Life In 19x19 http://prod.lifein19x19.com/ |
|
5 kyu Wbaduk Fighting Play Questions http://prod.lifein19x19.com/viewtopic.php?f=4&t=15976 |
Page 1 of 1 |
Author: | eksert [ Sat Aug 18, 2018 2:53 am ] |
Post subject: | 5 kyu Wbaduk Fighting Play Questions |
The questions are inside the sgf file. Thanks for your comments in advance. I was white this game. The link is at http://eidogo.com/#pMDVeJb5. |
Author: | Uberdude [ Sat Aug 18, 2018 2:57 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
Quote my post to see how to use sgf tags (it needs either a url to download link of sgf, which for eidogo is http://eidogo.com/sgf/<id without leading #>.sgf, or the sgf content itself). |
Author: | eksert [ Sat Aug 18, 2018 6:06 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
Thank you @uberdude. |
Author: | BlindGroup [ Sat Aug 18, 2018 9:52 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
One thing that really jumps out at me in this game is that I think you understand that it is good to keep your opponent's groups weak and separated. That's a good thing to understand. However, I feel like you are over-estimating the strength of your own groups at times. These two moves felt uncomfortable to me: Move 26: My instinct would have been to play something like A because it reinforces the white stones in the middle left and it starts to limit the options of the black stone. Black has two weak groups, but I think you should also realize that your group in between is also not strong either. Move 28: I'm not sure that the corner is strong enough to make this pincer work. If the corner were strong, then this move would separate blacks approach stone from the stones on the right. However, if you see the corner as weak, the black stone is now splitting two weak white groups. I'd have played something to reinforce the corner like A. This forces black to extend to the left to prevent the pincer, and then you get sente. A lot of complications that developed in this area would have been eliminated if the corner were secure. Unfortunately, I wasn't entirely sure what to suggest for the rest of the game. It gets complicated quickly with so may weak groups, and I fear that your group in the lower left corner isn't settled even after ![]() |
Author: | Joaz Banbeck [ Sat Aug 18, 2018 1:21 pm ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
BlindGroup wrote: ... play something like A because it reinforces the white stones in the middle left and it starts to limit the options of the black stone. Black has two weak groups, but I think you should also realize that your group in between is also not strong either. ... I like eksert's choice here. A move at 'a' looks small, and has iffy shape. Sure, white's two stones on the side are not really strong, but they are stronger than the local black stones. The shoulder hit helps keep one of those stones weak. If you really want to indulge in the goals behind 'a', this will do it better: |
Author: | Gomoto [ Sun Aug 19, 2018 2:28 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
I prefer this double 4-4 fuseki for white: If you play the 3-4 fuseki you have to add a stone add a or b in this fuseki, to enter the right side and approach 4-4 is a mistake: (If you would have played the double 4-4 fuseki you could consider approaching the upper right from the OUTSIDE or enter 3-3 directly (which would have been my preference)) This kind of passive play in the middle of the side is not a good move in the fuseki most of the times. It is a mistake in this position. Much better is a play at a,b or c. This move is fine: a and b (if you insist on a local play) are good options for white: |
Author: | EdLee [ Sun Aug 19, 2018 3:07 am ] |
Post subject: | |
Hi Gomoto, |
Author: | Gomoto [ Sun Aug 19, 2018 3:37 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
They are good, the winrates ![]() |
Author: | Gomoto [ Sun Aug 19, 2018 3:41 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
B15 by black is a mistake, the white answer 86 (turn around in the middle 85) turn arround in the middle 92 attachment 85 (ELF) |
Author: | EdLee [ Sun Aug 19, 2018 5:28 pm ] |
Post subject: | |
Hi Gomoto, Thanks. ![]() ( I assume 'good' means it's either LZ's top choice or one of its top 3 choices for the given board. ![]() |
Author: | Bill Spight [ Sun Aug 19, 2018 7:48 pm ] |
Post subject: | Re: |
EdLee wrote: Hi Gomoto, Thanks. ![]() ( I assume 'good' means it's either LZ's top choice or one of its top 3 choices for the given board. ![]() Without many playouts, you can't count on Leela's third choice, or maybe even the second choice. |
Author: | EdLee [ Mon Aug 20, 2018 1:58 am ] |
Post subject: | |
Quote: Without many playouts, you can't count on Leela's third choice, or maybe even the second choice. Or not even the first choice ?...
