def _reward_fn(self, last_response, response, end_status):
"""Difference between previous gold and new gold."""
del end_status # Unused
old_gold = base._get(last_response, "Blstats.gold", 0)
gold = base._get(response, "Blstats.gold", old_gold)
time_penalty = self._get_time_penalty(last_response, response)
return gold - old_gold + time_penalty
def _reward_fn(self, last_response, response, end_status):
"""Difference between previous hunger and new hunger."""
del end_status # Unused
old_hunger = base._get(last_response, "You.uhunger", 0)
hunger = base._get(response, "You.uhunger", old_hunger)
reward = max(0, hunger - old_hunger)
time_penalty = self._get_time_penalty(last_response, response)
return reward + time_penalty
