I'll try to address the issues brought up as best as I can.
Regarding the game time messages (opponent resigned, please capture dead
stones, etc) not being easy to notice, I'm working on it. However the
language barrier shouldn't be a problem as the website recognizes locales
of jp, cn, kr and is
Nice, Thanks Henry!
I am in the process of bringing up my first bot(as are a bunch of people)
so I will probably try and connect in the next couple of days.
--Danny
On Mon, May 23, 2016 at 8:46 AM, Petr Baudis wrote:
> Hi!
>
> On Sun, May 22, 2016 at 08:56:09AM +, Henry
Hi!
On Sun, May 22, 2016 at 08:56:09AM +, Henry Hemming wrote:
> Hello, I would like to invite all you go bot developers to my new go server.
>
> http://goratingserver.appspot.com
I tried a game last evening too. I didn't mind not being able to
choose my nickname too much (I just
Hi Henry,
Thanks for your message. I tried your server (as a human player, not a bot). It
might become an interesting alternative to CGOS, mixing humans and bots
together.
I find it annoying that I cannot choose my pseudo. I understand that you might
not want to worry about moderating
On 23-05-16 13:57, "Ingo Althöfer" wrote:
> Hi Gian-Carlo,
>
>> Unsurprisingly, self-play favors extreme selectivity, but this does not
>> hold against other opponents.
>
> is this just your personal experience, or are there systematic experiments on
> this?
> Is it true "only" for MCTS (and
Hi Gian-Carlo,
> Unsurprisingly, self-play favors extreme selectivity, but this does not
> hold against other opponents.
is this just your personal experience, or are there systematic experiments on
this?
Is it true "only" for MCTS (and vairants) or also for game tree search in chess?
Ingo.
On 22/05/2016 23:07, Álvaro Begué wrote:
> Disclaimer: I haven't actually implemented MCTS with NNs, but I have
> played around with both techniques.
>
> Would it make sense to artificially scale down the values before the
> SoftMax is applied, so the probability distribution is not as
>
I think one of the main problems is that the network learns good replies
to good moves. The training set does not have good replies to bad moves,
but the search tree is full of bad moves that need to be punished.
Alvaro's suggestion looks good. This is one of the experiments I want to
try.