Laskos wrote: ↑Tue Mar 26, 2019 4:44 pm
Haven't checked, but in the past I did check Leela performance with 6-men Syzygy on tricky endgames, they helped, but seemed to help no more than TBs help the regular engines, although strong regular engines play much better endgames even without TBs. It seems there are much more TB cut-offs with regular engines, and they actually use TBs better than Leela does. So, I wouldn't expect that using TBs something changes, and the games, though having more piece trading, are not going close to TBs very quickly, the importance of TBs being low even here.
Reminds me of this.
jp wrote: ↑Sat Mar 23, 2019 10:02 pm
The LC0 blog talks about a TCEC 7-man endgame, KNPP vs kbp, which Lc0 couldn't win.
SF showed +14 & +153.
TCEC gives the engines 6-man TBs.
Here I'm guessing TBs didn't help because Leela's lines weren't trading off a piece to get to the TBs.
Raphexon wrote: ↑Tue Mar 26, 2019 2:50 pm
I guess this confirms my suspicion that aggressively trading against NN engines (by an AB engine) is advantageous to the AB engine.
Basically you have the positional stage when there are still a lot of pieces on the board and the tactical stage when there are few.
And by shortening the positional stage through aggressive exchanges, the NN-engine is given less time to estabilish a positional advantage.
corres wrote: ↑Tue Mar 26, 2019 2:48 pm
In the old time developers think the opened positions give more opportunity to utilize the better tactical capacity of (AB type) engines.
Maybe reacting to the effect of NN engines the developers of AB engines should think such a manner than it was in the past.
But I think for a long run the competition will be won by the NN engines.
Yes, NN is still young stuff at top level, especially in Chess. In Go they, for example, have "ladder networks" helping the main Zero network, as Zero approach seems unable to cope well with ladders. Soon, such mixed approaches, to compensate for tactical and endgame weaknesses of Lc0 will probably be adopted in Chess too.
Raphexon wrote: ↑Tue Mar 26, 2019 2:50 pm
I guess this confirms my suspicion that aggressively trading against NN engines (by an AB engine) is advantageous to the AB engine.
Basically you have the positional stage when there are still a lot of pieces on the board and the tactical stage when there are few.
And by shortening the positional stage through aggressive exchanges, the NN-engine is given less time to estabilish a positional advantage.
Great work!
To me, the more interesting and revolutionary option is to set ThothFish to avoid trades. This is the real ground-breaker here. An engine that has a preference for closed positions and thriving on them. This is unlike anything we've seen before from an AB-type engine.
carldaman wrote: ↑Tue Mar 26, 2019 10:19 pm
To me, the more interesting and revolutionary option is to set ThothFish to avoid trades. This is the real ground-breaker here. An engine that has a preference for closed positions and thriving on them. This is unlike anything we've seen before from an AB-type engine.
But if you set that option, won't TF have a preference for closed positions but NOT thrive on them?
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
carldaman wrote: ↑Tue Mar 26, 2019 10:19 pm
To me, the more interesting and revolutionary option is to set ThothFish to avoid trades. This is the real ground-breaker here. An engine that has a preference for closed positions and thriving on them. This is unlike anything we've seen before from an AB-type engine.
But if you set that option, won't TF have a preference for closed positions but NOT thrive on them?
Against Leela it seems to NOT thrive at all, seemed weaker than the default SF against Leela, tried it for fun. And it will lose to default SF in direct match too. I think that was obvious to expect: seek open positions with less pieces on the board to beat Leela.
Laskos wrote: ↑Tue Mar 26, 2019 10:54 pm
I think that was obvious to expect: seek open positions with less pieces on the board to beat Leela.
I think even open positions with more pieces on the board (just trade lots of pawns) may be okay, but SF lost some very bad TCEC games with lots of pawns locked.
Laskos wrote: ↑Tue Mar 26, 2019 10:54 pm
I think that was obvious to expect: seek open positions with less pieces on the board to beat Leela.
I think even open positions with more pieces on the board (just trade lots of pawns) may be okay, but SF lost some very bad TCEC games with lots of pawns locked.
Tried only pawn trades too, seemed to not improve much against Leela. But always with some doubt, always more games are needed.
carldaman wrote: ↑Tue Mar 26, 2019 10:19 pm
To me, the more interesting and revolutionary option is to set ThothFish to avoid trades. This is the real ground-breaker here. An engine that has a preference for closed positions and thriving on them. This is unlike anything we've seen before from an AB-type engine.
But if you set that option, won't TF have a preference for closed positions but NOT thrive on them?
Maybe not against Leela, but it would certainly thrive vs other AB engines. SF is already (and has been for a long time) the best AB engine in closed positions, but it doesn't necessarily aim for them. Thothfish would take care of that, with the right settings.