本文整理汇总了Python中memory.Memory方法的典型用法代码示例。如果您正苦于以下问题：Python memory.Memory方法的具体用法？Python memory.Memory怎么用？Python memory.Memory使用的例子？那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类memory的用法示例。
在下文中一共展示了memory.Memory方法的10个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: __init__
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def __init__(self, mmemory, typeaccess=0) :
if not isinstance(mmemory, Memory):
raise TypeError("ERREUR")
self.mmemory = mmemory
self.mmemory.open("r", typeaccess)
try :
fichier = open("/usr/include/asm/unistd.h", "r")
except IOError :
print "No such file /usr/include/asm/unistd.h"
sys.exit(-1)
liste = fichier.readlines()
fichier.close()
count = 0
for i in liste :
if(re.match("#define __NR_", i)) :
l = string.split(i)
if(l[2][0].isdigit()) :
count = string.atoi(l[2], 10)
self.lists_syscalls.append([count, l[1][5:]])
else :
count = count + 1
self.lists_syscalls.append([count, l[1][5:]]) 示例2: get_memory
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def get_memory(self):
cls = memory.LSHMemory if self.use_lsh else memory.Memory
return cls(self.rep_dim, self.memory_size, self.vocab_size) 示例3: __init__
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def __init__(self, global_model, global_average_model, global_optimizer, global_ep, global_ep_r, res_queue, name):
super(Worker, self).__init__()
self.env = gym.make(env_name)
self.env.seed(500)
self.name = 'w%i' % name
self.global_ep, self.global_ep_r, self.res_queue = global_ep, global_ep_r, res_queue
self.global_model, self.global_average_model, self.global_optimizer = global_model, global_average_model, global_optimizer
self.local_model = LocalModel(self.env.observation_space.shape[0], self.env.action_space.n)
self.num_actions = self.env.action_space.n
self.memory = Memory(replay_memory_capacity) 示例4: __init__
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def __init__(self, mmemory, typeaccess=0) :
if not isinstance(mmemory, Memory):
raise TypeError("ERREUR")
self.mmemory = mmemory
self.mmemory.open("r", typeaccess) 示例5: __init__
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def __init__(self, mmemory, typeaccess=0) :
if not isinstance(mmemory, Memory):
raise TypeError("ERREUR")
self.mmemory = mmemory
self.symbols = Symbols(self.mmemory)
self.typeaccess = typeaccess 示例6: run
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def run(self):
epsilon = 1.0
steps = 0
while self.global_ep.value < max_episode:
if self.global_ep_r.value > goal_score:
break
done = False
score = 0
state = self.env.reset()
state = torch.Tensor(state).to(device)
state = state.unsqueeze(0)
memory = Memory(async_update_step)
while not done:
steps += 1
action = self.get_action(state, epsilon)
next_state, reward, done, _ = self.env.step(action)
next_state = torch.Tensor(next_state)
next_state = next_state.unsqueeze(0)
mask = 0 if done else 1
reward = reward if not done or score == 499 else -1
action_one_hot = np.zeros(2)
action_one_hot[action] = 1
memory.push(state, next_state, action_one_hot, reward, mask)
score += reward
state = next_state
epsilon -= 0.00001
epsilon = max(epsilon, 0.1)
if len(memory) == async_update_step or done:
batch = memory.sample()
loss = QNet.train_model(self.online_net, self.target_net, self.optimizer, batch)
memory = Memory(async_update_step)
if done:
self.record(score, epsilon, loss)
break
if steps % update_target == 0:
self.update_target_model()
score = score if score == 500.0 else score + 1
self.res_queue.put(None) 示例7: main
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def main():
env = gym.make(env_name)
env.seed(500)
torch.manual_seed(500)
num_inputs = env.observation_space.shape[0]
num_actions = env.action_space.n
env.close()
global_target_model = Model(num_inputs, num_actions)
global_online_model = Model(num_inputs, num_actions)
global_target_model.train()
global_online_model.train()
global_target_model.load_state_dict(global_online_model.state_dict())
global_target_model.share_memory()
global_online_model.share_memory()
global_memory = Memory(replay_memory_capacity)
global_ep, global_ep_r, res_queue, global_memory_pipe = mp.Value('i', 0), mp.Value('d', 0.), mp.Queue(), mp.Queue()
writer = SummaryWriter('logs')
n = 2
epsilons = [(i * 0.05 + 0.1) for i in range(n)]
