うーん…
新評価関数で学習させても結果が芳しくないので、旧評価関数でやった様に、終局まで指し切ったうえで序盤50手だけを学習ってのをやってみました。
で、事前に学習した進行度計算パラメータを使うとこうなります。
20局終了 +---+---+---+---+---+---+---+---+---+ |^圭|v飛|v王| |v銀| | |v歩| | +---+---+---+---+---+---+---+---+---+ | | |v金|v銀|v金| | |v角|v香| +---+---+---+---+---+---+---+---+---+ |v香|v歩|v歩|v歩|v歩|v歩|v歩| |v桂| +---+---+---+---+---+---+---+---+---+ | | | | | | | | | | +---+---+---+---+---+---+---+---+---+ |^香| | | | | |^歩| | | +---+---+---+---+---+---+---+---+---+ | | | | |^歩| | | |v歩| +---+---+---+---+---+---+---+---+---+ |vと|^歩|^歩|^歩| |^歩|^桂|v香| | +---+---+---+---+---+---+---+---+---+ | |^角|^金|^銀|^金| | | |vと| +---+---+---+---+---+---+---+---+---+ | | |^桂| |^王| |v龍| | | +---+---+---+---+---+---+---+---+---+ 歩,香,桂,銀,金,角,飛 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 1, 0, 0, 0, 手番:先手 手数:71 経過時間 :3m14sec
んでもって、対局と並行して進行度計算を学習させるとこうです。
610局終了 +---+---+---+---+---+---+---+---+---+ |v香|v桂|v銀| | | | | |v香| +---+---+---+---+---+---+---+---+---+ | |v王| | | |v飛|v銀| | | +---+---+---+---+---+---+---+---+---+ |v歩| | |^圭| | |v歩| |v角| +---+---+---+---+---+---+---+---+---+ | |^飛|^馬|v歩|v歩| |v金|v歩|v歩| +---+---+---+---+---+---+---+---+---+ |^銀| | | |^歩|v歩|^歩| | | +---+---+---+---+---+---+---+---+---+ |^歩| |^歩| | |^歩| |^歩|^歩| +---+---+---+---+---+---+---+---+---+ | | |vと|^歩|^金|^王|^金| | | +---+---+---+---+---+---+---+---+---+ | | | | |^銀| | |^金| | +---+---+---+---+---+---+---+---+---+ |^香| | | |^桂| | | |^香| +---+---+---+---+---+---+---+---+---+ 歩,香,桂,銀,金,角,飛 2, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 手番:後手 手数:116 経過時間 :50m59sec
…進行度計算の学習ルーチン、どっかバグってるのかな?