Life In 19x19 http://www.lifein19x19.com/ |
|
Engine Tournament http://www.lifein19x19.com/viewtopic.php?f=18&t=13322 |
Page 20 of 20 |
Author: | q30 [ Sat Mar 25, 2023 3:35 am ] |
Post subject: | Re: Engine Tournament |
New type KataGo weight file (b18c384nbt-uec.bin) is a bit stronger than the 2022 year winner of its "middleweight" category (kata1-b40c256-s12350780416-d3055274313 - b18c384nbt-uec 9 - 11) and markedly stronger than "heavyweight" one: kata1-b60c320-s6782286336-d3070935549 - b18c384nbt-uec 3 - 17 (details). New rate of KataGo strongest in their "categories" weight files: "bantamweight" <|= 2 ^ 23 B (< 12 MiB) - g170e-b10c128-s1141046784-d204142634.bin (6) "featherweight" 2 ^ 24 B (12 - 24 MiB) - I haven't "lightweight" 2 ^ 25 B (24 - 48 MiB) - g170e-b15c192-s1672170752-d466197061.bin (5) "welterweight" 2 ^ 26 B (48 - 96 MiB) - g170e|kata1-b20c256x2-s5303129600-d1228401921.bin (4) "middleweight" 2 ^ 27 B (96 - 192 MiB) - b18c384nbt-uec.bin (1) "light heavyweight" 2 ^ 28 B (192 - 384 MiB) - g170-b30c320x2-s4824661760-d1229536699.bin (3) "heavyweight" 2 ^ 29 B (384 - 768 MiB) - kata1-b60c320-s6782286336-d3070935549.bin (2) "super heavyweight" >|= 2 ^ 30 B (> 768 MiB) - I haven't |
Author: | q30 [ Sat May 13, 2023 5:55 am ] |
Post subject: | Re: Engine Tournament |
LeelaZero became a bit stronger: 57-41 in summary (details). |
Author: | q30 [ Sat Jul 01, 2023 12:28 am ] |
Post subject: | Re: Engine Tournament |
Pachi (DCNN) became a bit stronger, but still is weaker than Zen (details). |
Author: | q30 [ Sat Jul 29, 2023 7:11 am ] |
Post subject: | Re: Engine Tournament |
Pachi (--nodcnn) became a bit stronger than Ray (details). So the rating of Go engines without using GPU is next: Top level 1) KataGo 2) LeelaZero 3) SAI High level 4) Leela 5) Rayon 6) Zenith 7) Pachi_DCNN 8) Hiratuka Middle level 9) Pachi 10)Ray 11)MoGo |
Author: | q30 [ Sat Aug 12, 2023 12:06 am ] |
Post subject: | Re: Engine Tournament |
KataGo v1.13.0 is stronger (17-11) than previous version (details). |
Author: | q30 [ Sat Sep 09, 2023 2:42 am ] |
Post subject: | Re: Engine Tournament |
KataGo b18c384nbt-optimisticv13-s5971M.bin - b18c384nbt-uec.bin: 13 - 7 (details). So the rate of KataGo "weight categories" is next: "bantamweight" <|= 2 ^ 23 B (< 12 MiB) - g170e-b10c128-s1141046784-d204142634.bin (6) "featherweight" 2 ^ 24 B (12 - 24 MiB) - I haven't "lightweight" 2 ^ 25 B (24 - 48 MiB) - g170e-b15c192-s1672170752-d466197061.bin (5) "welterweight" 2 ^ 26 B (48 - 96 MiB) - g170e|kata1-b20c256x2-s5303129600-d1228401921.bin (4) "middleweight" 2 ^ 27 B (96 - 192 MiB) - b18c384nbt-optimisticv13-s5971M.bin (1) "light heavyweight" 2 ^ 28 B (192 - 384 MiB) - g170-b30c320x2-s4824661760-d1229536699.bin (3) "heavyweight" 2 ^ 29 B (384 - 768 MiB) - kata1-b60c320-s6782286336-d3070935549.bin (2) "super heavyweight" >|= 2 ^ 30 B (> 768 MiB) - I haven't |
Author: | q30 [ Sat Oct 28, 2023 3:30 am ] |
Post subject: | Re: Engine Tournament |
Pachi doesn't become stronger (details). |
Author: | q30 [ Sat Feb 17, 2024 12:47 am ] |
Post subject: | Re: Engine Tournament |
During last year KataGo weight files of "middleweight category" became stronger, but of "heavyweight" - not (details). The rate of KataGo "weight categories" with their strongest weight files: "bantamweight" <|= 2 ^ 23 B (< 12 MiB) - g170e-b10c128-s1141046784-d204142634.bin (6) "featherweight" 2 ^ 24 B (12 - 24 MiB) - I haven't "lightweight" 2 ^ 25 B (24 - 48 MiB) - g170e-b15c192-s1672170752-d466197061.bin (5) "welterweight" 2 ^ 26 B (48 - 96 MiB) - g170e|kata1-b20c256x2-s5303129600-d1228401921.bin (4) "middleweight" 2 ^ 27 B (96 - 192 MiB) - kata1-b18c384nbt-s8526915840-d3929217702.bin & kata1-b40c256-s12860905472-d3197353276.bin (1) "light heavyweight" 2 ^ 28 B (192 - 384 MiB) - g170-b30c320x2-s4824661760-d1229536699.bin (3) "heavyweight" 2 ^ 29 B (384 - 768 MiB) - kata1-b60c320-s6782286336-d3070935549.bin (2) "super heavyweight" >|= 2 ^ 30 B (> 768 MiB) - I haven't |
Author: | q30 [ Sat Mar 02, 2024 2:49 am ] |
Post subject: | Re: Engine Tournament |
KataGo became stronger: 18 - 12 in sparring of v.1.14.0 vs v.1.13.0 (details). |
Author: | q30 [ Sat Mar 02, 2024 2:56 am ] |
Post subject: | Re: Engine Tournament |
Ray didn't become stronger (details). |
Author: | q30 [ Sat Apr 20, 2024 7:57 am ] |