|
Author: | eksert [ Thu Aug 23, 2018 2:32 am ] |
Post subject: | Re: Re: |
Bill Spight wrote: EdLee wrote: Hi Gomoto, Thanks. ![]() ( I assume 'good' means it's either LZ's top choice or one of its top 3 choices for the given board. ![]() Without many playouts, you can't count on Leela's third choice, or maybe even the second choice. How many playouts 1k is enough generally? I analyze my games with LZ and after 1k generally moves do not change.. Only win percentages change 0.5%-1%. |
Author: | EdLee [ Thu Aug 23, 2018 3:06 am ] |
Post subject: | |
Quote: How many playouts 1k is enough generally? Interesting question. My understanding is it requires much more advanced hardware during the neural net training phase, but much less hardware requirements afterwards, during the playing ?
|
Author: | dfan [ Thu Aug 23, 2018 4:53 am ] |
Post subject: | Re: Re: |
eksert wrote: How many playouts 1k is enough generally? I analyze my games with LZ and after 1k generally moves do not change.. Only win percentages change 0.5%-1%. LZ with 1k playouts is certainly quite strong, but it is incorrect to believe that the moves don't change with more visits. People have done many experiments, and LZ with more playouts beats LZ with fewer playouts consistently. My recollection is that it's a logarithmic sort of relation, meaning that doubling the amount of time searching adds a fixed amount of strength, so you do get diminishing returns after a while (going from 1k to 2k playouts adds about the same amount of strength as going from 100k to 200k). |
Author: | Uberdude [ Thu Aug 23, 2018 7:03 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
There are certainly some situations where LZ needs 10s of thousands of playouts to find the better move, and in some of these (e.g. involving ladders) I can do so in a matter of seconds. Of course even with these inferior moves its overall strength is still very high, but on 1k playouts there will be rare times it suggests a move 4d me can quickly and confidently say is wrong. Presumably a stronger player than myself would find more such occurrences, though care is needed your judgement isn't based on wrong human knowledge. In these ladder cases you can escape from the ladder LZ mistakenly thinks is working and see its winrate plunge as it realises its delusion. |
Author: | Bill Spight [ Thu Aug 23, 2018 11:10 am ] |
Post subject: | Re: Re: |
eksert wrote: Bill Spight wrote: EdLee wrote: Hi Gomoto, Thanks. ![]() ( I assume 'good' means it's either LZ's top choice or one of its top 3 choices for the given board. ![]() Without many playouts, you can't count on Leela's third choice, or maybe even the second choice. How many playouts 1k is enough generally? I analyze my games with LZ and after 1k generally moves do not change.. Only win percentages change 0.5%-1%. If you are talking about Leela, not Leela Zero, with only 1,000 playouts I would have doubts about even the first choice. ![]() Leela Zero is more accurate, but I would still want to run it with at least a 10k setting. |
Author: | Bill Spight [ Thu Aug 23, 2018 11:13 am ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
Uberdude wrote: There are certainly some situations where LZ needs 10s of thousands of playouts to find the better move, and in some of these (e.g. involving ladders) I can do so in a matter of seconds. Of course even with these inferior moves its overall strength is still very high, but on 1k playouts there will be rare times it suggests a move 4d me can quickly and confidently say is wrong. Presumably a stronger player than myself would find more such occurrences, though care is needed your judgement isn't based on wrong human knowledge. In these ladder cases you can escape from the ladder LZ mistakenly thinks is working and see its winrate plunge as it realises its delusion. I only have some preliminary results, but Leela Zero run at a 200k setting can come up with a different first choice than Leela Zero run at a 100k setting more than 15% of the time. ![]() |
Author: | Uberdude [ Thu Aug 23, 2018 12:24 pm ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
Bill Spight wrote: I only have some preliminary results, but Leela Zero run at a 200k setting can come up with a different first choice than Leela Zero run at a 100k setting more than 15% of the time. ![]() How often does Leela Zero run at 100k come up with a different 1st choice than Leela Zero run at 100k another time? i.e. how consistent is it under the same settings. Control groups ![]() |
Author: | Bill Spight [ Thu Aug 23, 2018 1:03 pm ] |
Post subject: | Re: 5 kyu Wbaduk Fighting Play Questions |
Uberdude wrote: Bill Spight wrote: I only have some preliminary results, but Leela Zero run at a 200k setting can come up with a different first choice than Leela Zero run at a 100k setting more than 15% of the time. ![]() How often does Leela Zero run at 100k come up with a different 1st choice than Leela Zero run at 100k another time? i.e. how consistent is it under the same settings. Control groups ![]() Dunno. I am waiting both for having more time and more powerful hardware to do some real research. {shrug} |
Page 1 of 1 | All times are UTC - 8 hours [ DST ] |
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |