def step(self, state, player_id):
cardstr, one_last, two_last, three_last, legal_card = numpytostr(state)
curreny_hand = rltorh(cardstr)
one_last_hand = rltorh(one_last)
two_last_hand = rltorh(two_last)
three_last = rltorh(three_last)
upcard = []
if self.num == 0:
if len(two_last_hand) == 0:
upcard = three_last
else:
upcard = two_last_hand
else:
if len(one_last_hand) != 0:
upcard = one_last_hand
elif len(one_last_hand) == 0 and len(two_last_hand) != 0:
upcard = two_last_hand
putcard = to_char(
CEnv.step_auto_static(Card.char2color(curreny_hand),
to_value(upcard)))
putcard = rhtorl(putcard)
#print("player:", player_id, "手牌 is:", cardstr, "上一局:", upcard,"出牌:",putcard)
self.num = self.num + 1
return putcard
def step(self,state,player_id):
s = np.array(state['obs'])
cardstr, one_last, two_last, three_last, legal_card = numpytostr(state)
curreny_hand = rltorh(cardstr)
values = CEnv.get_cards_ass(Card.char2color(curreny_hand))
type,mymax,num = values[0],values[1],values[2]
actionzong = cardcouple(type,mymax,num)
for ca in actionzong:
print("组合牌面：：：：",ACTION_ID_TO_STR[ca])
hou_legal_action = []
if 308 in state["legal_actions"]:
hou_legal_action.append(308)
for a in actionzong:
if a in state["legal_actions"]:
hou_legal_action.append(a)
else:
hou_legal_action = actionzong
hou_legal_action = sorted(hou_legal_action)
#action = self.anet.choose_action(np.expand_dims(s, 0),hou_legal_action)
action = hou_legal_action[0]
print("---玩家:#", player_id, "#手牌是:", cardstr)
print("---出牌:", ACTION_ID_TO_STR[action], "id:", action)
return int(action)
def step(self, state, player_id):
s = np.array(state['obs'])
action = self.lnet.choose_action(np.expand_dims(s, 0),
state["legal_actions"], 0)[0]
cardstr, one_last, two_last, three_last, legal_card = numpytostr(state)
print("player:", player_id, "手牌 is:", cardstr)
#print("出牌:",ACTION_ID_TO_STR[action],"id":action)
return int(action)
def step(state, player_id):
cardstr, one_last, two_last, three_last, legal_card = numpytostr(state)
#print("player:", player_id, "手牌 is:", cardstr)
''' Predict the action given the curent state in gerenerating training data.
Args:
state (numpy.array): an numpy array that represents the current state
Returns:
action (int): the action predicted (randomly chosen) by the random agent
'''
return int(np.min(state['legal_actions']))
def step(self, state, player_id):
s = np.array(state['obs'])
zhuaction = self.lnet.choose_action(
np.expand_dims(self.normalizer.normalize(s), 0),
state["legal_actions"])
cardstr, one_last, two_last, three_last, legal_card = numpytostr(state)
print("player:", player_id, "手牌 is:", cardstr)
print("出牌:", ACTION_ID_TO_STR[zhuaction], "id:", zhuaction)
return int(zhuaction)
def step(state, player_id):
cardstr, one_last, two_last, three_last, legal_card = numpytostr(state)
#print("player:", player_id, "手牌 is:", cardstr)
''' Predict the action given the curent state in gerenerating training data.
Args:
state (numpy.array): an numpy array that represents the current state
Returns:
action (int): the action predicted (randomly chosen) by the random agent
'''
action = np.min(state['legal_actions'])
print("---玩家:#", player_id, "#手牌是:", cardstr)
print("---出牌:", ACTION_ID_TO_STR[action], "id:", action)
return int(action)