Life In 19x19

See previous answers. Your 4 is thé wrong hane. Then a diagonal decent nu white threatens to connect or make a 2nd eye

Even with 4 at kosumi white dies.

https://cdn.discordapp.com/attachments/709013107661602906/961731020682580038/unknown.png

I've submitted 8 variations of solutions in this ...

I added this tsumego to 101weiqi, please take a look and give it a rating!

https://www.101weiqi.com/auditquestion/318584/

Thanks, I want to put this problem on a tsumego site, but I have no experience in making tsumego. How should I simplify this position into a tsumego?

Seems like this is a hard reading problem, plus the time pressure and the whole board position in a live game makes it even harder. I think black can kill in this way, is this correct?

$$ Solution 1
$$ . O X . . X . . . . . . . . |
$$ . . X X X O X X . . . X . . |
$$ . X O . 9 O O X ...

I've been practising a lot of tsumego but missed this kill in a game. Wondering if I should be able to solve it (meaning I should learn to recognize tsumego in live games) or if it's above my rank?

Black to kill:

Servers down? It says "No more instances available. You may want to try a different gpu server."

A normal move, -90% winrate. Why?

You do have a point. And just for clarity, I do think KataGo would actually be playing that move if it had to make one, even though it has a low amount of playouts.

Afaik, by default KataGo just takes the move with the highest LCB, as long as it has more than 10 playouts. Although, I wouldn't be ...

Sometimes a low playout high winrate move gets picked (N11 in this case) instead of the good move (L18).

Problem with your averaging: both bots has H16 as top choice but B11 is chosen overall. H16 is shown in orange while B11 is in blue.

Some strong bots have been updated:
AQ v2.1.1
Leela 0.11.0
Ray 4.32
Pachi 12.00

Leela Zero is much stronger now, can you test it again please?

Zen 7 is really strong, even though it only uses cpu. Here are two games against Leela 0.11.0 and AQ 2.0.3. Hardware: i5-2500K, GTX 1060

(;GM[1]FF[4]CA[UTF-8]AP[Sabaki:0.31.5]KM[7.5]SZ[19]DT[2017-12-23]PB[Leela 0.11.0]PW[Zen 7]RE[W+Resign]
;B[cd]
;W[pc]
;B[dq]
;W[pp]
;B[ec]
;W[do]
;B[co]
;W[cn]
;B ...

Satorian wrote:Ah, OK. Makes sense. Good observation.

I'll give that Linux Leela a fresh run now as the script seems working fine. Going to run the whole night and probably until tomorrow evening.

It's L19_Sat_Leela_v2.

Are you running leela 0.10.0 or 0.10.5?

On a related note, Leela doesn't seem to utilize resources very well on Windows. Right now I'm running Leela_Test, and CPU load rarely exceeds 70% at peak, GPU load rarely exceeds 80%. RAM usage is at about 250 MB and GPU memory at 474 MB (including OS and other apps). Also, it doesn't seem to ...