Search found 1229 matches

by chrisw
Mon Apr 12, 2021 8:16 pm
Forum: Computer Chess Club: General Topics
Topic: Can somebody compare the AMD Ryzen processors to the intel processors
Replies: 41
Views: 4857

Re: Can somebody compare the AMD Ryzen processors to the intel processors

Sure can.
Air pouring out the fan, AMD box, 35C. Whereas, i7 box, 27C. AMD has liquid cooling system with a big radiator.
by chrisw
Sat Apr 10, 2021 11:34 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: TD learning by self play (TD-Gammon)
Replies: 2
Views: 202

TD learning by self play (TD-Gammon)

I’m a bit stuck to see the conceptual difference between TD (as described in various TD-Gammon write ups) and RL as used in AZ, LC0 and NNUE-SF, unless: it’s the position target value derivation used to train on. NNUE-SF uses a lambda mix of game result and SF-eval at N-ply (as far as I can tell). L...
by chrisw
Tue Mar 16, 2021 9:34 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Reinforcement learning project
Replies: 48
Views: 8189

Re: Reinforcement learning project

This looks very much at what I was doing 4 years ago when I first started with Texel-tuning. It did the job, the only drawback is that it was slow, tuning about 800 weights from scratch (including PST's) with 4 million examples took almost a day on my old 10 core machine. When I tuned from scratch ...
by chrisw
Thu Mar 11, 2021 3:15 pm
Forum: Computer Chess Club: General Topics
Topic: NNUE Research Project
Replies: 61
Views: 4776

Re: NNUE Research Project

NNUE Research Project March 10, 2021 (...) http://rebel13.nl/home/nnue.html That's an interesting initiative, thank you ! Just to say few words about Orion 0.7 (the site states that "Orion 0.7 - From the information on the website the nnue origin of version 0.7 is unclear, the RMS implies a strong ...
by chrisw
Thu Mar 11, 2021 3:09 pm
Forum: Computer Chess Club: General Topics
Topic: NNUE Research Project
Replies: 61
Views: 4776

Re: NNUE Research Project

What is called "Minic" here is a net if I understand well. But is it "Napping Nexus" or "Nascent Nutrient" ? The first one is based on SF data, the second on Minic (the engine) ones. Ed can answer, because the names listed are the filenames he used to save the data. I think, not entirely sure, he t...
by chrisw
Thu Mar 11, 2021 11:58 am
Forum: Computer Chess Club: General Topics
Topic: NNUE Research Project
Replies: 61
Views: 4776

Re: NNUE Research Project

What is called "Minic" here is a net if I understand well. But is it "Napping Nexus" or "Nascent Nutrient" ? The first one is based on SF data, the second on Minic (the engine) ones. Ed can answer, because the names listed are the filenames he used to save the data. I think, not entirely sure, he t...
by chrisw
Thu Mar 11, 2021 10:06 am
Forum: Computer Chess Club: General Topics
Topic: NNUE Research Project
Replies: 61
Views: 4776

Re: NNUE Research Project

Interesting results. If you're just directly applying RMSE to the raw low depth evaluations, you're likely picking up mostly on the scale of the evaluation rather than the uniqueness of the evaluations (which could explain the high similarity of the FF2 net assuming ChessBase's claims about its ori...
by chrisw
Mon Mar 01, 2021 7:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: SEE (Static Exchange Evaluation) pruning nodes
Replies: 5
Views: 643

Re: SEE (Static Exchange Evaluation) pruning nodes

SEE pruning is most commonly used to prune bad captures (all moves with negative SEE) in the qsearch. In general, as with most search heuristics, SEE pruning has desirable behavior on average, but there are rare exceptional cases. If your search wasn't deep enough to resolve the tactic above before...
by chrisw
Mon Mar 01, 2021 1:03 pm
Forum: Computer Chess Club: General Topics
Topic: Are neural nets (the weights file) copyrightable?
Replies: 473
Views: 17560

Re: Are neural nets (the weights file) copyrightable?

Which side you prefer to argue can hardly be called relevant. That there were previous Stockfish versions that did not have a NNUE interpreter in them is not relevant: it has one now, and we are arguing about this version, not the previous one (which is not included as a copy). That it does more th...
by chrisw
Mon Mar 01, 2021 12:26 pm
Forum: Computer Chess Club: General Topics
Topic: How to make a double-sized net as good as SF NNUE in a few easy steps
Replies: 39
Views: 2319

Re: How to make a double-sized net as good as SF NNUE in a few easy steps

But you would also be able to do that when doubling the width of the second layer. For every new cell you add anywhere you can take random weights on the output connections and zero weights on the input connections. To make sure they initially do not alter the network output (for lack of input), an...