Page 1 of 6
LZ help for position
Posted: Fri Oct 19, 2018 6:48 am
by Pippen
Can somebody run LZ for an analysis of the position below (Black to play) to check the difference with Leela 11 (see pic)? Just a couple of minutes. I hope LZ also offers something similiar of an Analysis Window like Leela to see deeper into its thought process.
- Leela.jpg (252.05 KiB) Viewed 18027 times
Re: LZ help for position
Posted: Fri Oct 19, 2018 7:18 am
by Bill Spight
Looks like it's hard for Black to go wrong in this position.
Re: LZ help for position
Posted: Fri Oct 19, 2018 7:51 am
by Pippen
Bill Spight wrote:Looks like it's hard for Black to go wrong in this position.
Well, first it's interesting how LZ evaluates this position since Leela thinks it's White's favour already. Secondly, I like to see what LZ proposes: g3/f3 or d5/c5 or something completely different. I think d5 is best here, but I have a feeling a brutal tenuki might be even better.
Re: LZ help for position
Posted: Fri Oct 19, 2018 7:55 am
by Tryss
A LZ analysis at around 100k playouts with network 181 :
(;GM[1]FF[4]CA[UTF-8]AP[CGoban:3]ST[2]
RU[Chinese]SZ[19]KM[7.00]
PW[White]PB[Black]AW[qp][dq][lq]AB[dd][pd][jq]LB[co:B][do:E][dp:F][op:A][pp:D][nq:G][oq:C]TR[np][gq]C[This analysis is done with LZ 0.15, network #181
The total amount of playouts is close to 100k
A to G are the only moves truly considered, the triangles have less than 100 playouts. The winrates for all of these moves are between 45.7% and 46.8% ]
(;B[op]C[LZ first choice :
46.8% winrate with 85k playouts
Her idea is interesting, see the continuation ]
;W[oq]
;B[nq]
;W[pq]
;B[np]
;W[qn]
;B[co]C[Play this exchange now]
;W[cp]
;B[do]
;W[fq]
;B[jo]C[And reinforce the group. M4 was also explored.]
;W[cc]C[White invade with the sente, after that, it start to be too deep to be meaningful, because there is many joseki choices possible here.])
(;B[co]C[LZ second choice :
46.3% winrate with 10k playouts
]
;W[fp]
;B[op]C[But now, white response is different ]
;W[lo]
;B[qq]C[And black settle in the corner, but it start to be too deep to be reliable (her "best" continuation for reference) ]
;W[pq]
;B[pp]
;W[qr]
;B[rq]
;W[qo])
(;B[oq]C[LZ third choice
45.7% winrate with 2.4k playouts
A very different idea from A and B
Not sure up to what depth the following sequence is reasonnable]
;W[np]
;B[qq]
;W[rq]
;B[op]
;W[qr]
;B[no]
;W[mo]
;B[nn]
;W[mn]
;B[mm]
;W[nm]
;B[pn]
;W[nq]
;B[lm]
;W[kn]
;B[nr])
(;B[pp]C[LZ fourth choice
46.0% winrate with 1.3k playouts
Here LZ wants to create a running "fight"]
;W[pq]
;B[oq]
;W[qq]
;B[co]C[A tenuki/probe]
;W[fp]
;B[op]
;W[pn]
;B[kp]C[Close to the ]
;W[mp]
;B[on]
;W[mn]
;B[om]
;W[pm]
;B[ol]
;W[qk])
(;B[do]C[45.8% winrate, 1.1k playouts, ]
;W[co]
;B[cn]
;W[cp]
;B[dn]
;W[fq]
;B[op]
;W[oq]
;B[nq]
;W[pq]
;B[np]
;W[qn]
;B[cf])
(;B[dp]C[46.3% winrate, 849 playouts]
;W[cp]
;B[eq]
;W[cq]
;B[op]
;W[oq]
;B[nq]
;W[pq]
;B[np]
;W[qn]
;B[ep]
;W[cn]
;B[em]
;W[cc])
(;B[nq]C[45.7% winrate, 780 playouts]
;W[pq]
;B[co]
;W[fp]
;B[po]
;W[rn]
;B[hq]
;W[mp]
;B[pp]
;W[qq]
;B[np]
;W[mn]
;B[no]
;W[mo]
;B[ol]
;W[ml]
;B[qm]))
Note that on an empty board, Black has 46.5% winrate (basically, it didn't move much, the splitting move was a little over -2%, but the white answer balanced this
Re: LZ help for position
Posted: Fri Oct 19, 2018 8:47 am
by Pippen
Thx a lot. Wow, would never have thought of playing on the right lower side.
Is 100K playouts much or little in terms of analysis depth?
Re: LZ help for position
Posted: Fri Oct 19, 2018 9:07 am
by Tryss
It's a lot. LZ training games and matches are played at 3200 playouts.
