Fat Fritz Flummoxed

Discussion of anything and everything relating to chess playing software and machines.

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
shrapnel
Posts: 1340
Joined: Fri Nov 02, 2012 8:43 am
Location: New Delhi, India

Re: Fat Fritz Flummoxed

Post by shrapnel » Sat Nov 16, 2019 4:24 pm

Werewolf wrote:
Sat Nov 16, 2019 3:08 pm
Different weights = different results I suppose.
But don't you think there is a possibility that there may be something wrong with the NN that your Fat Fritz shipped with, if its getting flummoxed so easily ?
Maybe you should ask Albert Silver or Chessbase for an urgent Update for the Fat Fritz NN....
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis

Werewolf
Posts: 1314
Joined: Thu Sep 18, 2008 8:24 pm

Re: Fat Fritz Flummoxed

Post by Werewolf » Sat Nov 16, 2019 5:18 pm

shrapnel wrote:
Sat Nov 16, 2019 4:24 pm
Werewolf wrote:
Sat Nov 16, 2019 3:08 pm
Different weights = different results I suppose.
But don't you think there is a possibility that there may be something wrong with the NN that your Fat Fritz shipped with, if its getting flummoxed so easily ?
Maybe you should ask Albert Silver or Chessbase for an urgent Update for the Fat Fritz NN....
I'll bet there are other positions FF can solve which Lc0 can't. Give it time and FF will probably sort itself out.

On the other hand has there been a direct head to head between the two to show elo difference? Presumably Lc0 has more resources behind it so the weights are more mature?

User avatar
Master Om
Posts: 400
Joined: Wed Nov 24, 2010 9:57 am
Location: INDIA

Re: Fat Fritz Flummoxed

Post by Master Om » Sat Nov 16, 2019 5:24 pm

Werewolf wrote:
Sat Nov 16, 2019 10:34 am
Fat Fritz on dual 2080 Ti (multiplexing, 4 threads, average nps = 55 Kn/s on this position) can't seem to find the right move after:

1.e4 Nf6
2.e5 Nd5
3.d4 d6
4.Nf3 de
5.Ne5 Nd7

and 6.Nf7 doesn't come even after 20 minutes thought...one of those quirks


[d]r1bqkb1r/pppnpppp/8/3nN3/3P4/8/PPP2PPP/RNBQKB1R w KQkq - 0 1[d]

Analysis by Fat Fritz (in lc0) rtx:

6.c4 N5f6 7.Nf3 e6 8.Nc3 c5 9.d5 Nb6 10.dxe6 Qxd1+ 11.Nxd1 Bxe6 12.b3 Be7 13.g3 a5 14.Bg2 a4 15.Rb1 Bf5 16.Rb2 axb3 17.axb3 Ra1 18.Bf4 0-0 19.0-0 Rfa8 20.Nh4 Bc8 21.Ne3 g6 22.Nf3 Rxf1+ 23.Kxf1 Ng4 24.Re2 Ra1+ 25.Re1 Rxe1+ 26.Kxe1 Nxe3 27.Bxe3 Nd7 28.Bf4 Nf8 29.Ng5 b6 30.Bd5 Ne6 31.Nxe6 fxe6 32.Bc6 Bd8 33.Ke2 Kf7 34.Kd3 Ke7 35.Ke4
White is slightly better: +/= (57% +39% =35% -26%) Depth: 25/72 00:12:31 34857kN
(Bicknell, 16.11.2019)
Even Lc0 on weight 49921 doesnot find.

