def run_one_game_rebuys(deepbot, nb_hands = 500, ini_stack = 3000, sb_amount = 50, opponents = 'default', verbose=False, cst_decks=None):
if opponents == 'default':
opp_algos = [CallBot(), ConservativeBot(), EquityBot(), ManiacBot()]
opp_names = ['call_bot','conservative_bot', 'equity_bot','maniac_bot']
else:
opp_algos = opponents['opp_algos']
opp_names = opponents['opp_names']
earnings = OrderedDict()
## for each bot to oppose
for opp_algo, opp_name in zip(opp_algos, opp_names):
#first match
max_round = nb_hands
my_game_result_1 = 0
cst_deck_match=cst_decks.copy()
deepbot.model.reset()
while True:
config = setup_config(max_round=max_round, initial_stack=ini_stack, small_blind_amount=sb_amount)
config.register_player(name=deepbot.opponent, algorithm=opp_algo)
config.register_player(name="deepbot", algorithm= deepbot)
game_result_1 = start_poker(config, verbose=0, cheat = True, cst_deck_ids = cst_deck_match, return_last_two=False)
#Hacky fix for missing last SB in some results
if game_result_1['players'][1]['stack'] == 2*ini_stack-sb_amount:
game_result_1['players'][1]['stack'] = 2*ini_stack
my_game_result_1 += game_result_1['players'][1]['stack'] - ini_stack
max_round-=(deepbot.round_count+1)
if max_round<=0:
break
if verbose:
print("Stack after first game: "+ str(game_result_1))
#return match
max_round = nb_hands
my_game_result_2 = 0
cst_deck_match=cst_decks.copy()
deepbot.model.reset()
while True:
config = setup_config(max_round=max_round, initial_stack=ini_stack, small_blind_amount=sb_amount)
config.register_player(name="deepbot", algorithm=deepbot)
config.register_player(name=deepbot.opponent, algorithm=opp_algo)
game_result_2 = start_poker(config, verbose=0, cheat = True, cst_deck_ids = cst_deck_match, return_last_two=False)
#Hacky fix for missing last SB in some results
if game_result_2['players'][0]['stack'] == 2*ini_stack-sb_amount:
game_result_2['players'][0]['stack'] = 2*ini_stack
my_game_result_2 += game_result_2['players'][0]['stack'] - ini_stack
max_round-=(deepbot.round_count+1)
if max_round<=0:
break
if verbose:
print("return game: "+ str(game_result_2['players'][0]['stack']))
earnings[opp_name] = my_game_result_1 + my_game_result_2
# print('Done with game of bot number: '+ str(deepbot.id))
return earnings
def run_one_game_reg(simul_id , gen_id, deepbot, verbose=False, cst_decks=None, nb_sub_matches =10):
"""old"""
#CONFIGURATION
nb_hands = 500
ini_stack = 2000
sb_amount = 50
opp_algos = [CallBot(), ConservativeBot(), EquityBot(), ManiacBot()]
opp_names = ['call_bot','conservative_bot', 'equity_bot','maniac_bot']
earnings = OrderedDict()
## for each bot to oppose
for opp_algo, opp_name in zip(opp_algos, opp_names):
#first match
my_game_result_1 = 0
cst_deck_match=cst_decks.copy()
deepbot.model.reset()
for i in range(nb_sub_matches):
config = setup_config(max_round=int(nb_hands/nb_sub_matches)-1, initial_stack=ini_stack, small_blind_amount=sb_amount)
config.register_player(name=deepbot.opponent, algorithm=opp_algo)
config.register_player(name="deepbot", algorithm= deepbot)
game_result_1 = start_poker(config, verbose=0, cheat = True, cst_deck_ids = cst_deck_match, return_last_two=False)
#Hacky fix for missing last SB in some results
if game_result_1['players'][1]['stack'] == 2*ini_stack-sb_amount:
game_result_1['players'][1]['stack'] = 2*ini_stack
my_game_result_1 += game_result_1['players'][1]['stack']
if verbose:
print("Stack after first game: "+ str(game_result_1))
#return match
my_game_result_2 = 0
cst_deck_match=cst_decks.copy()
deepbot.model.reset()
for i in range(nb_sub_matches):
config = setup_config(max_round=int(nb_hands/nb_sub_matches)-1, initial_stack=ini_stack, small_blind_amount=sb_amount)
config.register_player(name="deepbot", algorithm=deepbot)
config.register_player(name=deepbot.opponent, algorithm=opp_algo)
game_result_2 = start_poker(config, verbose=0, cheat = True, cst_deck_ids = cst_deck_match, return_last_two=False)
##Fixing issue with missing last SB in certain wins
if game_result_2['players'][0]['stack'] == 2*ini_stack-sb_amount:
game_result_2['players'][0]['stack'] = 2*ini_stack
my_game_result_2 += game_result_2['players'][0]['stack']
if verbose:
print("return game: "+ str(game_result_2['players'][0]['stack']))
earnings[opp_name] = my_game_result_1 + my_game_result_2 - 2*ini_stack
return earnings
def play_round(self, players):
"""
Input:
players: a list of 2 tuples- (player, num) where num is the index of player in self.pop
Output:
a list of the 2 players' payoffs
"""
