Lukas van de Wiel
> > wrote:
>
> A project similar to the Great Mersenne Prime search might be a
> possibility to distribute the work of training the network among many
> enthousiasts, and to keep improving it by self play.
>
> On 1/6/17, Andy  > wrote:
> > What is Ray? Strongest open source bot? Anyone have a link to it?
> >
> > On Fri, Jan 6, 2017 at 3:39 AM, Hiroshi Yamashita
> > wrote:
> >
> >> If value net is the most important part for over pro level, the
> problem
> >> is
> >> making strong selfplay games.
> >>
> >> 1. make 30 million selfplay games.
> >> 2. make value net.
> >> 3. use this value net for selfplay program.
> >> 4. go to (1)
> >>
> >> I don't know when the progress will stop by this loop.
> >> But if once strong enough selfplay games are published,
> everyone can make
> >> pro level program.
> >> 30 million is big number. It needs many computers.
> >> Computer Go community may be able to share this work.
> >> I can offer Aya, it is not open-source though. Maybe
> Ray(strongest open
> >> source so far)  is better choice.
> >>
> >> Thanks,
> >> Hiroshi Yamashita
> >>
> >> - Original Message - From:  >
> >> To:  >
> >> Sent: Friday, January 06, 2017 4:50 PM
> >> Subject: Re: [Computer-go] it's alphago
> >>
> >>
> >> Competitive with Alpha-go, one developer, not possible. I do
> think it is
> >> possible to make a pro level program with one person or a small
> team.
> >> Look
> >> at Deep Zen and Aya for example. I expect I’ll get there (pro
> level) with
> >> Many Faces as well.
> >>
> >> David
> >>
> >> ___
> >> 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

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

Re: [Computer-go] it's alphago

[Hiroshi Yamashita]

Ray was Japanese student program that went on 7th, UEC cup 2016.

Ray
http://computer-go-ray.com/eng/index.html
Thare is a stronger version of Ray, with policy net and value net.
https://github.com/zakki/Ray/tree/nn
CGOS BayesElo is 3463 (Rn.3.3-4c).
http://www.yss-aya.com/cgos/19x19/bayes.html

Hiroshi Yamashita

- Original Message - 
From: "Andy" 

To: "computer-go" 
Sent: Friday, January 06, 2017 11:48 PM
Subject: Re: [Computer-go] it's alphago


What is Ray? Strongest open source bot? Anyone have a link to it?


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

Re: [Computer-go] it's alphago (How to get a strong value network)

[Hiroshi Yamashita]

Hi,


If I understood correctly you would try to use a program using value net
with (let's say 2000 playouts) in selfplay? Using only one result, or


Yes. 2000 playouts/move MCTS with policy net and value net.


doing some games per position? Or are you thinking of using only the win


I thought one game per position, but some games per position looks nice option.


doing some games per position? Or are you thinking of using only the win
percentage such a program gives from his own mixing of SL network,


In my experience, game result is better than win percentage.


usage per game (at least if Rn3.3-4c is Ray on CGOS with 4 cores, NG04b


Oh NG04b is oakfoam. AlphaGo RL is about 2800 (CGOS BayesElo).
So around this rating program seems nice. And many computers don't have GPU.
To calculate DCNN on CPU, maybe we can not use big network(filter 192), but
smaller one(filter 64 or 32).

Hiroshi Yamashita

- Original Message - 
From: "Detlef Schmicker" 

To: 
Sent: Friday, January 06, 2017 7:20 PM
Subject: Re: [Computer-go] it's alphago (How to get a strong value network)



Hi,

this sounds interesting! AlphaGo paper plays only with RL network, if I
understood correctly. If we start this huge approach we should try to
carefully discuss the way (and hopefully get some hints from people
tried with much computational power :)

If I understood correctly you would try to use a program using value net
with (let's say 2000 playouts) in selfplay? Using only one result, or
doing some games per position? Or are you thinking of using only the win
percentage such a program gives from his own mixing of SL network,
search and value net?