Nodes: 456k, N/s: 2.0k, Time: 3m 45s

+0.57 c4 N5f6 Bf4 Nxe5 Bxe5 Ng4 Bg3 Nh6 d5 Nf5 Nc3 g6 Be5 f6 Bf4 e5 Bd2 Bc5 Bd3 O-O O-O a5 Kh1 b6 f4 Ne3 Bxe3 Bxe3 fxe5 fxe5 Rxf8+ (N: 80.83% [369k], P: 9.08%)
+0.50 Nf3 e6 c4 Bb4+ Bd2 Bxd2+ Qxd2 N5f6 Nc3 b6 Be2 Bb7 Rd1 O-O O-O c5 dxc5 Nxc5 Qe3 Qe7 Ne5 Rfd8 (N: 8.61% [39.3k], P: 16.77%)
+0.49 Nxd7 Bxd7 c4 Nf6 Nc3 e6 Be2 c5 Be3 cxd4 Qxd4 Be7 O-O O-O Rfd1 Bc6 Qf4 Qa5 a3 Rfd8 b4 Rxd1+ (N: 5.22% [23.8k], P: 11.08%)
+0.34 Bc4 Nxe5 dxe5 c6 Nc3 Be6 Qf3 g6 O-O Bg7 Rd1 Qc7 Bxd5 cxd5 Nxd5 Bxd5 Rxd5 Bxe5 g3 (N: 1.19% [5.4k], P: 7.06%)
+0.17 Qf3 Nxe5 dxe5 c6 Bc4 Qc7 Bxd5 Qxe5+ Be4 f5 Nd2 fxe4 Qxe4 Qxe4+ Nxe4 Bf5 f3 g6 Be3 Bg7 O-O-O (N: 0.44% [2.0k], P: 4.39%)
+0.24 Be2 Nxe5 dxe5 Bf5 c4 Nb6 Qb3 e6 O-O c6 Be3 Qc7 Nd2 Nd7 f4 a5 (N: 0.30% [1.4k], P: 2.46%)
+0.34 Nd3 g6 c4 N5b6 Qb3 Nf6 c5 Be6 Qc2 Nbd5 c6 b6 (N: 0.29% [1.3k], P: 1.65%)
+0.17 Bd3 Nxe5 dxe5 Nb4 O-O Nxd3 cxd3 Bf5 Qb3 b6 Nc3 e6 d4 Be7 d5 (N: 0.24% [1.1k], P: 2.36%)
+0.27 Ng4 g6 c4 N5f6 Nxf6+ Nxf6 Nc3 Bg7 Be2 O-O O-O c6 Be3 (N: 0.21% [944], P: 1.56%)
+0.22 Nxf7 Kxf7 Qh5+ Ke6 c4 N5f6 d5+ Kd6 c5+ Nxc5 Bf4+ Kd7 Bb5+ c6 dxc6+ bxc6 Qxc5 Qb6 Qxb6 axb6 Be2 Nd5 Be5 e6 O-O Ba6 (N: 0.20% [899], P: 1.73%)
+0.04 f4 e6 Be2 c5 c4 Ne7 d5 Nxe5 fxe5 exd5 O-O d4 Nd2 (N: 0.19% [847], P: 2.44%)
+0.22 a3 Nxe5 dxe5 g6 c4 Nb6 Qxd8+ Kxd8 Be3 Bg7 f4 f6 exf6 Bxf6 Nc3 Be6 O-O-O+ Ke8 (N: 0.18% [840], P: 1.61%)
+0.12 Nc4 g6 Be2 Bg7 O-O O-O Bf3 c6 Re1 b5 Ne3 e6 Nc3 (N: 0.16% [708], P: 1.76%)
+0.16 a4 Nxe5 dxe5 c6 Bc4 Bf5 O-O e6 Qe2 Qh4 Nc3 Bc5 Bd3 Nxc3 (N: 0.14% [624], P: 1.38%)
+0.01 Bb5 c6 Nxd7 Bxd7 Be2 Bf5 O-O e6 c4 Nb4 Na3 Be7 Be3 O-O Qb3 a5 Qc3 Bf6 (N: 0.14% [617], P: 1.9%)
+0.09 g3 Nxe5 dxe5 Nb4 Qxd8+ Kxd8 Na3 Bf5 c3 Nd3+ Bxd3 Bxd3 Bf4 Kc8 O-O-O Bf5 Rd2 (N: 0.13% [615], P: 1.6%)
-0.07 Nd2 Nxe5 dxe5 Bf5 c3 c6 Bc4 e6 O-O Nb6 Be2 Qc7 (N: 0.13% [610], P: 2.17%)
+0.11 Qd2 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 Qd7 O-O e6 Qe2 O-O-O Nd2 Qc6 (N: 0.13% [605], P: 1.52%)
+0.14 h3 Nxe5 dxe5 Nb4 Qxd8+ Kxd8 Na3 Bf5 c3 Nd3+ Bxd3 Bxd3 Be3 Ke8 O-O-O Be2 Rd2 (N: 0.13% [598], P: 1.42%)