# if (len(players) != 2):
# raise ValueError('The genetic algo fitness game has more than 2 players.')
config = setup_config(max_round=MAX_ROUNDS,
initial_stack=INITIAL_STACK,
small_blind_amount=SMALL_BLIND_AMOUNT)
# print("Setting up a new game")
for player, num in players:
# print("Registering player {0}".format(num))
config.register_player(name=num, algorithm=player)
results = start_poker(config, verbose=0)
# print("The final results of the poker game are: ", results)
fitnesses = [0] * self.size
for player in results['players']:
fitnesses[player['name']] = player['stack']
return fitnesses
def trainingFactor(initialValue, adjustingSize, trainingTimes, numOfValues):
RaisedPlayer.raiseThreshold = initialValue
maxValue = 0
for j in range(0, numOfValues):
totalResult = 0
print("The factor:", RaisedPlayer.raiseThreshold)
for k in range(0, trainingTimes):
config = setup_config(max_round=100, initial_stack=10000, small_blind_amount=10)
# config.register_player(name="call_player", algorithm=CallPlayer())
# config.register_player(name="random", algorithm=RandomPlayer())
config.register_player(name="runable", algorithm=RaisedPlayer())
config.register_player(name="search", algorithm=SearchPlayer())
game_result = start_poker(config, verbose=1)
print game_result
totalResult += game_result['players'][0]['stack']
# To change the variables of the function
# print("The factor:", AlphaBeta.factors[0])
aveResult = totalResult/trainingTimes
print("The aveResult:", aveResult)
maxValue = max(aveResult, maxValue)
if (aveResult == maxValue):
optimalFactor = RaisedPlayer.raiseThreshold
print("Current Max", maxValue, "current optimalFactor:", optimalFactor)
trainingResult = "raiseThreshold value:" + str(RaisedPlayer.raiseThreshold) + "aveResult:" + str(aveResult)
import cPickle
result = open('trainingResult' + ' raiseThreshold' + str(RaisedPlayer.raiseThreshold)+ '.pkl', 'w')
cPickle.dump(trainingResult, result)
print game_result
RaisedPlayer.raiseThreshold += adjustingSize
return optimalFactor
def create_bachtes(num, num_players):
# check if their are already existing saves
if (os.path.exists(path + state_to_save + "/save0.pickle")):
files = (glob.glob(path + state_to_save + "/save*.pickle"))
last_number = []
for l in range(len(files)):
last_number.append(int(files[l].split('.pic')[0].split('save')[-1]))
# if so set the the save numer to the highest existing savenumber + 1
last_number = max(last_number) + 1
else:
last_number = 0
# notice each game lasts only one round and a new game starts. This should be okay, since most cash players
# usually top their chip amount up if they lost either way (e.g.will always play with a minimum of 50€)
for i in range(num):
config = setup_config(max_round=1, initial_stack=100, small_blind_amount=5)
# determine the seat of the player randomly
own_seat = randint(1, num_players)
for x in range(1, own_seat):
# the current algorithm(fishPlayer) will always call, hence we will definetly reach the end state
# of the game if the TrainingsGenerator won't go all-in
config.register_player(name=("p", x), algorithm=FishPlayer())
config.register_player("trainingsGenerator", algorithm=TrainingsGenerator(-1, state_to_save,
path, last_number))
for x in range(own_seat+1, num_players+1):
config.register_player(name=("p", x), algorithm=FishPlayer())
# if set to 1 the winner of each game will be printed with some additionally information
game_result = start_poker(config, verbose = 0)
# print("I: ", i)
last_number += 1
def play_round(players):
"""
Input:
players: a list of PLAYERS_PER_TABLE tuples- (player, num) where num is the index of player in population
Output:
- order of the players
- their corresponding payoff probabilities
"""
