guitarremote
commented
4 years ago Run the rough work part
On Wed, Mar 18, 2020 at 1:29 PM Vaibhav Ojha notifications@github.com wrote:
@DragonWarrior15 commented on this pull request.
In game_env.py https://github.com/DragonWarrior15/othello-rl/pull/4#discussion_r394161317 :
- s2_nd = self._converter.convert(s[1], input_format='bitboard_single', output_format='ndarray')
- legal_moves_nd = self._converter.convert(legal_moves, input_format='bitboard_single',\
- output_format='ndarray')
- a_nd = self._converter.convert(a, input_format='bitboard_single', output_format='ndarray')
- next_s1_nd = self._converter.convert(next_s[0], input_format='bitboard_single',\
- output_format='ndarray')
- next_s2_nd = self._converter.convert(next_s[1], input_format='bitboard_single',\
- output_format='ndarray')
- next_legal_moves_nd = self._converter.convert(next_legal_moves, input_format='bitboard_single',\
- output_format='ndarray')
- nd_list = [s1_nd, s2_nd, legal_moves_nd, a_nd, next_s1_nd, next_s2_nd, next_legal_moves_nd]
- transition_dict = {}
- transition_dict['normal_bottom'] = transition_list
- self._nd_trans_fn = {'normal_left': lambda x: np.rot90(x, 1),
Also add diagonal and anti diagonal filps here, after all the transitions, we also need to see which of this modifications are resulting in the same board, we might be able to avoid some operations this way
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DragonWarrior15/othello-rl/pull/4#discussion_r394161317, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFNNEKMZ53NKIYYOA6ZKFZTRIB5MJANCNFSM4LMQ6OAA .
Added board representation method in Game Env and corrected transition list