TCEC used to have a section dedicated to NNUE engines in its rules:
Guidelines for use of NNUE at TCEC:
1. NNUE code can be used and considered as if it was a library (even if it is not literally one).
2. Custom modifications to the basic NNUE code are strongly encouraged, it should be considered rather like a starting point.
3. All NNUE training data should be generated by the unique engine's own search and/or eval code.
Do the NNUE guidelines apply outside NNUE technology?
No, of course not.
As of 22 October 2021, the TCEC NNUE guidelines have been depreciated and removed from the rules:
Madeleine Birchfield wrote: ↑Fri Oct 22, 2021 6:23 pm
TCEC used to have a section dedicated to NNUE engines in its rules:
Guidelines for use of NNUE at TCEC:
1. NNUE code can be used and considered as if it was a library (even if it is not literally one).
2. Custom modifications to the basic NNUE code are strongly encouraged, it should be considered rather like a starting point.
3. All NNUE training data should be generated by the unique engine's own search and/or eval code.
Do the NNUE guidelines apply outside NNUE technology?
No, of course not.
This means a starting programmer first must write a good HCE eval before getting into NNUE.
Ridiculous.
Edit, also, is SF still allowed to use Leela data under the new rule?
90% of coding is debugging, the other 10% is writing bugs.
Rebel wrote: ↑Fri Oct 22, 2021 10:13 pm
This means a starting programmer first must write a good HCE eval before getting into NNUE.
Ridiculous.
Edit, also, is SF still allowed to use Leela data under the new rule?
I think you misunderstand my post: those rules used to exist at TCEC but no longer exist anymore. SF is allowed to use Leela's data now because the rules preventing SF from using Leela's data at TCEC have been removed.
Rebel wrote: ↑Fri Oct 22, 2021 10:13 pm
This means a starting programmer first must write a good HCE eval before getting into NNUE.
Ridiculous.
Edit, also, is SF still allowed to use Leela data under the new rule?
I think you misunderstand my post: those rules used to exist at TCEC but no longer exist anymore. SF is allowed to use Leela's data now because the rules preventing SF from using Leela's data at TCEC have been removed.
I think you are right, my bad
90% of coding is debugging, the other 10% is writing bugs.
BTW, I remember that Peter Osterlund experimented with Giraffe's evaluation four years ago. I tested his Texel GI after he made it available, not long ago -- it was very slow, but it beat easily engines that achieve similar depths -- and it seems strange now that we had to wait for the NNUE and SF to embrace NNs for the evaluation.