+0.04 Qe2 Nxe5 Qxe5 Qd6 Qxd6 exd6 Bd2 Bf5 Na3 O-O-O O-O-O Nf6 f3 d5 c3 (N: 0.13% [573], P: 1.67%)
+0.15 h4 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 Qd7 Nc3 O-O-O Bd2 Nb4 (N: 0.12% [564], P: 1.31%)
+0.08 c3 Nxe5 dxe5 c6 Bc4 Bf5 Nd2 e6 O-O Nb6 Be2 Qc7 (N: 0.12% [540], P: 1.46%)
0.00 Bd2 Nxe5 dxe5 Bf5 Qf3 e6 Nc3 c6 O-O-O Nb4 Ne4 Qd5 Bxb4 (N: 0.10% [468], P: 1.47%)
-0.05 Na3 Nxe5 dxe5 c6 Be2 Bf5 O-O Qc7 Bg4 Bxg4 Qxg4 Qxe5 Nc4 (N: 0.10% [464], P: 1.59%)
-0.09 Qd3 g6 Nf3 Bg7 Be2 c5 O-O O-O Rd1 cxd4 Nxd4 (N: 0.10% [445], P: 1.62%)
-0.02 Bg5 Nxe5 dxe5 h6 Bd2 Bf5 Qf3 e6 Bd3 Bxd3 Qxd3 Qd7 Nc3 (N: 0.08% [346], P: 1.12%)
-0.16 b3 Nxe5 dxe5 Nb4 Qxd8+ Kxd8 Na3 Bf5 c3 Nd5 Nb5 c6 Nd4 e6 Nxf5 exf5 (N: 0.07% [339], P: 1.36%)
-0.42 Qh5 g6 Qf3 Nxe5 dxe5 Bg7 Bc4 c6 Nc3 Be6 O-O Bxe5 Re1 Bxc3 bxc3 (N: 0.07% [318], P: 1.73%)
-0.53 Nc3 Nxc3 bxc3 Nxe5 dxe5 Qxd1+ Kxd1 Bg4+ f3 O-O-O+ Bd3 Be6 Ke2 Rd5 f4 g5 (N: 0.05% [245], P: 1.48%)
-0.52 f3 Nxe5 dxe5 Bf5 Bd3 e6 Bxf5 exf5 Qd3 Qd7 Nc3 O-O-O Bg5 Nxc3 Bxd8 (N: 0.04% [163], P: 0.97%)
-0.60 Rg1 g6 c4 Nb4 Nxd7 Bxd7 a3 Nc6 d5 (N: 0.04% [162], P: 1.04%)
-0.78 g4 Nxe5 dxe5 Nb4 Qe2 Bxg4 Qxg4 Nxc2+ Ke2 Nxa1 Bg2 c6 e6 (N: 0.03% [134], P: 0.99%)
-1.69 Be3 Nxe3 fxe3 Nxe5 dxe5 Qxd1+ Kxd1 g6 Nd2 Bg7 Nf3 Bg4 Ke1 Bxf3 gxf3 Bxe5 Rb1 O-O-O (N: 0.02% [88], P: 1.09%)
-1.90 b4 Nxe5 dxe5 Nxb4 Qxd8+ Kxd8 Na3 Bf5 c3 Nd5 Nb5 c6 (N: 0.02% [76], P: 1.01%)
-8.25 Qg4 Nxe5 Qe4 Ng6 Be2 Nf6 Bb5+ c6 Bxc6+ bxc6 (N: 0.02% [74], P: 1.37%)
-1.67 Ke2 Nxe5 dxe5 Bf5 Ke1 Nb4 Na3 Qxd1+ Kxd1 O-O-O+ (N: 0.01% [61], P: 0.75%)
-1.79 Kd2 Nxe5 dxe5 Ne3+ Kxe3 Qxd1 Bb5+ Qd7 Bxd7+ Bxd7 Nc3 O-O-O Rd1 g6 (N: 0.01% [59], P: 0.75%)
-2.97 Ba6 Nxe5 dxe5 bxa6 Qf3 Rb8 O-O e6 Rd1 Qh4 Nc3 (N: 0.01% [48], P: 0.82%)
-18.04 Nc6 bxc6 c4 N5f6 Qa4 Bb7 c5 e5 Ba6 Bxa6 (N: 0.01% [40], P: 0.77%)
-7.85 Bf4 Nxf4 Nxf7 Kxf7 Qf3 g5 h4 e6 g3 Qf6 (N: 0.01% [34], P: 0.64%)
-22.52 Ng6 hxg6 Qf3 e6 Qg3 c6 Be2 c5 (N: 0.01% [33], P: 0.63%)
-2.34 Bh6 Nxe5 dxe5 gxh6 Qf3 c6 c4 Nb4 Qc3 (N: 0.01% [28], P: 0.44%)
Always Expect the Unexpected