config = setup_config(max_round=MAX_GAME_ROUNDS,
initial_stack=INITIAL_STACK,
small_blind_amount=SMALL_BLIND_AMOUNT)
# print("Setting up a new game")
for player, num in players:
config.register_player(name=num, algorithm=player)
results = start_poker(config, verbose=0)
# print("The final results of the poker game are: ", results)
playerOrder = []
fitnesses = []
for player in results['players']:
playerOrder.append(player['name'])
fitnesses.append(player['stack'])
return playerOrder, normalize_list(fitnesses)
def init():
print("hello world")
qtable = make_table()
config = setup_config(max_round=10,
initial_stack=10000,
small_blind_amount=10)
player = LearningPlayer()
config.register_player(name="rand", algorithm=RandomPlayer())
config.register_player(name="learn", algorithm=player)
game_result = start_poker(config, verbose=1)
for k in range(len(player.gameHistory)):
game = player.gameHistory[k]
print(("=" * 5 + " Game {} " + "=" * 5).format(k + 1))
for n in range(len(game['action_history'])):
i = game['action_history'][n]
print(
("{:<10} {:<4} {:<2} {:<2} - {:<20} ({:<4}) {:<5} {}").format(
i['street'], i['pot'], i['hole_cards'][0],
i['hole_cards'][1], " ".join(i['community_cards']),
i['ehs'], i['action'], i['valid_actions']))
# print("\n".join(list(map(lambda p: "{}: {}".format(p['name'], p['stack']), game_result['players']))))
print(game['result']['name'], 'won pot of', game['pot'])
print("stack sizes: {} {}".format(game['self_stack'],
game['opp_stack']))
def test_set_blind_structure(self):
config = G.setup_config(1, 100, 10)
config.register_player("p1", FoldMan())
config.register_player("p2", FoldMan())
config.set_blind_structure({ 1: { "ante":5, "small_blind":10 } })
result = G.start_poker(config)
p1, p2 = [result["players"][i] for i in range(2)]
self.eq(115, p1["stack"])
self.eq(85, p2["stack"])
def test_set_blind_structure(self):
config = G.setup_config(1, 100, 10)
config.register_player("p1", FoldMan())
config.register_player("p2", FoldMan())
config.set_blind_structure({1: {"ante": 5, "small_blind": 10}})
result = G.start_poker(config)
p1, p2 = [result["players"][i] for i in range(2)]
self.eq(115, p1["stack"])
self.eq(85, p2["stack"])
def test_set_blind_structure(self):
config = G.setup_config(1, 100, 10)
config.register_player('p1', FoldMan())
config.register_player('p2', FoldMan())
config.set_blind_structure({1: {'ante': 5, 'small_blind': 10}})
result = G.start_poker(config)
p1, p2 = [result['players'][i] for i in range(2)]
self.eq(115, p1['stack'])
self.eq(85, p2['stack'])
def play_game(agentOne, agentTwo, rounds):
# Game Settings
config = setup_config(max_round=rounds,
initial_stack=10000,
small_blind_amount=20)
config.register_player(name="Player 1 ", algorithm=agentOne)
config.register_player(name="Player 2 ", algorithm=agentTwo)
game_result = start_poker(config, verbose=0)
return game_result
def test_start_poker(self):
config = G.setup_config(1, 100, 10)
config.register_player("p1", FoldMan())
config.register_player("p2", FoldMan())
result = G.start_poker(config)
p1, p2 = [result["players"][i] for i in range(2)]
self.eq("p1", p1["name"])
self.eq(110, p1["stack"])
self.eq("p2", p2["name"])
self.eq(90, p2["stack"])
def test_start_poker(self):
config = G.setup_config(1, 100, 10)
config.register_player('p1', FoldMan())
config.register_player('p2', FoldMan())
result = G.start_poker(config)
p1, p2 = [result['players'][i] for i in range(2)]
self.eq('p1', p1['name'])
self.eq(110, p1['stack'])
self.eq('p2', p2['name'])
self.eq(90, p2['stack'])
def test_start_poker(self):
config = G.setup_config(1, 100, 10)
config.register_player("p1", FoldMan())
config.register_player("p2", FoldMan())
result = G.start_poker(config)
p1, p2 = [result["players"][i] for i in range(2)]
