Life In 19x19

I'll just drop this here. Analyzed (by Deep Zen) most recent game between Iyama and Ke. More info at viewtopic.php?f=13&t=13751&p=225118#p225118

johnsmith wrote:Thanks! Game between Ke Jie and Iyama is on on Fox. FineArt says W has a lead of 54.4% after the first move

pookpooi wrote:The 54.4% winrate for white come from black play 3-4 opening so we can't use this number

But now we know how large the grain of salt has to be.

Here are some more games of Zenith 7, this time against Zenith 6.
It seems that the 7d level is a little bit slower in Zenith 7, but also stronger (at least against Zenith 6 with level 7d).
On my machine Zenith 7 mostly used 3 or 4 seconds per move, in rare cases up to 6 seconds.









Zenith 7 in 9d mode against Zenith 6 in 7d mode—easy game for Zenith 7.
7d is the highest selectable level in Zenith 6.
Zenith 7 used between some seconds up to 20 seconds per move.

Interesting games, thanks for sharing. In the last game Zen 7 took the lead after 86 moves, just wow.

johnsmith wrote:I'll have to correct myself, seems there will be some new features. Link provided by pookpooi, google translation is:

Seventh Battle - It is equipped with this work, "7th game" which was loaded from previous work and gained popularity. Every time you win, your opponent's skill is raised, images and comments are changed by victory or defeat, so you can enjoy details.
Self-battle score - "Self-battle score" that became a topic with Alpha Go. Let the computer fight against each other, and you can obtain game cards that will serve as a model.
Review mode - In addition to your own game record, you can load the game of Go AI or the game of professional game and read it. It shows the evaluation value of the start and candidate hand of "Zen".

I am really looking forward to seeing all these new features, especially Review mode. I think it's totally worth the money.
As for leela, which is a different topic - download engine only from leela's website, download Sabaki editor for example, and just load that engine to the editor.

I just bought the English version of Zen 7 today, and have not seen any of these new features. Interestingly, the description for the English version does not mention any of these features. It only states:

Zenith Go 7 is a full-fledged Go software that is equipped with the pro level, record-breaking computer Go program Zen.
Its playing level is 9 dan. Depending on the user’s computer hardware, with the Specify Time setting, even higher playing levels can be reached.
In addition to a high level of play, the software is packed with features such as best-of-seven matches with the 25th Honinbo Cho Chikun and Mannami Nao 3 dan, and a collection of One-glance Tsumego.

Please enjoy our ever-improving full-scale Go software

Could it be that the review function is only available in the Japanese version? This would be hugely disappointing as I had purchased Zen 7 for the review function.

Caesura wrote:

johnsmith wrote:I'll have to correct myself, seems there will be some new features. Link provided by pookpooi, google translation is:

Seventh Battle - It is equipped with this work, "7th game" which was loaded from previous work and gained popularity. Every time you win, your opponent's skill is raised, images and comments are changed by victory or defeat, so you can enjoy details.
Self-battle score - "Self-battle score" that became a topic with Alpha Go. Let the computer fight against each other, and you can obtain game cards that will serve as a model.
Review mode - In addition to your own game record, you can load the game of Go AI or the game of professional game and read it. It shows the evaluation value of the start and candidate hand of "Zen".

I am really looking forward to seeing all these new features, especially Review mode. I think it's totally worth the money.
As for leela, which is a different topic - download engine only from leela's website, download Sabaki editor for example, and just load that engine to the editor.

I just bought the English version of Zen 7 today, and have not seen any of these new features. Interestingly, the description for the English version does not mention any of these features. It only states:

Zenith Go 7 is a full-fledged Go software that is equipped with the pro level, record-breaking computer Go program Zen.
Its playing level is 9 dan. Depending on the user’s computer hardware, with the Specify Time setting, even higher playing levels can be reached.
In addition to a high level of play, the software is packed with features such as best-of-seven matches with the 25th Honinbo Cho Chikun and Mannami Nao 3 dan, and a collection of One-glance Tsumego.

Please enjoy our ever-improving full-scale Go software

Could it be that the review function is only available in the Japanese version? This would be hugely disappointing as I had purchased Zen 7 for the review function.

Just use Leela(it's much better) and also help us to improve Leela Zero

Caesura wrote: Could it be that the review function is only available in the Japanese version? This would be hugely disappointing as I had purchased Zen 7 for the review function.

You can open an existing game and click the Analyze button. Then you can see Zen's five best candidate moves in the analysis window. I think thats what they called review function.

goame wrote: Just use Leela(it's much better) and also help us to improve Leela Zero

Leela is great and it's free, but Zen 7 is much stronger. Leela sometimes doesn't understand life and death situations.

Waylon wrote:

Caesura wrote: Could it be that the review function is only available in the Japanese version? This would be hugely disappointing as I had purchased Zen 7 for the review function.

You can open an existing game and click the Analyze button. Then you can see Zen's five best candidate moves in the analysis window. I think thats what they called review function.

