lookahead.ranges_data tensor size

[DWingHKL]

I was also confuse about Code comments said it has six dimensional tensor, but lookahead.ranges_data[1] just an five dimensional tensor .look like lose batch dimensional

line 130

--data structures [actions x parent_action x grandparent_id x batch x players x range] self.lookahead.ranges_data[1] = arguments.Tensor(1, 1, 1, constants.players_count, game_settings.card_count):fill(1.0 / game_settings.card_count)

[Kiv]

I also wondered about this.

I think the comment refers to a different version of the code. A batch dimension would allow to store data for different community (public) cards in that same tensor, since the different public cards don't affect the actions available in poker.

[lifrordi]

Yes, the comment is from different version of the code where we had another dimension for batch.