self.eq("p1", p1["name"])
self.eq(110, p1["stack"])
self.eq("p2", p2["name"])
self.eq(90, p2["stack"])
def simulate(self, nGames, nRounds, initial_stack, small_blind_amount):
# since no api is provided for simulate games with known hole_card value, we just simulate nRounds using random player for convenience purpose
results = []
config = setup_config(max_round=nRounds, initial_stack=initial_stack, small_blind_amount=small_blind_amount)
config.register_player(name="f1", algorithm=RandomPlayer())
config.register_player(name="FT2", algorithm=RandomPlayer())
for i in range(nGames):
game_result = start_poker(config, verbose=0)
results.append(game_result['players'][0]['stack'] - game_result['rule']['initial_stack'])
print(sum(results)/len(results))
def play_game(agent_one, agent_two, rounds):
# Game Settings
agent_one_name = agent_one[0]
agent_one_player = agent_one[1].getPlayer()
agent_two_name = agent_one[0]
agent_two_player = agent_two[1].getPlayer()
config = setup_config(max_round=rounds, initial_stack=10000, small_blind_amount=20)
config.register_player(name=agent_one_name, algorithm=agent_one_player)
config.register_player(name=agent_two_name, algorithm=agent_two_player)
game_result = start_poker(config, verbose=0)
return game_result
def trail_play():
from pypokerengine.api.game import setup_config, start_poker
config = setup_config(max_round=10,
initial_stack=100,
small_blind_amount=5)
config.register_player(name="p1", algorithm=FishPlayer())
config.register_player(name="p2", algorithm=FishPlayer())
config.register_player(name="p3", algorithm=FishPlayer())
game_result = start_poker(config, verbose=1)
print(game_result)
def play_round(self, players):
## Input: players is a list of tuples (player, num) where num is the index of player in self.pop
## Output: a list of their payoffs
config = setup_config(max_round=20, initial_stack=200, small_blind_amount=1)
print("Setting up a new table")
for player, num in players:
print("Welcoming player {0}".format(num))
config.register_player(name=num, algorithm=player)
results = start_poker(config, verbose=0)
print("The final results of the poker tournament are: ", results)
fitnesses = [0] * self.size
for player in results['players']:
fitnesses[player['name']] = player['stack']
return fitnesses
def init():
config = setup_config(max_round=MAX_ROUND, initial_stack=10000, small_blind_amount=10)
player = LearningPlayer(LOG_LEVEL)
player.load_qtable_from_file("q-table-rvp-400.txt")
player2 = RVPlayer()
# player2.load_qtable_from_file("gen-0-700-fixed.txt")
atexit.register(exithandler, player, player2)
config.register_player(name="RVPlayer", algorithm=player2)
config.register_player(name="learn1", algorithm=player)
total_start_time = time.time()
total_rounds = 0
last_unseen = 0
last_seen = 0
for i in range(NUM_GAMES):
start_time = time.time()
game_result = start_poker(config, verbose=LOG_LEVEL)
window['game_count'] += 1
num_rounds = player.num_rounds_this_game
total_rounds += num_rounds
print("Game #{} {} rounds - {}s".format(i + 1, num_rounds, time.time() - start_time))
# print("Unseen hands: {}".format(player.unseen_hands))
print("Seen/Unseen states: {} {}".format(player.seen_states - last_seen, player.unseen_states - last_unseen))
last_unseen = player.unseen_states
last_seen = player.seen_states
print("\n".join(list(map(lambda p: "{}: {}".format(p['name'], p['stack']), game_result['players']))))
print()
if i > 0 and i % DUMP_INTERVAL == 0:
player.write_table(i)
print("Unseen hands: {}".format(player.unseen_hands))
print("Unseen states: {}".format(player.unseen_states))
print("Total - {} games, {} rounds, {}s".format(NUM_GAMES, total_rounds, time.time() - total_start_time))
