textworld/envs/tw.py [68:88]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
                self.state["policy_commands"] = self._inform7.gen_commands_from_actions(self._current_winning_policy)

        if self.infos.intermediate_reward:
            self.state["intermediate_reward"] = 0
            if self.state["won"]:
                # The last action led to winning the game.
                self.state["intermediate_reward"] = 1

            elif self.state["lost"]:
                # The last action led to losing the game.
                self.state["intermediate_reward"] = -1

            elif self._previous_winning_policy is None:
                self.state["intermediate_reward"] = 0

            else:
                diff = len(self._previous_winning_policy) - len(self._current_winning_policy)
                self.state["intermediate_reward"] = int(diff > 0) - int(diff < 0)  # Sign function.

        if self.infos.facts:
            self.state["facts"] = list(map(self._inform7.get_human_readable_fact, self.state["_facts"]))
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



textworld/envs/wrappers/tw_inform7.py [233:253]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
                self.state["policy_commands"] = self._inform7.gen_commands_from_actions(self._current_winning_policy)

        if self.infos.intermediate_reward:
            self.state["intermediate_reward"] = 0
            if self.state["won"]:
                # The last action led to winning the game.
                self.state["intermediate_reward"] = 1

            elif self.state["lost"]:
                # The last action led to losing the game.
                self.state["intermediate_reward"] = -1

            elif self._previous_winning_policy is None:
                self.state["intermediate_reward"] = 0

            else:
                diff = len(self._previous_winning_policy) - len(self._current_winning_policy)
                self.state["intermediate_reward"] = int(diff > 0) - int(diff < 0)  # Sign function.

        if self.infos.facts:
            self.state["facts"] = list(map(self._inform7.get_human_readable_fact, self.state["_facts"]))
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -