Life In 19x19

This is basically to Remi C.

1) When will CS 2015 come out for Windows? Will CS 2013 drop in price and how much? (80$ is just pretty expensive though not overpriced)

2) Can you analyze the current FJ vs. Zen challenge like you did one year ago with your CS using deep analyzation mode?

Pippen wrote:This is basically to Remi C.

1) When will CS 2015 come out for Windows? Will CS 2013 drop in price and how much? (80$ is just pretty expensive though not overpriced)

I don't think Remi has much control over this. He provided the Go Engine and Unbalance are responsible for the GUI and pricing. I've found reporting bugs to Unbalance not too helpful (saying they'll fix it, and then NOT), but have been able to speak to Remi via his email address which you can find on the web.

Not sure how much CS 2015 is just a UI revamp, without much increase in Engine strength?

As longshanks wrote, Unbalance is in charge of selling the program, making new releases, customer support, etc. I'd rather not interfere with their communication.

As far as I know, no new version of the English Crazy Stone is planned for 2015. They made a new Japanese version. The main difference in the new Japanese version is that I managed to make weak levels that are much more human-like, in order to make the program more pleasant for beginners. Strength improved since 2013, but not enough to be worth releasing a new English version. They will release a new English version when there is a big strength improvement. I don't know yet when this will happen, but I am working on it.

The new deep learning version of Crazy Stone will be released on May 16th.

It's not yet possible to buy it, but its web page is already up:
http://www.unbalance.co.jp/igo/eng/contents.html

and you can check the demo video on youtube:

I wanna see the rematch between CrazyStone deep learning edition and Franz-Josef Dickhut, for once and for all

A question: Does Crazystone run on linux under wine?

Phobos wrote:A question: Does Crazystone run on linux under wine?

No, sorry, it is Windows only. Maybe a virtual machine would work better, but I am not even sure.

Marcel Grünauer wrote:This sounds very exciting!

On faster hardware, does "Crazy Stone Deep Learning" play better or just faster?

I'm asking because the iPad version has been said to have the same strength on all models.

The top levels are in fact defined as time per move, so they will be stronger with more powerful hardware.

How are weaker levels calibrated? Is it purely a function of time spent per move?

Kirby wrote:How are weaker levels calibrated? Is it purely a function of time spent per move?

No. Weaker levels play instantly.

All the levels up to 1k are pure neural networks. They play almost instantly. I collected game records of weak players from KGS, and produced neural networks that imitate them. So the weak levels are considerably more human-like than before. Strength was calibrated by playing games on KGS.

As you can see on the attached sgf, it produces a very human-like way of being weak. I am really happy with the result.

It is interesting that the 13k neural network was produced by learning from game records of 15k players. I suppose it may have become stronger because it plays instantly, and because it can read ladders better.

Thanks, Rémi. That's very fascinating. I think making the weaker levels "naturally weak" is a great idea.

How do levels between 1d and 7d work?

Kirby wrote:Thanks, Rémi. That's very fascinating. I think making the weaker levels "naturally weak" is a great idea.

How do levels between 1d and 7d work?

1d and beyond is MCTS as usual, except that the neural network makes it a lot stronger.

Can it be used in studying mode - meaning setting up any arbitrary position and ask it to play either white or black from there on?

sorin wrote:Can it be used in studying mode - meaning setting up any arbitrary position and ask it to play either white or black from there on?

Yes, my CS2012 is able to do so, and I do not assume that this feature has been erased in the meantime.

I created the starting position with another SGF editor, opened this SGF with CS, and everything went smoothly.
Seems to be the same as opening any other SGF (containing a "real" game), and asking CS about "his" suggestions for move 123.