User avatar
Master Om
Posts: 400
Joined: Wed Nov 24, 2010 9:57 am
Location: INDIA

Re: Fat Fritz Flummoxed

Post by Master Om » Sat Nov 16, 2019 5:27 pm

Werewolf wrote:
Sat Nov 16, 2019 10:34 am
Fat Fritz on dual 2080 Ti (multiplexing, 4 threads, average nps = 55 Kn/s on this position) can't seem to find the right move after:

1.e4 Nf6
2.e5 Nd5
3.d4 d6
4.Nf3 de
5.Ne5 Nd7

and 6.Nf7 doesn't come even after 20 minutes thought...one of those quirks


[d]r1bqkb1r/pppnpppp/8/3nN3/3P4/8/PPP2PPP/RNBQKB1R w KQkq - 0 1[d]

Analysis by Fat Fritz (in lc0) rtx:

6.c4 N5f6 7.Nf3 e6 8.Nc3 c5 9.d5 Nb6 10.dxe6 Qxd1+ 11.Nxd1 Bxe6 12.b3 Be7 13.g3 a5 14.Bg2 a4 15.Rb1 Bf5 16.Rb2 axb3 17.axb3 Ra1 18.Bf4 0-0 19.0-0 Rfa8 20.Nh4 Bc8 21.Ne3 g6 22.Nf3 Rxf1+ 23.Kxf1 Ng4 24.Re2 Ra1+ 25.Re1 Rxe1+ 26.Kxe1 Nxe3 27.Bxe3 Nd7 28.Bf4 Nf8 29.Ng5 b6 30.Bd5 Ne6 31.Nxe6 fxe6 32.Bc6 Bd8 33.Ke2 Kf7 34.Kd3 Ke7 35.Ke4
White is slightly better: +/= (57% +39% =35% -26%) Depth: 25/72 00:12:31 34857kN
(Bicknell, 16.11.2019)
But the T40B.4160 net does...