for k in range(len(player.gameHistory)):
game = player.gameHistory[k]
print(("=" * 5 + " Game {} " + "=" * 5).format(k + 1))
for n in range(len(game['action_history'])):
i = game['action_history'][n]
print("{:<10} {:<4} {:<2} {:<2} - {:<20} ({:<4}) {:<5} {} {}".format(i['street'], i['pot'], i['hole_cards'][0], i['hole_cards'][1], " ".join(i['community_cards']), i['ehs'], i['action'], i['decision'], i['choices']))
# print("\n".join(list(map(lambda p: "{}: {}".format(p['name'], p['stack']), game_result['players']))))
print(game['result']['name'], 'won pot of', game['pot'])
print("stack sizes: {} {}".format(game['self_stack'], game['opp_stack']))
def initiate_game(config):
# Init pot of players
win_cnt = 0
for game in range(1, 2):
print "Game number: %d" % (game)
start = time.time()
game_result = start_poker(config, verbose=0)
end = time.time()
print "Time take to play: %.4f seconds" % (end - start)
agent1_pot = game_result['players'][0]['stack']
agent2_pot = game_result['players'][1]['stack']
print 'Player 1 pot: %d' % (agent1_pot)
print 'Player 2 pot: %d' % (agent2_pot)
if agent1_pot > agent2_pot:
win_cnt += 1
return win_cnt
def play_game(max_round, names, value_functions, with_me=False, verbose=1):
if with_me:
assert len(value_functions) == 9
else:
assert len(value_functions) == 10
task = TexasHoldemTask()
config = setup_config(max_round=max_round,
initial_stack=10000,
small_blind_amount=25)
config.set_blind_structure(blind_structure)
for name, value_func in zip(names, value_functions):
config.register_player(
name, PokerPlayer(task, value_func, debug=verbose == 1))
if with_me:
config.register_player("console", ConsolePlayer())
return start_poker(config, verbose=verbose)
def train(rank, shared_model, optimizer):
my_player = A3Cplayer(rank, shared_model, optimizer)
config = setup_config(max_round=50,
initial_stack=1500,
small_blind_amount=15)
config.register_player(name="p1", algorithm=my_player)
config.register_player(name="p2", algorithm=HonestPlayer())
config.register_player(name="p3", algorithm=AggressivePlayer())
config.register_player(name="p4", algorithm=RandomPlayer())
config.register_player(name="p5", algorithm=CallerPlayer())
for i_episode in range(1000000):
game_result = start_poker(config, verbose=0)
if my_player.epsilon > 0:
my_player.epsilon -= 0.05
def main(coef_raise, coef):
model1 = NN_3J()
model1.load_state_dict(torch.load('model_hyppo_3J.pth'))
model1.eval()
mean3J = np.asarray([
25.62989439, 506.31789281, 500.85171349, 492.83039371, 8.46048567,
29.11114304, 35.52415763, 127.08888929
])
var3J = np.asarray([
77.21836346, 20549.21166419, 23143.48462078, 23994.43942467,
208.35494771, 6671.09516214, 8899.36275071, 22232.29371878
])
model2 = NN_2J()
model2.load_state_dict(torch.load('model_hyppo_2J.pth'))
model2.eval()
mean2J = np.asarray([
37.99850064, 665.80849344, 692.35777389, 15.20394244, 56.24048568,
160.47079922
])
var2J = np.asarray([
226.25168453, 59823.15358233, 91017.39574507, 440.05912923,
23585.63323956, 41242.62439903
])
config = setup_config(max_round=1000,
initial_stack=500,
small_blind_amount=10)
config.register_player(name="h1",
algorithm=HipoPlayer(model3J=model1,
mean3J=mean3J,
var3J=var3J,
model2J=model2,
mean2J=mean2J,
var2J=var2J,
coef_raise=coef_raise,
coef=coef))
#config.register_player(name="r1", algorithm=RandomPlayer())
config.register_player(name="f1", algorithm=CallBot())
config.register_player(name="f2", algorithm=CallBot())
game_result = start_poker(config, verbose=0)
if (game_result["players"][0]["stack"] > 1000):
return 1
else:
return 0
def start_real_game(self, players, ai_version='5'):
ai_params = {
'3': {
'class': DQNPlayerV3And4,
'model': '3200'
},
'4': {
'class': DQNPlayerV3And4,
'model': '9999'
},