By the way to make some promotion :) oakfoam is not far away from Ray
for this kind of approach, where you will probably try to reduce cpu/gpu
usage per game (at least if Rn3.3-4c is Ray on CGOS with 4 cores, NG04b
is oakfoam on CGOS with 10k and saving GPU usage by using only 50% of GX970)

Detlef


Am 06.01.2017 um 10:39 schrieb Hiroshi Yamashita:

If value net is the most important part for over pro level, the problem
is making strong selfplay games.

1. make 30 million selfplay games.
2. make value net.
3. use this value net for selfplay program.
4. go to (1)

I don't know when the progress will stop by this loop.
But if once strong enough selfplay games are published, everyone can
make pro level program.
30 million is big number. It needs many computers.
Computer Go community may be able to share this work.
I can offer Aya, it is not open-source though. Maybe Ray(strongest open
source so far)  is better choice.

Thanks,
Hiroshi Yamashita

- Original Message - From: 
To: 
Sent: Friday, January 06, 2017 4:50 PM
Subject: Re: [Computer-go] it's alphago


Competitive with Alpha-go, one developer, not possible. I do think it is
possible to make a pro level program with one person or a small team.
Look at Deep Zen and Aya for example. I expect I’ll get there (pro
level) with Many Faces as well.

David


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

Re: [Computer-go] it's alphago

[daniel rich]

Oh sorry I mispoke, corporate players losing interest is a bad thing in my
mind but also more or less inevitable(to some degree anyway). I was simply
saying that as delighted as I am that google and other players are putting
so much money and research into go I suspect eventually the resources will
be re-allocated to other things, and having a large community network would
help push things forward even after/if resources dwindle.

Just as you point out deepmind is not selling their bot and I think aren't
motivated as much by the game of go so much as the AI breakthroughs it
represents. Even if many others players stay fairly active I think the
resources currently invested in go are larger than the market for go bots.

For example in the chess world there are some valuable chess engines but
the engines are now limited mostly to companies that directly sell the
engines or community/non-profit efforts. I suspect at some point go will be
the same.

On Fri, Jan 6, 2017 at 10:49 AM, Marc Landgraf  wrote:

> And why would it be desirable that 'the big corporate players lose
> interest to devote computer power'?
> And who are those big corporate players? Deepmind? Who are not even
> selling their bot? Or are you talking about CS/Zen who are having indeed
> financial interests here?
> What would be the benefit of any of those parties in losing interest?
>
> Am 06.01.2017 17:22 schrieb "daniel rich" :
>
>> A closer example than the mersenne prime search is fishtest from the
>> chess engine world. My understanding is that it is a key part of why
>> stockfish is such a strong chessengine.
>>
>> https://github.com/glinscott/fishtest
>>
>> A large group of volunteers that essentially donate compute power to test
>> changes and improve the bot. That would be a fairly cool way compute time
>> to be made available to the community. The plus is that eventually big
>> corporate players may lose interest to devote the same level of spending
>> and compute that we have seen so far.
>>
>>
>>
>> On Fri, Jan 6, 2017 at 8:01 AM, Lukas van de Wiel <
>> lukas.drinkt.t...@gmail.com> wrote:
>>
>>> A project similar to the Great Mersenne Prime search might be a
>>> possibility to distribute the work of training the network among many
>>> enthousiasts, and to keep improving it by self play.
>>>
>>> On 1/6/17, Andy  wrote:
>>> > What is Ray? Strongest open source bot? Anyone have a link to it?
>>> >
>>> > On Fri, Jan 6, 2017 at 3:39 AM, Hiroshi Yamashita 
>>> wrote:
>>> >
>>> >> If value net is the most important part for over pro level, the
>>> problem
>>> >> is
>>> >> making strong selfplay games.
>>> >>
>>> >> 1. make 30 million selfplay games.
>>> >> 2. make value net.
>>> >> 3. use this value net for selfplay program.
>>> >> 4. go to (1)
>>> >>
>>> >> I don't know when the progress will stop by this loop.
>>> >> But if once strong enough selfplay games are published, everyone can
>>> make
>>> >> pro level program.
>>> >> 30 million is big number. It needs many computers.
>>> >> Computer Go community may be able to share this work.
>>> >> I can offer Aya, it is not open-source though. Maybe Ray(strongest
>>> open
>>> >> source so far)  is better choice.
>>> >>
>>> >> Thanks,
>>> >> Hiroshi Yamashita
>>> >>
>>> >> - Original Message - From: 
>>> >> To: 
>>> >> Sent: Friday, January 06, 2017 4:50 PM
>>> >> Subject: Re: [Computer-go] it's alphago
>>> >>
>>> >>
>>> >> Competitive with Alpha-go, one developer, not possible. I do think it
>>> is
>>> >> possible to make a pro level program with one person or a small team.
>>> >> Look
>>> >> at Deep Zen and Aya for example. I expect I’ll get there (pro level)
>>> with
>>> >> Many Faces as well.
>>> >>
>>> >> David
>>> >>
>>> >> ___
>>> >> 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
>>
>
> ___
> 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] it's alphago

