Breaking down Texel's tuning method - automated evaluation function tuning in chess engines

Hey what's up guys, Code Monkey King's here. Texel's tuning is a very sensitive topic for me - I got stuck with it for months but now finally realized how it works thankfully to outstanding explanations by Ronald Friederich, the author of chess engines RofChade and PeSTO (both 3000+ Elo) Texel's tuning pipeline explained: https://github.com/maksimKorzh/wukongJS/blob/main/docs/TEXEL'S_TUNING.MD Contents: 03:33 - intro 05:07 - describing evaluation in Wukong JS 10:36 - issue definition 14:57 - when automated tuning comes onto the stage 18:30 - input & output data examples explanation 21:58 - some notes on my Texel's tuning implementation 31:22 - generating dataset 48:14 - calculating evaluation error (mean square error) 59:54 - adjusting evaluation parameters 01:10:45 - SUMMARY: HOW TEXEL'S TUNING WORKS 01:19:45 - BONUS: ideas for Wukong JS tuning WukongJS - javascript chess engine with UCI support: https://github.com/maksimKorzh/wukongJS Evaluation tuner: https://github.com/maksimKorzh/wukongJS/tree/main/tools/eval_tuner Datasets: http://talkchess.com/forum3/viewtopic.php?f=7&t=76251