'5': {
'class': DQNPlayerV5,
'model': '19999'
},
}
path = './models/v' + ai_version + '/model_v' + ai_version + '-' + ai_params[
ai_version]['model'] + '.ckpt'
tf.reset_default_graph()
main_qn = ai_params[ai_version]['class'](
learning_rate=self.learning_rate,
discount=self.y,
nb_players=self.nb_players,
start_stack=self.start_stack,
max_round=self.max_rounds)
init = tf.global_variables_initializer()
self.saver = tf.train.Saver()
with tf.Session() as sess:
sess.run(init)
main_qn.set_session(sess)
self.saver.restore(sess, path)
config = setup_config(max_round=self.max_rounds,
initial_stack=self.start_stack,
small_blind_amount=5)
i = 1
for player in players:
config.register_player(
name='p' + str(i),
algorithm=player['class'](**player['kwargs']))
i += 1
config.register_player(name='p' + str(i), algorithm=main_qn)
game_result = start_poker(config, verbose=0)
return game_result
def eval_player(num_evals=5, other_bots=[HonestPlayer(1000) for _ in range(8)],
my_bot=HonestPlayer2(1000)):
config = setup_config(max_round=50, initial_stack=1500, small_blind_amount=15)
# add my player
config.register_player(name="my_bot", algorithm=my_bot)
# add other players
for i, b in enumerate(other_bots):
config.register_player(name="other_{}".format(i + 1), algorithm=b)
result_dict = {}
for _ in range(num_evals):
game_result = start_poker(config, verbose=0)
# add results
for k in game_result['players']:
result_dict.setdefault(k['name'], 0)
result_dict[k['name']] += k['stack'] / num_evals
return result_dict
def init():
config = setup_config(max_round=MAX_ROUND,
initial_stack=10000,
small_blind_amount=10)
player = LearningPlayer(LOG_LEVEL)
player.load_qtable_from_file("gen-2-100.txt")
player2 = RVPlayer()
# player2.load_qtable_from_file("gen-0-700-fixed.txt")
atexit.register(exithandler, player, player2)
config.register_player(name="RVPlayer", algorithm=player2)
config.register_player(name="learn1", algorithm=player)
total_start_time = time.time()
total_rounds = 0
last_unseen = 0
for i in range(NUM_GAMES):
start_time = time.time()
game_result = start_poker(config, verbose=LOG_LEVEL)
window['game_count'] += 1
num_rounds = player.num_rounds_this_game
total_rounds += num_rounds
print("Game #{} {} rounds - {}s".format(i + 1, num_rounds,
time.time() - start_time))
print("Unseen hands: {}".format(player.unseen_hands))
print("Unseen states: {}".format(player.unseen_states - last_unseen))
last_unseen = player.unseen_states
print("\n".join(
list(
map(lambda p: "{}: {}".format(p['name'], p['stack']),
game_result['players']))))
print()
if i > 0 and i % DUMP_INTERVAL == 0:
player.write_table(i)
print("Unseen hands: {}".format(player.unseen_hands))
print("Unseen states: {}".format(player.unseen_states))
print("Total - {} games, {} rounds, {}s".format(
NUM_GAMES, total_rounds,
time.time() - total_start_time))
def play_game(game_no, conf):
n_players = conf['n_players']
config = setup_config(
max_round=conf['max_round'],
initial_stack=conf['initial_stack'],
small_blind_amount=conf['small_blind_amount'],
summary_file=None)
bots = list()
for bot in conf['bots']:
m = importlib.import_module(bot['module'])
cl = getattr(m, bot['class'])
props = bot.get('properties', dict())
bots.append((cl, props))
players = np.random.choice(len(bots), n_players, replace=False)
for idx in players:
pl_cls, pl_props = bots[idx]
pl = pl_cls(pl_props)
config.register_player(name=pl.__class__.__name__, algorithm=pl)
print('Starting game {}'.format(game_no))
game_result = start_poker(config, verbose=0)
pre_result = dict()
for p in game_result['players']:
nm = p['name']
stack = p['stack']
if nm in pre_result:
pre_result[nm] = (pre_result[nm][0] + stack, pre_result[nm][1] + 1)
else:
pre_result[nm] = (stack, 1)
