Forums

Leela gambit in Nimzo Indian defense! Rook sacrifice in move 7!

Sort:
drmrboss
drmrboss
Jun 11, 2020
0
#1
In Nimzo indian defense Novelty, 7. e5N! was seen in computer chess since 2 years ago where a newly trained leela beat Gull. 
People have tested this position several times. and last TCEC season 17, both Stockfish and Leela ended as draw in both colors. 
However, today, Allie beat Stockfish with this rook gambit.  In general, you cant bluff a computer, but if your sacrifice is too deep beyond the search horizon of the depth of engines, it works. happy.png 
bong711
bong711
Closed: Inactive
Jun 11, 2020
0
#2

Thanks for sharing the novelty. Someday I will try it in blitz.

barretoff
barretoff
Jun 20, 2020
0
#3

Hello master, how are you?

In this case, the TCEC super final started and Leela is taking a beating from Stockfish, in less than 10 games she already has 3 losses. But what I don't understand is this: why is TCEC using a Leela network that is not the strongest?

According to the website https://www.sp-cc.de/nn-vs-sf-testing.htm, the strongest version of Leela is currently the t60-4082 (30x384). However, they are using version t60-3972 (30x384) at the moment, on the superfinal, which on this website I mentioned is in the 11th position of the raking and has 32 elo points less than the version that won the last TCEC, the t60-3010 (30x384) .

The network t60-4082 (30x384) has a rating of 3589, while the network used on the superfinal (t60-3972 (30x384)) has 3550. What is the logic behind this? I did not understand

barretoff
barretoff
Jun 20, 2020
0
#4

1  Lc0 0.24.1 LS 14.3 (20x256)     : 3644 513 (+311,=  0,-202), 60.6 %
2  Lc0 0.25.1 LS 15 (20x256)       : 3643 512 (+310,=  0,-202), 60.5 %
3  Lc0 0.24.1 LS 14.2 (20x256)     : 3633 520 (+308,=  0,-212), 59.2 %
4  Lc0 0.25.1 t60-4082 (30x384)    : 3589 510 (+270,=  0,-240), 52.9 %
5  Lc0 0.25.1 t40-1541 (20x256)    : 3583 516 (+269,=  0,-247), 52.1 %
6  Lc0 0.25.1 t60-3010 (30x384)    : 3582 514 (+267,=  0,-247), 51.9 %
** Stockfish 200418 ************** : 3568 SPCC-Elo *******************
7  Allie 0.6 LS 14.3 (20x256)      : 3558 519 (+252,=  0,-267), 48.6 %
8  Lc0 0.25.1 42850 (20x256)       : 3556 522 (+252,=  0,-270), 48.3 %
9  Lc0 0.25.1 63651 (24x320)       : 3554 517 (+248,=  0,-269), 48.0 %
10 Lc0 0.25.1 702820 (10x128)      : 3552 518 (+247,=  0,-271), 47.7 %
11 Lc0 0.25.1 t60-3972 (30x384)    : 3550 514 (+244,=  0,-270), 47.5 %
12 Fat Fritz 1.1 (20x256)          : 3530 523 (+233,=  0,-290), 44.6 %
13 Lc0 0.25.1 63305 (24x320)       : 3530 512 (+228,=  0,-284), 44.5 %
14 Lc0 0.25.1 32930 (20x256)       : 3483 515 (+196,=  0,-319), 38.1 %
15 Lc0 0.25.1 714435 (19x256)      : 3463 517 (+183,=  0,-334), 35.4 %
16 Lc0 0.25.1 11260 (20x256)       : 3408 521 (+149,=  0,-372), 28.6 %

 

https://www.sp-cc.de/nn-vs-sf-testing.htm