[Marc Landgraf]

And why would it be desirable that 'the big corporate players lose interest
to devote computer power'?
And who are those big corporate players? Deepmind? Who are not even selling
their bot? Or are you talking about CS/Zen who are having indeed financial
interests here?
What would be the benefit of any of those parties in losing interest?

Am 06.01.2017 17:22 schrieb "daniel rich" :

> A closer example than the mersenne prime search is fishtest from the chess
> engine world. My understanding is that it is a key part of why stockfish is
> such a strong chessengine.
>
> https://github.com/glinscott/fishtest
>
> A large group of volunteers that essentially donate compute power to test
> changes and improve the bot. That would be a fairly cool way compute time
> to be made available to the community. The plus is that eventually big
> corporate players may lose interest to devote the same level of spending
> and compute that we have seen so far.
>
>
>
> On Fri, Jan 6, 2017 at 8:01 AM, Lukas van de Wiel <
> lukas.drinkt.t...@gmail.com> wrote:
>
>> A project similar to the Great Mersenne Prime search might be a
>> possibility to distribute the work of training the network among many
>> enthousiasts, and to keep improving it by self play.
>>
>> On 1/6/17, Andy  wrote:
>> > What is Ray? Strongest open source bot? Anyone have a link to it?
>> >
>> > On Fri, Jan 6, 2017 at 3:39 AM, Hiroshi Yamashita 
>> wrote:
>> >
>> >> If value net is the most important part for over pro level, the problem
>> >> is
>> >> making strong selfplay games.
>> >>
>> >> 1. make 30 million selfplay games.
>> >> 2. make value net.
>> >> 3. use this value net for selfplay program.
>> >> 4. go to (1)
>> >>
>> >> I don't know when the progress will stop by this loop.
>> >> But if once strong enough selfplay games are published, everyone can
>> make
>> >> pro level program.
>> >> 30 million is big number. It needs many computers.
>> >> Computer Go community may be able to share this work.
>> >> I can offer Aya, it is not open-source though. Maybe Ray(strongest open
>> >> source so far)  is better choice.
>> >>
>> >> Thanks,
>> >> Hiroshi Yamashita
>> >>
>> >> - Original Message - From: 
>> >> To: 
>> >> Sent: Friday, January 06, 2017 4:50 PM
>> >> Subject: Re: [Computer-go] it's alphago
>> >>
>> >>
>> >> Competitive with Alpha-go, one developer, not possible. I do think it
>> is
>> >> possible to make a pro level program with one person or a small team.
>> >> Look
>> >> at Deep Zen and Aya for example. I expect I’ll get there (pro level)
>> with
>> >> Many Faces as well.
>> >>
>> >> David
>> >>
>> >> ___
>> >> 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
>
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] it's alphago

[daniel rich]

A closer example than the mersenne prime search is fishtest from the chess
engine world. My understanding is that it is a key part of why stockfish is
such a strong chessengine.

https://github.com/glinscott/fishtest

A large group of volunteers that essentially donate compute power to test
changes and improve the bot. That would be a fairly cool way compute time
to be made available to the community. The plus is that eventually big
corporate players may lose interest to devote the same level of spending
and compute that we have seen so far.