result = dict((nm, stat[0]/float(stat[1])) for nm, stat in pre_result.items())
for name, score in result.items():
print('Game #{}, {}: {}'.format(game_no, name, score))
return result
def test(rank, shared_model):
torch.manual_seed(123)
my_player = A3Cplayer(rank, shared_model, training=False)
my_player.epsilon = 0
config = setup_config(max_round=50,
initial_stack=1500,
small_blind_amount=15)
config.register_player(name="p1", algorithm=my_player)
config.register_player(name="p2", algorithm=HonestPlayer())
config.register_player(name="p3", algorithm=AggressivePlayer())
config.register_player(name="p4", algorithm=RandomPlayer())
config.register_player(name="p5", algorithm=CallerPlayer())
for i_episode in range(1000000):
game_result = start_poker(config, verbose=0)
print(
rank, game_result['players'][0],
"random {} real {}".format(my_player.random_action,
my_player.real_action))
my_player.random_action = 0
my_player.real_action = 0
def run_evaluation(self, verbose: int = 0) -> Dict[str, int]:
if len(self.config.players_info) <= 1:
raise AssertionError("At least two players have to be seated.")
game_result = start_poker(self.config,
verbose=verbose,
cashgame=True,
log_file_location=self.log_file_location)
# Prettify game result:
for player in game_result['players']:
self.player_final_stack[player['name']] = player['cashgame_stack']
self.player_final_stack = {
k: v
for k, v in sorted(self.player_final_stack.items(),
key=lambda item: item[1],
reverse=True)
}
# Print out game result
for rank, (name, stack) in enumerate(self.player_final_stack.items()):
print(f"{rank + 1:2}. Player: {name:>25}, Stack: {stack:n}")
return self.player_final_stack
def test_start_poker_validation_when_one_player(self):
config = G.setup_config(1, 100, 10)
config.register_player("p1", FoldMan())
with self.assertRaises(Exception) as e:
result = G.start_poker(config)
self.assertIn("only 1 player", str(e.exception))
def test_start_poker_validation_when_one_player(self):
config = G.setup_config(1, 100, 10)
config.register_player('p1', FoldMan())
with self.assertRaises(Exception) as e:
result = G.start_poker(config)
self.assertIn('only 1 player', str(e.exception))
def test_start_poker_validation_when_no_player(self):
config = G.setup_config(1, 100, 10)
with self.assertRaises(Exception) as e:
result = G.start_poker(config)
self.assertIn('no player', str(e.exception))
from pypokerengine.api.game import setup_config, start_poker from NNPlayer import NNPlayer config = setup_config(max_round=10, initial_stack=100, small_blind_amount=5) config.register_player(name="p1", algorithm=NNPlayer()) config.register_player(name="p2", algorithm=NNPlayer()) game_result = start_poker(config, verbose=1)
def test_start_poker_validation_when_no_player(self):
config = G.setup_config(1, 100, 10)
with self.assertRaises(Exception) as e:
result = G.start_poker(config)
self.assertIn("no player", str(e.exception))
import numpy as np
from pypokerengine.api.game import start_poker, setup_config
from .players.fish_player import FishPlayer
from .players.monte_carlo_player import MonteCarloPlayer
if __name__ == '__main__':
blogger_bot = MonteCarloPlayer()
# The stack log contains the stacks of the Data Blogger bot after each game (the initial stack is 100)
stack_log = []
for round in range(1000):
p1, p2 = blogger_bot, FishPlayer()
config = setup_config(max_round=5, initial_stack=100, small_blind_amount=5)
config.register_player(name="p1", algorithm=p1)
config.register_player(name="p2", algorithm=p2)
game_result = start_poker(config, verbose=0)
stack_log.append([player['stack'] for player in game_result['players'] if player['uuid'] == blogger_bot.uuid])
print('Avg. stack:', '%d' % (int(np.mean(stack_log))))