Post subject: | Re: Engine Tournament |
KataGo v. 1.14.1 - 1.14.0: 25 - 21 (details). |
Author: | q30 [ Sat May 04, 2024 2:37 am ] |
Post subject: | Re: Engine Tournament |
New KataGo weight files of "light heavyweight category" are weaker than only "middleweight" ones (details). |
Author: | q30 [ Sat May 04, 2024 6:02 am ] |
Post subject: | Re: Engine Tournament |
The current rate of KataGo "weight categories": "bantamweight" <|= 2 ^ 23 B (< 12 MiB) - g170e-b10c128-s1141046784-d204142634.bin (6) "featherweight" 2 ^ 24 B (12 - 24 MiB) - I haven't "lightweight" 2 ^ 25 B (24 - 48 MiB) - g170e-b15c192-s1672170752-d466197061.bin (5) "welterweight" 2 ^ 26 B (48 - 96 MiB) - g170e|kata1-b20c256x2-s5303129600-d1228401921.bin (4) "middleweight" 2 ^ 27 B (96 - 192 MiB) - kata1-b18c384nbt-s9131461376-d4087399203.bin (1) "light heavyweight" 2 ^ 28 B (192 - 384 MiB) - b28c512nbt-s5668008960-d4210144556.bin (2) "heavyweight" 2 ^ 29 B (384 - 768 MiB) - kata1-b60c320-s6782286336-d3070935549.bin (3) "super heavyweight" >|= 2 ^ 30 B (> 768 MiB) - I haven't |
Author: | xela [ Sat May 04, 2024 4:57 pm ] |
Post subject: | Re: Engine Tournament |
q30 wrote: New KataGo weight files of "light heavyweight category" are weaker than only "middleweight" ones (details). It looks like you're using very fast time limits. In slower games, the larger networks get more value from the extra time and will become relatively stronger. Compare the 1-minute and 5-minute rankings at https://lifein19x19.com/viewtopic.php?p=248817#p248817 and see how kata_20b overtakes kata_15b by a large margin given more time (and similar but less drastically for LZ188, 40 blocks versus LZ157, 15 blocks). Heavyweights are slower but more powerful :-) |
Author: | Mike Novack [ Sat May 04, 2024 6:10 pm ] |
Post subject: | Re: Engine Tournament |
And with computers, not absolute time but time and hardware. We really are going to have to come to some agreement about those. What will be considered a "standard machine" and what "standard time controls. And what sort of time controls? Allowing X per move is simple but I believe we will eventually want AI that can be making "time management" decisions << situation not critical; try to conserve time vs critical, use some of the saved time >> |
Author: | q30 [ Sat May 18, 2024 2:31 am ] |
Post subject: | Re: Engine Tournament |
Ray v. 11.1.0 - the strongest v. (from 02.09.19): 7 - 13 (details). |
Author: | q30 [ Sat May 25, 2024 3:51 am ] |
Post subject: | Re: Engine Tournament |
Quote: It looks like you're using very fast time limits. In slower games, the larger networks get more value from the extra time and will become relatively stronger. Compare the 1-minute and 5-minute rankings at https://lifein19x19.com/viewtopic.php?p=248817#p248817 and see how kata_20b overtakes kata_15b by a large margin given more time (and similar but less drastically for LZ188, 40 blocks versus LZ157, 15 blocks). Heavyweights are slower but more powerful :-) I'm testing with 2 minutes per move. You can find number of visits for different weights network files on this and this pages. Of course You are right. But the goal is to test with equal time and resources to each engine/weight. If You can test, how many time per move for this amount of visits will spend modern PC with 8-core CPU + modern video card GPU cores number, I will only welcome the release of these results data here. I think, that ~10 seconds per move is that a usual end user uses... |
Author: | q30 [ Sat May 25, 2024 3:59 am ] |
Post subject: | Re: Engine Tournament |
Quote: And with computers, not absolute time but time and hardware. We really are going to have to come to some agreement about those. What will be considered a "standard machine" and what "standard time controls. And what sort of time controls? Allowing X per move is simple but I believe we will eventually want AI that can be making "time management" decisions << situation not critical; try to conserve time vs critical, use some of the saved time >> Of course You are right too. What I think about "standard" for usual end user (that prefers the computer to spend time thinking evenly), I had wrote in message above... |
Page 20 of 20 | All times are UTC - 8 hours [ DST ] |
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group http://www.phpbb.com/ |