I am curious to kniw if someone tried to make self learning chess engine like Alhpazero or Leila but without the tree search part?
I understand that you could probably set for example Leila to have search depth=1 but that wouldn't be the same thing because without a search tree you need a lot fewer evaluations of positions per game and you could have a much larger neural network.
I am curious to kniw if someone tried to make self learning chess engine like Alhpazero or Leila but without the tree search part?
I understand that you could probably set for example Leila to have search depth=1 but that wouldn't be the same thing because without a search tree you need a lot fewer evaluations of positions per game and you could have a much larger neural network.