Oh, they are the same thing? I had the impression from the following quote from Gomoto on the first page of this thread that review mode was different (or at least an improved version of) analyse function:

Zenith 6 is already very strong and very useful for every amateur go player.
(Much much stronger is relative, but any increase is nice and welcome here)

Review mode is not really new:
You can do this with "Analyze" in Zenith 6 already. It is a very nice feature indeed.
Move through the game (for example with mouse wheel) and watch win value and move ideas by zen

Waylon wrote:

goame wrote: Just use Leela(it's much better) and also help us to improve Leela Zero

Leela is great and it's free, but Zen 7 is much stronger. Leela sometimes doesn't understand life and death situations.

Then you hardware configuration is wrong.
Leela is much stronger, that's also the reason why nearly no one is buying Zen 7.
https://sjeng.org/leela.html

goame wrote: Then you hardware configuration is wrong.
Leela is much stronger, that's also the reason why nearly no one is buying Zen 7.
https://sjeng.org/leela.html

This is a position from a game between Leela 0.10.0 (black) and Zen 6. The black dragon on the right side is dead. Zen 6, Zen 7 and Crazy Stone see this almost immediately.

After more than 800,000 playouts, Leela 0.11.0 believes blacks winning chances are at 52%. This has nothing to do with a wrong hardware configuration.

goame, the reason why YOU dont buy Zen 7 is, you do not want to spend the money. And you dont have to.

I for one am quite happy to spend a small amount of money on several nice go engines.

I also spent some money on human go teachers and loved to do so (and will continue to do so from time to time). Go engines are much cheaper than a real teacher.

Go is my main pastime and I would be stupid if I would not invest some of my resources for my passion.

But it is very nice you dont have to if you dont like to. I appreciate Leela and the other free go engines as well.

Just some unorganized notes for anyone who might be interested in buying Zen 7 and looking for info:

By my understanding and impressions of both, Zen 7 is significantly stronger than Leela 0.11 when using equal hardware, but Zen 7 commercial release doesn't support GPU, unlike Leela. So if you have a laptop like mine without any good graphics card anyways, Zen 7 is definitely stronger, but if you have a decent graphics card, Leela might well be as strong or stronger because Zen will be unable to use that hardware.

On CPU, stylewise Zen 7 so far seems on average a little more stable and self-consistent in its evaluations and very steady. Leela is more dynamic and prone to change its mind as you play down a variation, and continues to have occasional blind spots or misevaluations just like the older 0.10 Leela, although they are somewhat less frequent. I don't have any experience with GPU-accelerated Leela, but I would expect with the large speedup of a strong GPU it would continue to have rare large misevals in situations where the search isn't sufficient to correct it, but I could easily see the deeper search making it much sharper and competitive with Zen 7 in fights that it doesn't miseval.

Zen 7 doesn't appear to have a nice "offline" review interface right now, or at least I haven't found it, but I've found it okay for interactive review of games. The interface is a bit lacking through. By contrast, with Leela 0.11 GTP version there are scripts you can find online that will let you do things like have Leela deep-search your game SGF overnight and annotate it with what it thinks your mistakes were, as well as plug them into GUIs that let you navigate and make reviewing easy.

However, with Leela I always have to take its evals with a grain of salt when they get in certain kinds of capturing races or life and death due to its blind spots. From experience I have a sense now of when it's more likely to do this and in those cases will often interactively go to the position and "test" Leela by playing the moves to resolve a situation or that I think might expose a blind spot, before I can trust the eval. I've used Zen less, but I haven't noticed such major problems with Zen.

Also, one last detail - Leela has an interesting behavior when reviewing handicap games. For example, in no-komi games it thinks *white* has the advantage early on, and in low-handicap games it thinks Black has none or only a small advantage. This is presumably because it was trained on human games where in practice, no-komi games often are won by white since they are underhandicapped, etc. But this makes it hard to use Leela to "objectively" evaluate the position sometimes.

And yet I don't think you can always rely on it to give you a reasonable value of things like "how likely is black to win given that he is 2 stones weaker than white" either. Because from what I can tell from what gcp has said, Leela doesn't actually know that for example black is 2 stones weaker than white. It simply infers this from seeing the board position with black having the extra hoshi stone. But as the board fills up and captures happen, it's no longer clear from the board position alone that it was originally a handicap game, so I think Leela in some unknown and hard-to-predict way will gradually transition to no longer evaluating as if the players are different in strength.

Zen seems to report values closer to objective, at least in 2 and 3 stone games unlike Leela it reports a strong advantage for Black at the start. I haven't tried using Zen on a no-komi game yet.

I haven't see any Leela configuration surpass Zen-15.3-10c in http://www.yss-aya.com/cgos/19x19/bayes.html yet. So even without GPU advantage I believe Zenith Go 7 is still stronger than Leela 0.11. Of course you can prove me wrong by running your strongest Leela in CGOS at least 100 games.

My laptop doesn't have 10 cores either...

Anyways, going by the rough numbers in CGOS for overall strength Zen on CPU and Leela on strong GPU are close enough that any remaining difference is no big deal compared to all the other things I mentioned.