actors = [Actor(global_target_model, global_memory_pipe, global_ep, global_ep_r, epsilons[i], i) for i in range(n)]
[w.start() for w in actors]
learner = Learner(global_online_model, global_target_model, global_memory, global_memory_pipe, res_queue)
learner.start()
res = []
while True:
r = res_queue.get()
if r is not None:
res.append(r)
[ep, loss] = r
# writer.add_scalar('log/score', float(ep_r), ep)
writer.add_scalar('log/loss', float(loss), ep)
else:
break
[w.join() for w in actors] 示例8: run
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def run(self):
while self.global_ep.value < max_episode:
self.local_model.pull_from_global_model(self.global_model)
done = False
score = 0
steps = 0
state = self.env.reset()
state = torch.Tensor(state)
state = state.unsqueeze(0)
memory = Memory(n_step)
while True:
policy, value = self.local_model(state)
action = self.get_action(policy, self.num_actions)
next_state, reward, done, _ = self.env.step(action)
next_state = torch.Tensor(next_state)
next_state = next_state.unsqueeze(0)
mask = 0 if done else 1
reward = reward if not done or score == 499 else -1
action_one_hot = torch.zeros(2)
action_one_hot[action] = 1
memory.push(state, next_state, action_one_hot, reward, mask)
score += reward
state = next_state
if len(memory) == n_step or done:
batch = memory.sample()
loss = self.local_model.push_to_global_model(batch, self.global_model, self.global_optimizer)
self.local_model.pull_from_global_model(self.global_model)
memory = Memory(n_step)
if done:
running_score = self.record(score, loss)
break
self.res_queue.put(None) 示例9: main
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def main():
env = gym.make(env_name)
env.seed(500)
torch.manual_seed(500)
num_inputs = env.observation_space.shape[0]
num_actions = env.action_space.n
print('state size:', num_inputs)
print('action size:', num_actions)
net = TNPG(num_inputs, num_actions)
writer = SummaryWriter('logs')
net.to(device)
net.train()
running_score = 0
steps = 0
loss = 0
for e in range(30000):
done = False
memory = Memory()
score = 0
state = env.reset()
state = torch.Tensor(state).to(device)
state = state.unsqueeze(0)
while not done:
steps += 1
action = net.get_action(state)
next_state, reward, done, _ = env.step(action)
next_state = torch.Tensor(next_state)
next_state = next_state.unsqueeze(0)
mask = 0 if done else 1
reward = reward if not done or score == 499 else -1
action_one_hot = torch.zeros(2)
action_one_hot[action] = 1
memory.push(state, next_state, action_one_hot, reward, mask)
score += reward
state = next_state
loss = TNPG.train_model(net, memory.sample())
score = score if score == 500.0 else score + 1
running_score = 0.99 * running_score + 0.01 * score
if e % log_interval == 0:
print('{} episode | score: {:.2f}'.format(
e, running_score))
writer.add_scalar('log/score', float(running_score), e)
writer.add_scalar('log/loss', float(loss), e)
if running_score > goal_score:
break 示例10: main
# 需要导入模块: import memory [as 别名]
# 或者: from memory import Memory [as 别名]
def main():
env = gym.make(env_name)
env.seed(500)
torch.manual_seed(500)
num_inputs = env.observation_space.shape[0]
num_actions = env.action_space.n
print('state size:', num_inputs)
print('action size:', num_actions)
net = TRPO(num_inputs, num_actions)
writer = SummaryWriter('logs')
net.to(device)
net.train()
running_score = 0
steps = 0
loss = 0
for e in range(30000):
done = False
memory = Memory()
score = 0
state = env.reset()
state = torch.Tensor(state).to(device)
state = state.unsqueeze(0)
while not done:
steps += 1
action = net.get_action(state)
next_state, reward, done, _ = env.step(action)
next_state = torch.Tensor(next_state)
next_state = next_state.unsqueeze(0)
mask = 0 if done else 1
reward = reward if not done or score == 499 else -1
action_one_hot = torch.zeros(2)
action_one_hot[action] = 1
memory.push(state, next_state, action_one_hot, reward, mask)
score += reward
state = next_state
loss = TRPO.train_model(net, memory.sample())
score = score if score == 500.0 else score + 1
running_score = 0.99 * running_score + 0.01 * score
if e % log_interval == 0:
print('{} episode | score: {:.2f}'.format(
e, running_score))
writer.add_scalar('log/score', float(running_score), e)
writer.add_scalar('log/loss', float(loss), e)
if running_score > goal_score:
break