def test_make(self):
register(env_id='test_make', entry_point='rlcard.envs.tarot:TarotEnv')
env = rlcard.make('test_make')
_, player = env.init_game()
self.assertEqual(player, 0)
with self.assertRaises(ValueError):
make('test_random_make')
def test_make(self):
register(env_id='test_make',
entry_point='rlcard.envs.blackjack:BlackjackEnv')
env = rlcard.make('test_make')
_, player = env.reset()
self.assertEqual(player, 0)
with self.assertRaises(ValueError):
make('test_random_make')
from yaniv_rl.models.yaniv_rule_models import YanivNoviceRuleAgent, YanivIntermediateRuleAgent
from yaniv_rl import utils
from rlcard.envs.registration import register, make
import sys
register(
env_id='yaniv',
entry_point='yaniv_rl.envs.yaniv:YanivEnv',
)
agents = [RandomAgent(488), YanivNoviceRuleAgent(), YanivIntermediateRuleAgent()]
# Make environment
eval_num = 10000
env = make('yaniv', config={'seed': 0, 'starting_player': 0})
table = [[0 for i in range(3)] for i in range(3)]
for i in range(3):
# player v player
env.set_agents([agents[i], agents[i]])
res = utils.tournament(env, eval_num)
print("{} vs {}: ".format(i, i), res)
winrate = res['wins'][0] / eval_num
table[i][i] = winrate
for agent_1, agent_2 in combinations(agents, 2):
a1i = agents.index(agent_1)
a2i = agents.index(agent_2)
env.set_agents([agent_1, agent_2])
res = utils.tournament(env, eval_num)
print("{} vs {}: ".format(a1i, a2i), res)
from yaniv_rl import utils
from rlcard.envs.registration import register, make
register(
env_id='yaniv',
entry_point='yaniv_rl.envs.yaniv:YanivEnv',
)
agents = [
RandomAgent(488),
YanivNoviceRuleAgent(),
YanivIntermediateRuleAgent()
]
# Make environment
env = make('yaniv', config={'seed': 0})
eval_num = 10000
table = [[0 for i in range(3)] for i in range(3)]
for i in range(3):
# player v player
env.set_agents([agents[i], agents[i]])
res = utils.tournament(env, eval_num)
winrate = res['wins'][0] / eval_num
table[i][i] = winrate
for agent_1, agent_2 in combinations(agents, 2):
a1i = agents.index(agent_1)
a2i = agents.index(agent_2)
env.set_agents([agent_1, agent_2])
def test_make_modes(self):
register(env_id='test_env',
entry_point='rlcard.envs.blackjack:BlackjackEnv')
with self.assertRaises(ValueError):
make('test_env', config={'active_player': -1})
from rlcard.envs.registration import register, make
import numpy as np
import time
start = time.time()
battle_number = 100000
Reward = 0
for _ in range(battle_number):
env = make('no-limit-holdem',config={'single_agent_mode':True})
#print('---------------new game------------------------')
state = env.reset()
#dealer_id = env.game.dealer_id
#print("dealer id即小盲注位置为:",env.game.dealer_id)
#print('state:',state)
while True:
action = np.random.choice(state['legal_actions'])
#print('选择的动作为:',action)
state,reward,done = env.step(action)
#print('state:',state)
#print('reweard:',reward)
#print('done:',done)
if done:
Reward += reward
break
end = time.time()
print('the average payoff of {} battles TP Vs random opponent:{}'.format(battle_number,Reward/battle_number))
print('time is：',end-start)