Nodes: 342k, N/s: 1.6k, Time: 3m 30s

+0.80 Nxf7 Kxf7 Qh5+ Ke6 g3 N7f6 Bh3+ Kd6 Qe5+ Kc6 Bg2 e6 c4 Kb6 Nc3 a5 cxd5 exd5 O-O c6 Bf4 h6 a3 g5 Bd2 Bd6 Qe2 Qe7 Qd1 Kc7 Rc1 Kb8 Na4 Ne4 Re1 Qf6 (N: 65.28% [224k], P: 1.81%)
+0.59 c4 Nxe5 cxd5 Nd7 Nc3 Nb6 Qf3 g6 a4 Bd7 a5 Nc8 Bf4 Bg7 Be5 O-O Bd3 Nd6 O-O c6 Rfe1 cxd5 Qxd5 Bc6 Qc5 Bxe5 dxe5 Ne8 Red1 (N: 23.05% [78.9k], P: 12.99%)
+0.54 Nf3 e6 c4 N5f6 g3 b6 Bg2 Bb7 O-O Bd6 Nc3 O-O a3 Re8 Re1 c5 dxc5 Bxc5 (N: 5.26% [18.0k], P: 17.71%)
+0.54 Nxd7 Bxd7 c4 Nf6 Be2 c5 dxc5 e6 Be3 Qc7 b4 a5 b5 Bxc5 Bxc5 Qxc5 O-O e5 Nc3 Be6 Na4 Qe7 (N: 2.88% [9.9k], P: 9.07%)
+0.34 Bc4 Nxe5 dxe5 c6 Nc3 Be6 Nxd5 Bxd5 Bxd5 Qxd5 Qxd5 cxd5 Be3 e6 Ke2 Rc8 c3 Bc5 Bd2 f6 (N: 0.78% [2.7k], P: 7.6%)
+0.06 Qf3 Nxe5 dxe5 c6 Bc4 Qc7 Bf4 Bf5 Bxd5 cxd5 Qxd5 e6 Qb5+ Qc6 (N: 0.34% [1.2k], P: 5.64%)
+0.33 Be2 Nxe5 dxe5 Bf5 c4 Nb6 Qb3 e6 O-O c6 Be3 Qc7 Nd2 Qxe5 Nf3 (N: 0.26% [874], P: 2.49%)
+0.35 Nd3 g6 c4 N5b6 Qb3 Nb8 Be3 Bg7 Nc3 Bxd4 (N: 0.21% [729], P: 1.97%)
+0.13 Bd3 Nxe5 dxe5 Nb4 O-O Nxd3 cxd3 Qd4 Nc3 c6 Re1 Bf5 Qb3 Qb6 Qa4 (N: 0.12% [426], P: 1.9%)
+0.12 Nc4 g6 Be2 Bg7 O-O O-O Bf3 N7b6 Na5 c6 Re1 Re8 c3 e5 dxe5 (N: 0.12% [412], P: 1.84%)
+0.25 Ng4 g6 c4 N5f6 Nxf6+ Nxf6 Nc3 Bg7 Be2 O-O O-O c6 Be3 Ne8 (N: 0.12% [407], P: 1.47%)
+0.20 a3 Nxe5 dxe5 g6 Bc4 Be6 Qe2 Bg7 O-O c6 Re1 (N: 0.10% [350], P: 1.4%)
+0.18 h3 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 Qd7 O-O O-O-O Nc3 Nb4 Qe4 e6 Bg5 (N: 0.10% [336], P: 1.38%)
-0.06 Nd2 Nxe5 dxe5 Bf5 c3 e6 Qb3 Nb6 Nf3 Qd5 Be3 O-O-O Be2 (N: 0.09% [320], P: 1.83%)
+0.01 f4 e6 Be2 Nxe5 fxe5 c5 O-O cxd4 Qxd4 Qc7 Kh1 Bc5 Qg4 Qxe5 (N: 0.09% [318], P: 1.66%)
+0.02 Bd2 Nxe5 dxe5 Bf5 c4 Nb6 Qf3 e6 Qxb7 Be7 Qc6+ Qd7 Qxd7+ Nxd7 (N: 0.09% [306], P: 1.58%)
0.00 Qe2 Nxe5 Qxe5 Qd6 Qxd6 exd6 c4 Nb4 Na3 Bf5 Bd2 d5 c5 (N: 0.09% [305], P: 1.61%)
+0.12 Qd2 g6 Bc4 Nxe5 dxe5 Be6 O-O Bg7 Re1 c6 (N: 0.09% [302], P: 1.36%)
-0.02 Bb5 c6 Nxd7 Bxd7 Be2 Bf5 O-O e6 c4 Nb4 Na3 Be7 Be3 O-O (N: 0.08% [288], P: 1.58%)
+0.15 a4 Nxe5 dxe5 c6 Bc4 Bf5 O-O e6 Bd3 Bxd3 Qxd3 Qc7 Qe4 (N: 0.08% [287], P: 1.24%)
+0.05 c3 Nxe5 dxe5 Bf5 Qb3 Nb6 Be3 e6 Nd2 Be7 O-O-O O-O (N: 0.08% [283], P: 1.4%)
+0.12 h4 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 Qd7 Nc3 Nxc3 Qxc3 O-O-O O-O (N: 0.08% [274], P: 1.23%)
+0.01 g3 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 Qd7 O-O O-O-O Nc3 Nb4 Qe4 (N: 0.07% [251], P: 1.3%)
-0.08 Qd3 g6 Nxd7 Bxd7 Qb3 e6 Nc3 Bc6 Bb5 (N: 0.07% [233], P: 1.37%)
-0.03 Bg5 Nxe5 dxe5 h6 Bd2 Bf5 Qf3 e6 Bd3 Bxd3 Qxd3 Qd7 (N: 0.06% [213], P: 1.18%)
-0.04 Na3 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 e6 O-O Qd7 Nc4 Be7 Qg3 O-O-O (N: 0.06% [204], P: 1.15%)
-0.49 Qh5 g6 Qf3 Nxe5 dxe5 Bg7 Bc4 c6 Nc3 Be6 O-O Nxc3 Bxe6 fxe6 (N: 0.05% [181], P: 1.56%)
-0.29 b3 Nxe5 dxe5 Nb4 Qxd8+ Kxd8 Na3 Bf5 c3 Nd5 Nb5 e6 Bd2 Bc5 (N: 0.05% [163], P: 1.18%)
-0.52 Nc3 Nxc3 bxc3 Nxe5 dxe5 Qxd1+ Kxd1 Bg4+ f3 O-O-O+ Bd3 Be6 Ke2 Rd5 f4 g5 (N: 0.05% [158], P: 1.39%)
-0.46 f3 e6 c4 Bb4+ Kf2 Nxe5 cxd5 Nd7 dxe6 (N: 0.03% [94], P: 0.79%)
-0.79 g4 Nxe5 dxe5 Nb4 Qe2 Qd5 Rg1 h5 Nc3 (N: 0.03% [86], P: 0.91%)
-0.72 Rg1 g6 c4 Nb4 Nxd7 Qxd7 a3 Nc6 (N: 0.02% [79], P: 0.8%)
-1.76 Be3 Nxe3 fxe3 Nxe5 dxe5 Qxd1+ Kxd1 g6 Nd2 Bg7 Nf3 Bg4 Ke1 Bxf3 gxf3 Bxe5 Rb1 O-O-O Bd3 (N: 0.02% [71], P: 1.2%)
-6.74 Qg4 Nxe5 Qe4 Nc6 Nc3 Nf6 Bb5 Nxe4 Bd2 (N: 0.01% [49], P: 1.18%)
-1.81 b4 Nxe5 dxe5 Nxb4 Qe2 Bf5 Qb5+ Nc6 Qxb7 Nd4 Bb5+ Nxb5 (N: 0.01% [47], P: 0.82%)
-1.70 Ke2 Nxe5 dxe5 Bf5 c4 Nb4 Qa4+ Nc6 Be3 Qd3+ (N: 0.01% [45], P: 0.75%)
-2.06 Kd2 Nxe5 dxe5 Nb4+ Ke1 Bf5 Na3 Qxd1+ Kxd1 (N: 0.01% [34], P: 0.65%)
-2.29 Ba6 Nxe5 dxe5 bxa6 Qf3 Rb8 O-O e6 Rd1 Qh4 (N: 0.01% [34], P: 0.69%)
-15.89 Nc6 bxc6 c4 N5f6 Qa4 Bb7 c5 e5 dxe5 Nd5 (N: 0.01% [29], P: 0.72%)
-19.02 Ng6 hxg6 c4 N5f6 Qb3 e5 Qe3 Ng4 Qe2 Bb4+ (N: 0.01% [23], P: 0.58%)
-7.09 Bf4 Nxf4 Qf3 Nxe5 dxe5 Ng6 e6 Bxe6 Qxb7 (N: 0.01% [22], P: 0.54%)
-2.44 Bh6 Nxe5 dxe5 gxh6 Qf3 c6 Bc4 Bg7 Qg3 Rg8 (N: 0.01% [21], P: 0.46%)
Always Expect the Unexpected

