Stockfish with new NNUE architecture and bigger net released

[pohl4711]

https://abrok.eu/stockfish/

Author: Tomasz Sobczyk
Date: Tue May 18 18:06:23 2021 +0200
Timestamp: 1621353983

New NNUE architecture and net

Introduces a new NNUE network architecture and associated network parameters, as obtained by a new pytorch trainer. The network is already very strong at short TC, without regression at longer TC, and has potential for further improvements.


I started my 7000 games testrun... Result should go online next Friday.
Stay tuned!

[Spliffjiffer]

yes, had a quick test yesterday with Jörg Oster's HalfKAv2 (exp HalfKAv2 version) and it did score around 25 elo better than my cfish "pure" in a quick1min+0.2sec inc-testmatch...looking forward to a real test like yours

unfortunately i still cant run SF with incomplete TB's because it stops analsing then when TB's getting hit

[Modern Times]

Will they release it as SF14 I wonder.

[pohl4711]

Will they release it as SF14 I wonder.

Do not think so. The Elo gain is not that big... The "normal" LTC test in the SF-framework (with 60''+600ms thinkingtime) gave +5.85 Elo.

But, as written on abrok.eu, this new architecture and bigger netsize is a great base for further improvements. Thats the good news. Because, since the release of Stockfish 13, 3 months ago, the progress of SF was very small.

[peter]

Jörg Oster's HalfKAv2 (exp HalfKAv2 version)

Good to know that too, thanks regards

[Joerg Oster]

Jörg Oster's HalfKAv2 (exp HalfKAv2 version)

Good to know that too, thanks regards

Just to clarify, it was my compile, yet the net and code from @sopel (Tomasz Sobczyk).

[peter]

Jörg Oster's HalfKAv2 (exp HalfKAv2 version)

Good to know that too, thanks regards

Just to clarify, it was my compile, yet the net and code from @sopel (Tomasz Sobczyk).

Thought so anyhow, Jörg, thanks.
So we just have to wait for a MOJO- update to have Pure NNUE and Nullmove option again also and thanks btw for recent 1.5 version

[schack]

Is anybody seeing a difference in processor behavior with this new net? For some reason when I go from 16 to 32 threads the processor (5950x) now won't do much boosting at all. At 16 threads it's at 4.1ghz; at 32, it's at like 3.6. This was not the behavior with the version from just a couple of days ago.

[Cornfed]

Will they release it as SF14 I wonder.

Do not think so. The Elo gain is not that big... The "normal" LTC test in the SF-framework (with 60''+600ms thinkingtime) gave +5.85 Elo.

But, as written on abrok.eu, this new architecture and bigger netsize is a great base for further improvements. Thats the good news. Because, since the release of Stockfish 13, 3 months ago, the progress of SF was very small.

So...how is it a great "base for further improvements"?

[Nay Lin Tun]

Will they release it as SF14 I wonder.

Do not think so. The Elo gain is not that big... The "normal" LTC test in the SF-framework (with 60''+600ms thinkingtime) gave +5.85 Elo.

But, as written on abrok.eu, this new architecture and bigger netsize is a great base for further improvements. Thats the good news. Because, since the release of Stockfish 13, 3 months ago, the progress of SF was very small.

So...how is it a great "base for further improvements"?

Because they can tweek new network to gain another 5-10 elo..
They had been tweeking the same old network for 4-5 months and keep gaining small elo,? 5-10?.