mmt Vs. Ovyron (G4 D5 BG2)

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

Alayan
Posts: 550
Joined: Tue Nov 19, 2019 8:48 pm
Full name: Alayan Feh

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by Alayan »

Ethereal at depth 36 (10B nodes) with MultiPV=2 suggested as 1st move Bg5 (-1.00) and as 2nd best move Na4 (also -1.00, but the eval was worse for most of the search and Ethereal doesn't enforce eval consistency at a MultiPV depth).
User avatar
Ovyron
Posts: 4558
Joined: Tue Jul 03, 2007 4:30 am

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by Ovyron »

jp wrote: Mon Feb 03, 2020 12:21 pm This on its own is interesting. We wouldn't say this opening is very tactical, would we? That's the usual excuse for Lc0's failures. "Strategic" doesn't have meaning for computer play (it's all calculation), but if we pretend it does, I'm not convinced NN engines are so fantastic "strategically". The problem may be Lc0's weak endgame play. If lines aren't resolved by tactics, long searches will end up in endgames, which will harm Lc0's evaluations at the roots.
Lines are being resolved by tactics, you can tell because Stockfish will give some -2.00 score to some position Leela is very happy to play into, but if you look at her PV she's missing some key move, and after you insert this key move into her, her eval go under -1.60 very quick. If she could see those moves from the root her score would be much more accurate, but she can't see them even when the position is on her face!

1. g4 d5 2. c4 Bxg4 3. Bg2 c6 4. Qb3 e6 5. Qxb7 Nd7 6. Nc3 Ne7 7. cxd5 exd5 8. d4 Rb8 9. Qa6 Rb6 10. Qd3 Ng6 11. h3 Be6 12. Nf3 Bd6 13. h4 h5 14. b3 Nf6 15. Bg5 O-O

[d]3q1rk1/p4pp1/1rpbbnn1/3p2Bp/3P3P/1PNQ1N2/P3PPB1/R3K2R w KQ -
jp
Posts: 1480
Joined: Mon Apr 23, 2018 7:54 am

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by jp »

Ovyron wrote: Tue Feb 04, 2020 1:07 am Lines are being resolved by tactics, you can tell because Stockfish will give some -2.00 score to some position Leela is very happy to play into, but if you look at her PV she's missing some key move, and after you insert this key move into her, her eval go under -1.60 very quick.
But are these lines, by computer standards, "highly tactical"? It doesn't look like a position that's especially tactical.

What does it being "very happy to play into" mean? Evals of -1.0 or scores of 0.0?
Last edited by jp on Tue Feb 04, 2020 1:12 am, edited 1 time in total.
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by zullil »

Ovyron wrote: Tue Feb 04, 2020 1:07 am
1. g4 d5 2. c4 Bxg4 3. Bg2 c6 4. Qb3 e6 5. Qxb7 Nd7 6. Nc3 Ne7 7. cxd5 exd5 8. d4 Rb8 9. Qa6 Rb6 10. Qd3 Ng6 11. h3 Be6 12. Nf3 Bd6 13. h4 h5 14. b3 Nf6 15. Bg5 O-O

[d]3q1rk1/p4pp1/1rpbbnn1/3p2Bp/3P3P/1PNQ1N2/P3PPB1/R3K2R w KQ -
Komodo at depth 45 had

15...O-O (-1.18)
15...Rb8 (-0.98)
15...Qc8 (-0.87)
SheikhYerbouti

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by SheikhYerbouti »

An OxyMoron... by someone who isn't paying attention.
Ovyron wrote: Tue Feb 04, 2020 1:07 am
Lines are being resolved by tactics, you can tell because Stockfish will give some -2.00 score to some position Leela is very happy to play into, but if you look at her PV she's missing some key move, and after you insert this key move into her, her eval go under -1.60 very quick. If she could see those moves from the root her score would be much more accurate, but she can't see them even when the position is on her face!

1. g4 d5 2. c4 Bxg4 3. Bg2 c6 4. Qb3 e6 5. Qxb7 Nd7 6. Nc3 Ne7 7. cxd5 exd5 8. d4 Rb8 9. Qa6 Rb6 10. Qd3 Ng6 11. h3 Be6 12. Nf3 Bd6 13. h4 h5 14. b3 Nf6 15. Bg5 O-O

[d]3q1rk1/p4pp1/1rpbbnn1/3p2Bp/3P3P/1PNQ1N2/P3PPB1/R3K2R w KQ -
mmt
Posts: 343
Joined: Sun Aug 25, 2019 8:33 am
Full name: .

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by mmt »

I think it might be that LC0 hasn't seen many 1. g4 d5 2. Bg2 Bxg4 games in training and therefore is not as well prepared to play with the g2 pawn missing so early.
jp
Posts: 1480
Joined: Mon Apr 23, 2018 7:54 am

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by jp »

Well, that might weaken its opening play, but once it's around move 15 or 20, you'd hope it wouldn't be a problem. If it is, then we're seeing once again how opening-dependent NN engines are.
Last edited by jp on Tue Feb 04, 2020 2:38 am, edited 1 time in total.
SheikhYerbouti

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by SheikhYerbouti »

Exactly what a few have been saying in Leela chat. The solution is to put expansive opening book into her training. There are too many obscure openings including gambits that Leela chooses to ignore because of more preferential lines. What's more, high opening temperature doesn't resolve issue, instead makes her play too randomly.
mmt wrote: Tue Feb 04, 2020 2:28 am I think it might be that LC0 hasn't seen many 1. g4 d5 2. Bg2 Bxg4 games in training and therefore is not as well prepared to play with the g2 pawn missing so early.
mmt
Posts: 343
Joined: Sun Aug 25, 2019 8:33 am
Full name: .

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by mmt »

1. g4 d5 2. c4 Bxg4 3. Bg2 c6 4. Qb3 e6 5. Qxb7 Nd7 6. Nc3 Ne7 7. cxd5 exd5 8. d4 Rb8 9. Qa6 Rb6 10. Qd3 Ng6 11. h3 Be6 12. Nf3 Bd6 13. h4 h5 14. b3 Nf6 15. Bg5 O-O 16. e3

[d]3q1rk1/p4pp1/1rpbbnn1/3p2Bp/3P3P/1PNQPN2/P4PB1/R3K2R b KQ - 0 16
jp
Posts: 1480
Joined: Mon Apr 23, 2018 7:54 am

Re: mmt Vs. Ovyron (G4 D5 BG2)

Post by jp »

Wow, moves are coming fast.

The f8 R & d8 Q need to find their way to c6 and a6, resp., for obvious reasons. :wink: