Search found 1551 matches

by yanquis1972
Tue Dec 11, 2018 1:56 am
Forum: Computer Chess Club: General Topics
Topic: Lco road to 8000 or 10000! self play elo.
Replies: 27
Views: 2023

Re: Lco road to 8000 or 10000! self play elo.

Thats an interesting idea, hope someone (SF team?) tests it thoroughly. Id be surprised if there isn’t a similar elo increase though.
by yanquis1972
Tue Dec 11, 2018 12:56 am
Forum: Computer Chess Club: General Topics
Topic: Scaling of Lc0 at high Leela Ratio
Replies: 21
Views: 1843

Re: Scaling of Lc0 at high Leela Ratio

I get dramatically better results going from eg 0+15s to 0+60s (should work out to very roughly 100kn searches/move vs 400kn on my hardware), this with test30 and prior to the 19.1 update. Since your hardware is considerably faster I’d be curious about the results vs SF8 with the deepmind opening se...
by yanquis1972
Sat Dec 08, 2018 4:42 am
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 243
Views: 13019

Re: Alphazero news

I am not convinced the newest SF would win against it. The ELO is calculated against a pool of similar engines. It's not clear if 50 or 100 ELO more against this pool is equal to 50-100 ELO more against an opponent of a different type. While that's true, Lc0 with the best nets on my powerful GPU an...
by yanquis1972
Fri Dec 07, 2018 4:06 am
Forum: Computer Chess Club: General Topics
Topic: Alphazero news
Replies: 243
Views: 13019

Re: Alphazero news

While I sympathize with that statement, releasing A0 source code and networks for anyone to test sounds better. Many will not be satisfied with in-house testing with supposedly fair conditions. That would be good, but it would also be a lot of work for us (AZ is tightly-coupled with DM and Google's...
by yanquis1972
Thu Dec 06, 2018 3:00 am
Forum: Computer Chess Club: General Topics
Topic: Remarkable SF - Lc0 Sicilian Kan
Replies: 4
Views: 349

Re: Remarkable SF - Lc0 Sicilian Kan

i think it looks absolutely devastating & when the top AB engine evaluates it as +1.20, it's rather astonishing to have it demonstrated (OTB at least) incorrect.
by yanquis1972
Thu Dec 06, 2018 1:59 am
Forum: Computer Chess Club: General Topics
Topic: Remarkable SF - Lc0 Sicilian Kan
Replies: 4
Views: 349

Re: Remarkable SF - Lc0 Sicilian Kan

Id is 31830 btw, and the 800+ nets have drawn all 11 games so far against SF10 (0+60s, deepmind opening set)
by yanquis1972
Thu Dec 06, 2018 1:50 am
Forum: Computer Chess Club: General Topics
Topic: Remarkable SF - Lc0 Sicilian Kan
Replies: 4
Views: 349

Remarkable SF - Lc0 Sicilian Kan

Watching this, I assumed it was another embarrassing tactical oversight , with Leela inexplicably missing Nxg7+. [Event "Lc0id318xx vs SF10"] [Site "PC"] [Date "2018.12.05"] [Round "12"] [White "Stockfish 10"] [Black "Lc0v3"] [Result "1/2-1/2"] [BlackElo "2875"] [ECO "B43"] [Opening "Sicilian"] [Tim...
by yanquis1972
Sun Dec 02, 2018 11:46 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish 10
Replies: 1
Views: 1048

Re: Stockfish 10

is default contempt (I believe 24) really best (not harmful against strong engines)? running a match against the lc0 test30 & it's extremely close so far; book is very shallow (using the deepmind set) if that's a factor. asking the question because in the current game SF10 started from the caro-kann...
by yanquis1972
Wed Nov 28, 2018 4:23 pm
Forum: Computer Chess Club: General Topics
Topic: Scaling of Lc0 at high Leela Ratio
Replies: 21
Views: 1843

Re: Scaling of Lc0 at high Leela Ratio

curious about test30 scaling; imo it should be much closer to A0, at least at present. using DM openings, 128MB hash, 3 cores/~7MN/s for SF8 & default settings for Lc0 w/ a GTX 1080 it scored 2W-4L-18D (-28 elo) @ 0+12s in the first batch of 24 games i ran. should be very roughly equivalent to 1s/m...
by yanquis1972
Tue Nov 27, 2018 4:19 pm
Forum: Computer Chess Club: General Topics
Topic: Scaling of Lc0 at high Leela Ratio
Replies: 21
Views: 1843

Re: Scaling of Lc0 at high Leela Ratio

curious about test30 scaling; imo it should be much closer to A0, at least at present. using DM openings, 128MB hash, 3 cores/~7MN/s for SF8 & default settings for Lc0 w/ a GTX 1080 it scored 2W-4L-18D (-28 elo) @ 0+12s in the first batch of 24 games i ran. should be very roughly equivalent to 1s/mo...