User avatar
Master Om
Posts: 400
Joined: Wed Nov 24, 2010 9:57 am
Location: INDIA

Re: Fat Fritz Flummoxed

Post by Master Om » Sat Nov 16, 2019 5:38 pm

Werewolf wrote:
Sat Nov 16, 2019 10:34 am
Fat Fritz on dual 2080 Ti (multiplexing, 4 threads, average nps = 55 Kn/s on this position) can't seem to find the right move after:

1.e4 Nf6
2.e5 Nd5
3.d4 d6
4.Nf3 de
5.Ne5 Nd7

and 6.Nf7 doesn't come even after 20 minutes thought...one of those quirks


[d]r1bqkb1r/pppnpppp/8/3nN3/3P4/8/PPP2PPP/RNBQKB1R w KQkq - 0 1[d]

Analysis by Fat Fritz (in lc0) rtx:

6.c4 N5f6 7.Nf3 e6 8.Nc3 c5 9.d5 Nb6 10.dxe6 Qxd1+ 11.Nxd1 Bxe6 12.b3 Be7 13.g3 a5 14.Bg2 a4 15.Rb1 Bf5 16.Rb2 axb3 17.axb3 Ra1 18.Bf4 0-0 19.0-0 Rfa8 20.Nh4 Bc8 21.Ne3 g6 22.Nf3 Rxf1+ 23.Kxf1 Ng4 24.Re2 Ra1+ 25.Re1 Rxe1+ 26.Kxe1 Nxe3 27.Bxe3 Nd7 28.Bf4 Nf8 29.Ng5 b6 30.Bd5 Ne6 31.Nxe6 fxe6 32.Bc6 Bd8 33.Ke2 Kf7 34.Kd3 Ke7 35.Ke4
White is slightly better: +/= (57% +39% =35% -26%) Depth: 25/72 00:12:31 34857kN
(Bicknell, 16.11.2019)

Lc0 on Sergio net is fast...