Re: LZ help for position
Posted: Fri Oct 19, 2018 12:03 pm
by Uberdude
Pippen: 1 LZ playout is worth a lot more than 1 Leela simulation (and not just because LZ is stronger), they aren't equivalent as a "unit of work". Maybe a factor of 10 difference? Also LZ's principal variations stay sensible far deeper into the tree whereas Leela's tend to turn to junk maybe 6-10 moves deep.
Re: LZ help for position
Posted: Fri Oct 19, 2018 1:28 pm
by Pippen
Tryss wrote:
Note that on an empty board, Black has 46.5% winrate (basically, it didn't move much, the splitting move was a little over -2%, but the white answer balanced this
Maybe interesting to see what White should have answered instead of M3 and Leela's ideas. In any way my Split-Fuseki seems far more competitive than even I thought. It basically loses nothing, even in a scenerio like here where virtually nobody except me ever plays it (one pro game
ever^^). I can't wait till LZ is available without all that installing mambo-jambo to explore it myself. BTW: Any idea when LZ might be available like Leela 11 with or without payment? Did the developer say something?
Re: LZ help for position
Posted: Fri Oct 19, 2018 3:05 pm
by Tryss
LZ split fuseki possible continuation :
$$Wcm6 after
, white play at a or b
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . b . . |
$$ | . . . X . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . a . . |
$$ | . . 6 . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 2 4 . . . . . . . . . . . . . . . |
$$ | . . 3 , . . . . . , . . . . . , O . . |
$$ | . . . O . 5 . . . X . . . 1 . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------
- Click Here To Show Diagram Code
[go]$$Wcm6 after , white play at a or b
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . b . . |
$$ | . . . X . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . a . . |
$$ | . . 6 . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 2 4 . . . . . . . . . . . . . . . |
$$ | . . 3 , . . . . . , . . . . . , O . . |
$$ | . . . O . 5 . . . X . . . 1 . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]
But what Installing mambo-jumbo are you talking about? If you use windows and have a GPU
Step 1) download Lizzie (the windows GPU version) and unzip it to the folder you want :
https://github.com/featurecat/lizzie/releases/tag/0.5Launch lizzie.jar to start LZ with a nice interface
Step 2 (optionnal, but won't hurt) download and replace the Network file with the latest one. You can get it here :
http://zero.sjeng.org/best-network
Re: LZ help for position
Posted: Fri Oct 19, 2018 3:25 pm
by Gomoto
- badforblack.jpg (154.89 KiB) Viewed 17912 times
Re: LZ help for position
Posted: Fri Oct 19, 2018 6:32 pm
by Kirby
Elf v1 prefers black here (prefers black from empty board, too), but recommends the same move as LZ:
Re: LZ help for position
Posted: Fri Oct 19, 2018 6:37 pm
by Kirby
Also, even though Elf likes black at the beginning, it thinks black's split is a bad move, and then thinks white is winning:
Re: LZ help for position
Posted: Fri Oct 19, 2018 6:41 pm
by Kirby
All this being said, I will say that I've been screwing around with Elf lately, and I noticed when replaying the Lee Sedol - AlphaGo game 4, where Lee Sedol won...
Elf obviously thinks AlphaGo is majorly winning around the middle of the game, but after Lee Sedol's "divine move" (of 78), which doesn't actually work, Elf still thinks AlphaGo is winning. And while it's not Elf's top choice, AlphaGo's response still has a 95% or so winrate. In other words, even after showing Elf AlphaGo's mistake, Elf, like AlphaGo, still thinks that black was winning the game. As the game plays out, like AlphaGo, Elf's winrate dropped significantly when it finally realized that the result had changed.
So these types of analyses are interesting, but it's also interesting to note the mistakes that they make sometimes.
Re: LZ help for position
Posted: Fri Oct 19, 2018 11:47 pm
by Pippen
Tryss wrote:LZ split fuseki possible continuation :
$$Wcm6 after
, white play at a or b
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . b . . |
$$ | . . . X . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . a . . |
$$ | . . 6 . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 2 4 . . . . . . . . . . . . . . . |
$$ | . . 3 , . . . . . , . . . . . , O . . |
$$ | . . . O . 5 . . . X . . . 1 . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------
- Click Here To Show Diagram Code
[go]$$Wcm6 after , white play at a or b
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . b . . |
$$ | . . . X . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . a . . |
$$ | . . 6 . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . 2 4 . . . . . . . . . . . . . . . |
$$ | . . 3 , . . . . . , . . . . . , O . . |
$$ | . . . O . 5 . . . X . . . 1 . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]
What evaluation does LZ give Black after 6 and then for the position in the diagram? (remark for me: 46.5% for Black on an empty board)
Re: LZ help for position
Posted: Fri Oct 19, 2018 11:52 pm
by Pippen
I have Windows Vista and as far as I know it's not working there. So I would have to go some way to make LZ working...too much effort.