On Fri, Jan 6, 2017 at 8:01 AM, Lukas van de Wiel <
lukas.drinkt.t...@gmail.com> wrote:

> A project similar to the Great Mersenne Prime search might be a
> possibility to distribute the work of training the network among many
> enthousiasts, and to keep improving it by self play.
>
> On 1/6/17, Andy  wrote:
> > What is Ray? Strongest open source bot? Anyone have a link to it?
> >
> > On Fri, Jan 6, 2017 at 3:39 AM, Hiroshi Yamashita 
> wrote:
> >
> >> If value net is the most important part for over pro level, the problem
> >> is
> >> making strong selfplay games.
> >>
> >> 1. make 30 million selfplay games.
> >> 2. make value net.
> >> 3. use this value net for selfplay program.
> >> 4. go to (1)
> >>
> >> I don't know when the progress will stop by this loop.
> >> But if once strong enough selfplay games are published, everyone can
> make
> >> pro level program.
> >> 30 million is big number. It needs many computers.
> >> Computer Go community may be able to share this work.
> >> I can offer Aya, it is not open-source though. Maybe Ray(strongest open
> >> source so far)  is better choice.
> >>
> >> Thanks,
> >> Hiroshi Yamashita
> >>
> >> - Original Message - From: 
> >> To: 
> >> Sent: Friday, January 06, 2017 4:50 PM
> >> Subject: Re: [Computer-go] it's alphago
> >>
> >>
> >> Competitive with Alpha-go, one developer, not possible. I do think it is
> >> possible to make a pro level program with one person or a small team.
> >> Look
> >> at Deep Zen and Aya for example. I expect I’ll get there (pro level)
> with
> >> Many Faces as well.
> >>
> >> David
> >>
> >> ___
> >> 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] it's alphago

[Lukas van de Wiel]

A project similar to the Great Mersenne Prime search might be a
possibility to distribute the work of training the network among many
enthousiasts, and to keep improving it by self play.

On 1/6/17, Andy  wrote:
> What is Ray? Strongest open source bot? Anyone have a link to it?
>
> On Fri, Jan 6, 2017 at 3:39 AM, Hiroshi Yamashita  wrote:
>
>> If value net is the most important part for over pro level, the problem
>> is
>> making strong selfplay games.
>>
>> 1. make 30 million selfplay games.
>> 2. make value net.
>> 3. use this value net for selfplay program.
>> 4. go to (1)
>>
>> I don't know when the progress will stop by this loop.
>> But if once strong enough selfplay games are published, everyone can make
>> pro level program.
>> 30 million is big number. It needs many computers.
>> Computer Go community may be able to share this work.
>> I can offer Aya, it is not open-source though. Maybe Ray(strongest open
>> source so far)  is better choice.
>>
>> Thanks,
>> Hiroshi Yamashita
>>
>> - Original Message - From: 
>> To: 
>> Sent: Friday, January 06, 2017 4:50 PM
>> Subject: Re: [Computer-go] it's alphago
>>
>>
>> Competitive with Alpha-go, one developer, not possible. I do think it is
>> possible to make a pro level program with one person or a small team.
>> Look
>> at Deep Zen and Aya for example. I expect I’ll get there (pro level) with
>> Many Faces as well.
>>
>> David
>>
>> ___
>> 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] it's alphago

[Andy]

What is Ray? Strongest open source bot? Anyone have a link to it?

