Changes

He introduced new net architectures using the [[Stockfish NNUE#HalfKA|HalfKAv2HalfKA]] feature set , HalfKAv2 <ref>[https://github.com/official-stockfish/Stockfish/pull/3474 Update default net to nn-8a08400ed089.nnue by Sopel97 · Pull Request #3474 · official-stockfish/Stockfish · GitHub]</ref> <ref>[https://github.com/glinscott/nnue-pytorch/blob/master/docs/nnue.md#halfkav2-feature-set nnue-pytorch/nnue.md at master · glinscott/nnue-pytorch · GitHub - HalfKAv2 feature set]</ref>and HalfKAv2_hm <ref>[https://github.com/official-stockfish/Stockfish/pull/3646 New NNUE architecture and net by Sopel97 · Pull Request #3646 · official-stockfish/Stockfish · GitHub]</ref> <ref>[https://github.com/glinscott/nnue-pytorch/blob/master/docs/nnue.md#halfkav2_hm-feature-set nnue-pytorch/nnue.md at master · glinscott/nnue-pytorch · GitHub - HalfKAv2 feature set]</ref> feature sets, optimized the NNUE inference code for various [[SIMD and SWAR Techniques#SIMD Instruction Sets|SIMD instruction sets]] <ref>[https://github.com/official-stockfish/Stockfish/pull/3203 Optimization of the affine transformations. by Sopel97 · Pull Request #3203 · official-stockfish/Stockfish · GitHub]</ref> <ref>[https://github.com/official-stockfish/Stockfish/pull/3218 AVX-512 optimizations. by Sopel97 · Pull Request #3218 · official-stockfish/Stockfish · GitHub]</ref> <ref>[https://github.com/official-stockfish/Stockfish/pull/3663 Optimize and tidy up affine transform code. by Sopel97 · Pull Request #3663 · official-stockfish/Stockfish · GitHub]</ref>, and contributed to [[Gary Linscott|Gary Linscott's]] [[Gary Linscott#PyTorch NNUE|Pytorch NNUE]] project to train Stockfish , mostly working on optimizations which allowed nets to be trained within hours instead of days <ref>[https://github.com/glinscott/nnue-pytorch/pull/1 Defer data preparation to native code. Use sparse input tensors. · Pull Request #1 · glinscott/nnue-pytorch · GitHub]</ref> <ref>[https://github.com/glinscott/nnue-pytorch/pull/96 A custom kernel for the feature transformer. · Pull Request #96 · glinscott/nnue-pytorch · GitHub]</ref><ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=77503&start=55 Re: will Tcec allow Stockfish with a Leela net to play?] by [[Connor McMonigle]], [[CCC]], June 17, 2021</ref>. He also heavily contributed to the original NNUE trainer and data generator by [[Yu Nasu]], cleaning up the codebase, updating the data generator, optimizing the trainer, and adding other useful tools <ref>[https://github.com/nodchip/Stockfish/pulls?q=is%3Apr+author%3ASopel97 Pull Requests by Sopel97 · nodchip/Stockfish · GitHub]</ref> which are now available in the tools branch of the official-stockfish repository <ref>[https://github.com/official-stockfish/Stockfish/tree/tools official-stockfish tools branch · official-stockfish/Stockfish · GitHub]</ref>. He also introduced the [[Binpack]] storage format for training data <ref>[https://github.com/Sopel97/nnue_data_compress Sopel97/nnue_data_compress · GitHub]</ref> <ref>[https://github.com/nodchip/Stockfish/pull/129 Introduce sfen_format option for gensfen. Experimental support for binpack format in gensfen and learn. · Pull Request #129 · nodchip/Stockfish · GitHub]</ref>, comprising of position evaluations from chess games, which utilizes efficient delta encoding, and reduces the sizes of the datasets by 10 to 20 times compared to the previous solutions. His contributions span more than 200 pull requests over multiple repositories <ref>[https://github.com/nodchip/Stockfish/pulls?q=is%3Apr+author%3ASopel97 Pull Requests by Sopel97 · nodchip/Stockfish · GitHub]</ref> <ref>[https://github.com/official-stockfish/Stockfish/pulls?q=is%3Apr+author%3ASopel97 Pull Requests by Sopel97 · official-stockfish/Stockfish · GitHub]</ref> <ref>[https://github.com/glinscott/nnue-pytorch/pulls?q=is%3Apr+author%3ASopel97 Pull Requests by Sopel97 · glinscott/nnue-pytorch · GitHub]</ref>. He's also an author of a chess engine [[Fat Titz]], which is based on [[CFish]] and is a parody of [[Fat Fritz]] <ref>[https://github.com/Sopel97/FatTitz Sopel97/FatTitz · GitHub]</ref> <ref>[http://www.talkchess.com/forum3/viewtopic.php?f=2&t=78026 Fat Titz 1.0 released by [[Tomasz Sobczyk|Sopel]], [[CCC]], Aug 26, 2021]</ref>.

* [https://github.com/glinscott/nnue-pytorch/issuespull/73 get rid of some nodchip 1 Defer data preparation to native code . Use sparse input tensors. by Sopel97 · Issue Pull Request #73 1 · glinscott/nnue-pytorch · GitHub]* [https://github.com/glinscott/nnue-pytorch/pull/96 A custom kernel for the feature transformer. by Sopel97 · Pull Request #96 · glinscott/nnue-pytorch · GitHub]