Nodes: 21.0k, N/s: 335, Time: 1m 2s

+0.83 Nxf7 Kxf7 Qh5+ Ke6 c4 N5f6 d5+ Kd6 Qf7 Ne5 Bf4 Qd7 c5+ Kxc5 Bxe5 Qxd5 Qxd5+ Nxd5 Nd2 Kc6 Rc1+ Kd7 Bc4 c6 O-O e6 (N: 33.58% [7.0k], P: 3.32%)
+0.60 Nf3 e6 c4 N5f6 Nc3 b6 Be2 Bb7 O-O Bd6 Nb5 O-O Nxd6 cxd6 Bf4 d5 (N: 29.35% [6.2k], P: 11.83%)
+0.56 c4 Nxe5 cxd5 Nd7 Nc3 g6 g3 Bg7 Bg2 O-O O-O Nb6 Re1 e6 dxe6 Bxe6 d5 Bf5 Qb3 Re8 (N: 18.00% [3.8k], P: 15.75%)
+0.57 Nxd7 Bxd7 c4 Nf6 Nc3 e6 Be2 c5 O-O cxd4 Qxd4 Bc6 Qh4 Be7 Rd1 Qa5 Qg3 (N: 7.52% [1.6k], P: 6.62%)
+0.52 Bc4 Nxe5 dxe5 c6 Nc3 Be6 Nxd5 Bxd5 Bxd5 Qxd5 Qxd5 cxd5 Be3 e6 Ke2 f6 exf6 gxf6 (N: 5.46% [1.1k], P: 8.38%)
+0.02 Qf3 Nxe5 dxe5 Nb4 Na3 Qd5 Qxd5 Nxd5 Bd2 Bf5 O-O-O e6 Nb5 Bc5 c4 (N: 0.86% [181], P: 7.51%)
+0.23 Be2 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 Qd7 O-O O-O-O Qf3 (N: 0.63% [133], P: 3.56%)
+0.12 f4 e6 c4 Bb4+ Kf2 Ne7 a3 Bd6 Nc3 (N: 0.47% [99], P: 3.39%)
+0.09 Bd3 Nxe5 dxe5 Nb4 Be4 Qxd1+ Kxd1 f5 exf6 exf6 Nc3 (N: 0.45% [94], P: 3.51%)
+0.37 Nd3 g6 c4 N5b6 c5 Nd5 Nc3 N7f6 Ne5 Bg7 Bc4 (N: 0.40% [83], P: 1.51%)
+0.24 a3 Nxe5 dxe5 Bf5 Bd3 Qd7 O-O O-O-O Qf3 Bxd3 (N: 0.26% [54], P: 1.46%)
-0.03 Nd2 Nxe5 dxe5 Bf5 c3 e6 Qb3 Nb6 Be2 a5 a4 (N: 0.22% [47], P: 2.16%)
+0.18 a4 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 e6 Qb5+ c6 Qxb7 (N: 0.21% [44], P: 1.4%)
+0.23 Ng4 h5 Ne3 N7f6 Be2 Nxe3 Bxe3 g6 h3 (N: 0.20% [43], P: 1.18%)
+0.17 h3 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 e6 O-O Qd7 Rd1 O-O-O (N: 0.20% [43], P: 1.38%)
+0.16 g3 Nxe5 dxe5 Nb4 Qxd8+ Kxd8 Na3 Bf5 c3 Nd3+ Bxd3 Bxd3 (N: 0.19% [40], P: 1.34%)
-0.17 Bb5 c6 Be2 Nxe5 dxe5 g6 c4 Nb4 a3 (N: 0.17% [36], P: 1.91%)
+0.10 c3 Nxe5 dxe5 c6 Bc4 Bf5 O-O e6 Nd2 Be7 (N: 0.17% [35], P: 1.27%)
+0.07 Qe2 Nxe5 dxe5 c6 g3 Bf5 Bg2 (N: 0.17% [35], P: 1.37%)
+0.08 h4 Nxe5 dxe5 Bf5 Bd3 Qd7 Bxf5 Qxf5 (N: 0.16% [34], P: 1.27%)
+0.04 Nc4 g6 Be2 Bg7 O-O O-O Re1 c5 Bf3 e6 (N: 0.15% [32], P: 1.31%)
-0.03 Bd2 Nxe5 dxe5 Bf5 Qf3 e6 Bd3 Bxd3 Qxd3 Qd7 (N: 0.13% [28], P: 1.26%)
0.00 Qd2 Nxe5 dxe5 Bf5 Bd3 e6 Bxf5 exf5 O-O (N: 0.12% [26], P: 1.14%)
+0.02 Na3 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 e6 O-O Qd7 (N: 0.12% [25], P: 1.08%)
-0.13 Qd3 Nxe5 dxe5 c6 c4 Nb4 Qxd8+ (N: 0.10% [20], P: 1.08%)
-0.42 Nc3 Nxc3 bxc3 Nxe5 dxe5 Qxd1+ Kxd1 Bd7 Bd3 e6 Ke2 (N: 0.10% [20], P: 1.38%)
-0.48 Qh5 g6 Qf3 Nxe5 dxe5 Bg7 Bc4 c6 Nc3 Be6 (N: 0.10% [20], P: 1.4%)
-0.18 Bg5 Nxe5 dxe5 h6 Bd2 Bf5 Qf3 e6 Bd3 Bxd3 (N: 0.09% [19], P: 1.05%)
-0.40 b3 Nxe5 dxe5 Bf5 Bd3 Bxd3 Qxd3 e6 O-O Nb4 (N: 0.08% [16], P: 1.04%)
-0.63 f3 e6 c4 Bb4+ Kf2 Ne7 Nxd7 (N: 0.05% [10], P: 0.84%)
-0.67 Rg1 g6 c4 Nb4 a3 Nxe5 axb4 (N: 0.05% [10], P: 0.87%)
-0.85 g4 Nxe5 dxe5 h5 gxh5 Bf5 Bd3 (N: 0.05% [10], P: 0.92%)
-0.49 Ba6 bxa6 Nc6 Bb7 Nxd8 (N: 0.04% [8], P: 0.61%)
-1.35 Be3 Nxe5 dxe5 Nxe3 Qxd8+ (N: 0.03% [6], P: 0.74%)
-1.82 Ke2 Nxe5 dxe5 Bf5 Ke1 e6 (N: 0.02% [5], P: 0.7%)
-1.83 b4 Nxe5 dxe5 Nxb4 Qxd8+ Kxd8 (N: 0.02% [5], P: 0.73%)
-4.42 Qg4 Nxe5 Qe4 Nc6 Bb5 e6 (N: 0.02% [5], P: 1.03%)
-2.49 Kd2 Nxe5 dxe5 Ne3+ Bd3 (N: 0.02% [4], P: 0.68%)
-5.92 Bf4 Nxf4 Nxf7 Kxf7 (N: 0.01% [3], P: 0.61%)
-4.13 Bh6 gxh6 Nxf7 (N: 0.01% [2], P: 0.5%)
-17.49 Nc6 bxc6 c4 (N: 0.01% [2], P: 0.44%)
-20.37 Ng6 hxg6 c4 (N: 0.01% [2], P: 0.47%)
Always Expect the Unexpected