On Fri, Jan 6, 2017 at 3:39 AM, Hiroshi Yamashita  wrote:

> If value net is the most important part for over pro level, the problem is
> making strong selfplay games.
>
> 1. make 30 million selfplay games.
> 2. make value net.
> 3. use this value net for selfplay program.
> 4. go to (1)
>
> I don't know when the progress will stop by this loop.
> But if once strong enough selfplay games are published, everyone can make
> pro level program.
> 30 million is big number. It needs many computers.
> Computer Go community may be able to share this work.
> I can offer Aya, it is not open-source though. Maybe Ray(strongest open
> source so far)  is better choice.
>
> Thanks,
> Hiroshi Yamashita
>
> - Original Message - From: 
> To: 
> Sent: Friday, January 06, 2017 4:50 PM
> Subject: Re: [Computer-go] it's alphago
>
>
> Competitive with Alpha-go, one developer, not possible. I do think it is
> possible to make a pro level program with one person or a small team. Look
> at Deep Zen and Aya for example. I expect I’ll get there (pro level) with
> Many Faces as well.
>
> David
>
> ___
> 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] it's alphago

[Sebastian Scheib]

Maybe now we need AlphaGo vs. Tartrate to see who is the definitive Sai XD

2017-01-06 6:39 GMT-03:00 Hiroshi Yamashita :

> If value net is the most important part for over pro level, the problem is
> making strong selfplay games.
>
> 1. make 30 million selfplay games.
> 2. make value net.
> 3. use this value net for selfplay program.
> 4. go to (1)
>
> I don't know when the progress will stop by this loop.
> But if once strong enough selfplay games are published, everyone can make
> pro level program.
> 30 million is big number. It needs many computers.
> Computer Go community may be able to share this work.
> I can offer Aya, it is not open-source though. Maybe Ray(strongest open
> source so far)  is better choice.
>
> Thanks,
> Hiroshi Yamashita
>
> - Original Message - From: 
> To: 
> Sent: Friday, January 06, 2017 4:50 PM
> Subject: Re: [Computer-go] it's alphago
>
>
> Competitive with Alpha-go, one developer, not possible. I do think it is
> possible to make a pro level program with one person or a small team. Look
> at Deep Zen and Aya for example. I expect I’ll get there (pro level) with
> Many Faces as well.
>
> David
>
> ___
> Computer-go mailing list
> Computer-go@computer-go.org
> http://computer-go.org/mailman/listinfo/computer-go
>



-- 
Dracux
*http://www.dracux.com *
___
Computer-go mailing list
Computer-go@computer-go.org
http://computer-go.org/mailman/listinfo/computer-go

Re: [Computer-go] it's alphago (How to get a strong value network)

[Detlef Schmicker]

Hi,

this sounds interesting! AlphaGo paper plays only with RL network, if I
understood correctly. If we start this huge approach we should try to
carefully discuss the way (and hopefully get some hints from people
tried with much computational power :)

If I understood correctly you would try to use a program using value net
with (let's say 2000 playouts) in selfplay? Using only one result, or
doing some games per position? Or are you thinking of using only the win
percentage such a program gives from his own mixing of SL network,
search and value net?

By the way to make some promotion :) oakfoam is not far away from Ray
for this kind of approach, where you will probably try to reduce cpu/gpu
usage per game (at least if Rn3.3-4c is Ray on CGOS with 4 cores, NG04b
is oakfoam on CGOS with 10k and saving GPU usage by using only 50% of GX970)

Detlef


Am 06.01.2017 um 10:39 schrieb Hiroshi Yamashita:
> If value net is the most important part for over pro level, the problem
> is making strong selfplay games.
> 
> 1. make 30 million selfplay games.
> 2. make value net.
> 3. use this value net for selfplay program.
> 4. go to (1)
> 
> I don't know when the progress will stop by this loop.
> But if once strong enough selfplay games are published, everyone can
> make pro level program.
> 30 million is big number. It needs many computers.
> Computer Go community may be able to share this work.
> I can offer Aya, it is not open-source though. Maybe Ray(strongest open
> source so far)  is better choice.
> 
> Thanks,
> Hiroshi Yamashita
> 
> - Original Message - From: 
> To: 
> Sent: Friday, January 06, 2017 4:50 PM
> Subject: Re: [Computer-go] it's alphago
> 
> 
> Competitive with Alpha-go, one developer, not possible. I do think it is
> possible to make a pro level program with one person or a small team.
> Look at Deep Zen and Aya for example. I expect I’ll get there (pro
> level) with Many Faces as well.
> 
> David
> 
> ___
> 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