1226a2eee68336240a35fce9678320ca9430584f,snake/solver/dqn.py,DQNSolver,__choose_action,#DQNSolver#,185

Before Change

            )
            // Select an action with max value
            max_val = None
            for i, val in enumerate(q_eval_all[0]):
                direc = self.__actions[i]
                if direc != oppo_direc and (max_val is None or val > max_val):
                    max_val = val
                    action_idx = i

        return action_idx

    def __step(self, action_idx):
        direc = self.__actions[action_idx]

After Change

                    self.__state_eval: state
                }
            )
            action_idx = np.argmax(q_eval_all, axis=1)[0]
        return action_idx

    def __step(self, action_idx):
        direc = self.__snake_actions[action_idx]

Instances