shrapnel
Posts: 1340
Joined: Fri Nov 02, 2012 8:43 am
Location: New Delhi, India

Re: Fat Fritz Flummoxed

Post by shrapnel » Sat Nov 16, 2019 6:41 pm

Werewolf wrote:
Sat Nov 16, 2019 5:18 pm
shrapnel wrote:
Sat Nov 16, 2019 4:24 pm
Werewolf wrote:
Sat Nov 16, 2019 3:08 pm
Different weights = different results I suppose.
But don't you think there is a possibility that there may be something wrong with the NN that your Fat Fritz shipped with, if its getting flummoxed so easily ?
Maybe you should ask Albert Silver or Chessbase for an urgent Update for the Fat Fritz NN....
I'll bet there are other positions FF can solve which Lc0 can't. Give it time and FF will probably sort itself out.

On the other hand has there been a direct head to head between the two to show elo difference? Presumably Lc0 has more resources behind it so the weights are more mature?
True. lc0 certainly has its flaws, though I do think many of them get ironed out with
(1)very large NNs like Sergio's
(2) and top-end Hardware like ours
(3) AND the correct Tweaks for lc0 Parameters.
I very much doubt that most of the Default Settings decided upon by the lc0 Developers are really ideal for Top-Class Hardware. They were probably aimed at people with mid-level hardware.
This is where advice of guys like Ankan Banerjee is vital, but unfortunately he rarely seems to be around.
Direct Head-to-Head between lc0 and Fat Fritz ?
Well, I've seen around 3-4 guys using Fat Fritz hanging around in PlayChess, but they seemed strangely disinclined to play against me whenever I sent out Invitations for a 16"+1' game. :)
So, I've yet to play against Fat Fritz in the Engine Room.
But, win or lose, I promise to post a game here, when I finally get the chance to play against Fat Fritz.
BTW, what nick do you use on Playchess ?
I'd like to play a Friendly against you, if I see you there when I'm around.
Whoever wins, it hardly matters as it will be a learning experience and will be for the betterment of AI Chess.
Regards
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis

AndrewGrant
Posts: 813
Joined: Tue Apr 19, 2016 4:08 am
Location: U.S.A
Full name: Andrew Grant
Contact:

Re: Fat Fritz Flummoxed

Post by AndrewGrant » Sat Nov 16, 2019 10:49 pm

info depth 50 seldepth 128 multipv 1 score cp 98 time 24350096 nodes 972335631376 nps 39931000 tbhits 0 hashfull 1000 pv e5f7 e8f7 d1h5 f7e6 c2c4 d5f6 d4d5 e6d6 h5f7 d7c5 b2b4 c8f5 b4c5 d6d7 b1c3 d8e8 f7e8 a8e8 c1a3 e7e6 e1c1 h7h5 c1b2 f6e4 f1d3 e4c3 b2c3 f5d3 d1d3 e6d5 c4d5 e8e2 d3d2 e2d2 c3d2 h5h4 d2c3 c7c6 d5d6 h8h5 h1b1 b7b6 c5b6 a7b6 b1b6 f8d6 a3d6 d7d6 b6b4 h4h3 g2g3 h5c5 b4c4 c5a5 a2a4 c6c5 c3d2 d6d5 c4h4 d5e6 h4g4 e6f5 g4c4 f5e6 c4h4 a5a6 d2c3 e6d5 h4h5 d5d6 c3b3 a6b6 b3c2 b6a6 h5h4

Post Reply