From Chessprogramming wiki
Jump to: navigation, search

Home * Engines * Kristallweizen

Kristallweizen and Hefeweizen (right) [1] [2]

a Shogi engine, or more precisely a Shogi evaluation function by team Barrel house, a beer bar in front of the Okayama station [3], due to the developers around Seiji Shiba, Mitsunori Matsushita and Yasuhiro Tajima [4]. Kristallweizen became runner-up at the WCSC29 in 2019, where it run in the Amazon Elastic Compute Cloud and applied multi pondering. Kristallweizen is a NNUE HalfKP-256x2-32-32 type of evaluation function, used within YaneuraOu's search, a Shogi adaptation of Stockfish's search. Further CSA available engines and libraries used to train and prepare Kristallweizen mentioned on the WCSC29 site [5] were Apery, Tanuki, Qhapaq, Elmo, dlshogi [6], python-shogi [7] and the Android Go Player No. 18 [8]. Kristallweizen may also used with Dolphin, also based on YaneuraOu's (version 4.82) search [9][10].


Kristallweizen's forerunner, the more unfiltered Hefeweizen uses an Apery type of KPP/KKP indexed Piece-Square Tables approach, initially introduced by Bonanza. Hefeweizen won the WCSC28 in 2018, also on the Amazon Elastic Compute Cloud with multi ponder [11].

Forum Posts

External Links