Search found 286 matches

by jorose
Mon Aug 03, 2020 4:53 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: NNUE accessible explanation
Replies: 12
Views: 1889

Re: NNUE accessible explanation

1) If I want the NN to return an evaluation, which could be used in a traditional alphabeta search engine, how is this achieved most commonly? Do I have a last hidden layer that feeds into a single output neuron which uses e.g. a tanh activation function so -1 would be a loss and +1 would be a win?...
by jorose
Mon Aug 03, 2020 2:40 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: The White Room
Replies: 10
Views: 743

Re: The White Room

Oh wow! Thank you! I honestly only followed the first half of the tournament and mostly followed the gauntlet after that. After a very bad start I was just happy Winter was no longer last, but I guess with how close the field was Winter was able to make a fairly extreme comeback. Thanks for organizi...
by jorose
Fri Jul 31, 2020 12:00 am
Forum: Computer Chess Club: General Topics
Topic: Stockfish NNue [download ]
Replies: 19
Views: 1439

Re: Stockfish NNue [download ]

Thank you! I have an i7-9700k with 2080ti. Which version should I use? popcount, bmi2, or avx2? The GPU is irrelevant for SF. Use whichever on happens to run fastest on your system, it's not like you pay money per binary. Just run them each from a couple positions. Most likely BMI2 version will be ...
by jorose
Fri Jul 24, 2020 8:22 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: NNUE accessible explanation
Replies: 12
Views: 1889

Re: NNUE accessible explanation

I think in general a lot of why certain architectures are chosen over others boils down to "because we tried others and this worked the best". That's not to say there is no justification, just that it's not black and white. Perhaps, however, I can give you a bit of my intuition with different aspect...
by jorose
Thu Jul 23, 2020 2:12 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: NNUE accessible explanation
Replies: 12
Views: 1889

Re: NNUE accessible explanation

Hi Martin, cool to see you around here again! It is indeed a bit difficult to find non-Japanese sources for NNUE and it put me off a bit in the passed. You can find a nice compendium of sources here . Still a lot of them are in Japanese, but I still think its quite useful. Based on my understanding ...
by jorose
Wed Jul 01, 2020 1:31 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: knodes/s halved after improving eval
Replies: 6
Views: 425

Re: knodes/s halved after improving eval

Eval changes can be tested with fixed nodes per move games. Just wanted to chime in here to mention that Brian probably is referring to small changes and tweaks to the evaluation function. Larger changes can effect how fast the engine can search nodes, as you described finding out yourself in this ...
by jorose
Mon Jun 29, 2020 4:39 am
Forum: Computer Chess Club: General Topics
Topic: How good is your engine?
Replies: 32
Views: 1061

Re: How good is your engine?

I gave Winter the first position from this thread, not expecting it to find anything. Indeed it struggled and I gave up on depth 40, only to see that the bestmove UCI output was the right move! After starting the search again it was printing the correct line form the beginning :) info depth 28 selde...
by jorose
Thu Jun 18, 2020 3:30 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: Open Swiss Edition III
Replies: 62
Views: 4733

Re: Open Swiss Edition III

Hi Sergio,

I saw you are working on a new website! The design looks nice and I am looking forward to when it is closer to finished.

Naturally, I would like to enter Winter in the tournament! :D

Regards,
Jonathan
by jorose
Thu Jun 11, 2020 5:23 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: FGRL rating lists - Winter 0.8, Pirarucu 3.3.5, Marvin 3.6.0
Replies: 5
Views: 592

Re: FGRL rating lists - Winter 0.8, Pirarucu 3.3.5, Marvin 3.6.0

Nice writeup and I like your thought experiment with SF "do nothing". I agree that a large part of Winter's scaling probably comes from bad fast TC performance and low N/s relative to other classical engines. I do wonder how best to to measure scaling. Assuming equivalent definition of a node, one w...
by jorose
Thu Jun 11, 2020 7:36 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: FGRL rating lists - Winter 0.8, Pirarucu 3.3.5, Marvin 3.6.0
Replies: 5
Views: 592

Re: FGRL rating lists - Winter 0.8, Pirarucu 3.3.5, Marvin 3.6.0

Thank you for testing Winter at LTC!

It kind of fascinates me how Winter is between Texel and Igel in the rapid list which are 129 and 94 respective Elo above it in the bullet list.