Search found 71 matches

by thomasahle
Sun Feb 10, 2019 9:35 pm
Forum: Computer Chess Club: General Topics
Topic: What happened for Stockfish at Move 47...?
Replies: 3
Views: 800

Re: What happened for Stockfish at Move 47...?

Ah, that's interesting. In that case I would have thought it returned whatever was in the transposition table.
But maybe it doesn't even use the table in this case.
by thomasahle
Sun Feb 10, 2019 10:00 am
Forum: Computer Chess Club: General Topics
Topic: What happened for Stockfish at Move 47...?
Replies: 3
Views: 800

What happened for Stockfish at Move 47...?

In the latest TCEC game that Lc0 won, near the end, Stockfish showed a 2600cp advantage for white, but then for one move it dropped down to 176cp before going back up to an impressive 7913. Game here: http://legacy-tcec.chessdom.com/archive.php?se=14&sf&ga=29 Presumably this was a bug that got revea...
by thomasahle
Thu Jan 11, 2018 8:30 pm
Forum: Computer Chess Club: General Topics
Topic: Sunfish (Python Engine)
Replies: 3
Views: 1390

Re: Sunfish (Python Engine)

Hi Norbert,
I didn't write the UCI component myself, but I think it used to work. Maybe later changes to sunfish broke it.
If you make any fixed, I'll be happy to include them!

Best, Thomas
by thomasahle
Fri Dec 08, 2017 10:40 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Residual Networks for Computer Go
Replies: 1
Views: 974

Re: Residual Networks for Computer Go

This seems to be pre Alpha Go Zero. Since the Zero paper also used residual nets, and this paper doesn't seem to mention it.
by thomasahle
Mon Aug 28, 2017 1:12 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Approximating Stockfish's Evaluation by PSQTs
Replies: 9
Views: 4639

Re: Approximating Stockfish's Evaluation by PSQTs

Making different maps for black and white would be interesting. I chose this format because it made it easier to plug the tables into sunfish. My experiments also showed that sunfish played slightly stronger with the asymmetric tables. Do you have a link to the heat maps you ate referencing? http:/...
by thomasahle
Sun Aug 27, 2017 1:51 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Approximating Stockfish's Evaluation by PSQTs
Replies: 9
Views: 4639

Re: Approximating Stockfish's Evaluation by PSQTs

I see, but these are just based on where the piece most often goes, right? Not an indication if whether it goes there by its own will, or because it's forced.
by thomasahle
Sun Aug 27, 2017 9:49 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Approximating Stockfish's Evaluation by PSQTs
Replies: 9
Views: 4639

Re: Approximating Stockfish's Evaluation by PSQTs

Making different maps for black and white would be interesting. I chose this format because it made it easier to plug the tables into sunfish. My experiments also showed that sunfish played slightly stronger with the asymmetric tables.

Do you have a link to the heat maps you ate referencing?
by thomasahle
Wed Aug 23, 2017 1:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Approximating Stockfish's Evaluation by PSQTs
Replies: 9
Views: 4639

Approximating Stockfish's Evaluation by PSQTs

I did some experiments, mostly for fun, that you all may find fun as well. I made stockfish play 20,000,000 random games to ply 60 at 1ms per move. Each game was made different by always picking a random move from its top 6 pvs. For each (white) position I read Stockfish's score estimate and recorde...
by thomasahle
Mon Aug 21, 2017 10:15 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: high values of pst
Replies: 16
Views: 4953

Re: high values of pst

This is why I require the PST tables to sum to zero. Always nice to be able to remove some redundant dimensions when optimizing.
by thomasahle
Tue Feb 21, 2017 9:26 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Deep Pink: a chess engine using deep learning
Replies: 20
Views: 6698

Re: Deep Pink: a chess engine using deep learning

"Players will choose an optimal or near-optimal move" is a bad assumption for the FICS dataset. Players will choose better than random moves, which is part of the theory. I think this should depend on how robust your model and training is to noise. Say 10% of moves are blunders (more than 50cp off ...