(A5) 빅데이터 혁신공유대학 여름캠프 (최규빈)
강화학습, LunarLander
본 강의노트는 LG전자 인공지능연구소 고한규박사님께 공유받았습니다.
# !sudo apt-get install -y xvfb python-opengl ffmpeg > /dev/null 2>&1
# !pip install rarfile --quiet
# !pip install stable-baselines3[extra] ale-py==0.7.4 --quiet
# !pip install box2d-py --quiet
# !pip install gym pyvirtualdisplay --quiet
# !pip install pyglet --quiet
# !pip install piglet --quiet
from pyvirtualdisplay import Display
display = Display(visible=0, size=(1400, 900))
display.start()
import gym
from gym.wrappers import Monitor
import random
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
import matplotlib.pyplot as plt
import base64, io
import numpy as np
from collections import deque, namedtuple
# For visualization
from gym.wrappers.monitoring import video_recorder
from IPython.display import HTML
from IPython import display
import glob
env = gym.make('LunarLander-v2')
env.seed(0)
print('State shape: ', env.observation_space.shape)
print('Number of actions: ', env.action_space.n)
class QNetwork(nn.Module):
"""Actor (Policy) Model."""
def __init__(self, state_size, action_size, seed):
"""Initialize parameters and build model.
Params
======
state_size (int): Dimension of each state
action_size (int): Dimension of each action
seed (int): Random seed
"""
super(QNetwork, self).__init__()
self.seed = torch.manual_seed(seed)
self.layer1 = nn.Linear(state_size, 128)
self.layer2 = nn.Linear(128, 64)
self.layer3 = nn.Linear(64, 32)
self.layer4 = nn.Linear(32, action_size)
def forward(self, state):
"""Build a network that maps state -> action values."""
x = F.relu(self.layer1(state))
x = F.relu(self.layer2(x))
x = F.relu(self.layer3(x))
x = self.layer4(x)
return x
BUFFER_SIZE = 50000 # replay buffer size
BATCH_SIZE = 128 # minibatch size
GAMMA = 0.99 # discount factor
TAU = 0.001 # for soft update of target parameters
LR = 0.0001 # learning rate
UPDATE_EVERY = 1 # how often to update the network
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
device
class ReplayBuffer:
"""Fixed-size buffer to store experience tuples."""
def __init__(self, action_size, buffer_size, batch_size, seed):
"""Initialize a ReplayBuffer object.
Params
======
action_size (int): dimension of each action
buffer_size (int): maximum size of buffer
batch_size (int): size of each training batch
seed (int): random seed
"""
self.action_size = action_size
self.memory = deque(maxlen=buffer_size)
self.batch_size = batch_size
self.experience = namedtuple("Experience", field_names=["state", "action", "reward", "next_state", "done"])
self.seed = random.seed(seed)
def add(self, state, action, reward, next_state, done):
"""Add a new experience to memory."""
e = self.experience(state, action, reward, next_state, done)
self.memory.append(e)
def sample(self):
"""Randomly sample a batch of experiences from memory."""
experiences = random.sample(self.memory, k=self.batch_size)
# Convert to torch tensors
states = torch.from_numpy(np.vstack([e.state for e in experiences if e is not None])).float().to(device)
actions = torch.from_numpy(np.vstack([e.action for e in experiences if e is not None])).long().to(device)
rewards = torch.from_numpy(np.vstack([e.reward for e in experiences if e is not None])).float().to(device)
next_states = torch.from_numpy(np.vstack([e.next_state for e in experiences if e is not None])).float().to(device)
# Convert done from boolean to int
dones = torch.from_numpy(np.vstack([e.done for e in experiences if e is not None]).astype(np.uint8)).float().to(device)
return (states, actions, rewards, next_states, dones)
def __len__(self):
"""Return the current size of internal memory."""
return len(self.memory)
class Agent():
"""Interacts with and learns from the environment."""
def __init__(self, state_size, action_size, seed):
"""Initialize an Agent object.
Params
======
state_size (int): dimension of each state
action_size (int): dimension of each action
seed (int): random seed
"""
self.state_size = state_size
self.action_size = action_size
self.seed = random.seed(seed)
# Q-Network
self.qnetwork_local = QNetwork(state_size, action_size, seed).to(device)
self.qnetwork_target = QNetwork(state_size, action_size, seed).to(device)
self.optimizer = optim.Adam(self.qnetwork_local.parameters(), lr=LR)
# Replay memory
self.memory = ReplayBuffer(action_size, BUFFER_SIZE, BATCH_SIZE, seed)
# Initialize time step (for updating every UPDATE_EVERY steps)
self.t_step = 0
def step(self, state, action, reward, next_state, done):
# Save experience in replay memory
self.memory.add(state, action, reward, next_state, done)
# Learn every UPDATE_EVERY time steps.
self.t_step = (self.t_step + 1) % UPDATE_EVERY
if self.t_step == 0:
# If enough samples are available in memory, get random subset and learn
if len(self.memory) > BATCH_SIZE:
experiences = self.memory.sample()
self.learn(experiences, GAMMA)
def act(self, state, eps=0.):
"""Returns actions for given state as per current policy.
Params
======
state (array_like): current state
eps (float): epsilon, for epsilon-greedy action selection
"""
state = torch.from_numpy(state).float().unsqueeze(0).to(device)
self.qnetwork_local.eval()
with torch.no_grad():
action_values = self.qnetwork_local(state)
self.qnetwork_local.train()
if random.random() > eps:
return np.argmax(action_values.cpu().data.numpy())
else:
return random.choice(np.arange(self.action_size))
def learn(self, experiences, gamma):
"""Update value parameters using given batch of experience tuples.
Params
======
experiences (Tuple[torch.Variable]): tuple of (s, a, r, s', done) tuples
gamma (float): discount factor
"""
# Obtain random minibatch of tuples from D
states, actions, rewards, next_states, dones = experiences
## Compute and minimize the loss
### Extract next maximum estimated value from target network
q_targets_next = self.qnetwork_target(next_states).detach().max(1)[0].unsqueeze(1)
q_targets = rewards + gamma * q_targets_next * (1 - dones)
### Calculate expected value from local network
q_expected = self.qnetwork_local(states).gather(1, actions)
loss = F.mse_loss(q_expected, q_targets)
self.optimizer.zero_grad()
loss.backward()
self.optimizer.step()
# ------------------- update target network ------------------- #
self.soft_update(self.qnetwork_local, self.qnetwork_target, TAU)
def soft_update(self, local_model, target_model, tau):
"""Soft update model parameters.
θ_target = τ*θ_local + (1 - τ)*θ_target
Params
======
local_model (PyTorch model): weights will be copied from
target_model (PyTorch model): weights will be copied to
tau (float): interpolation parameter
"""
for target_param, local_param in zip(target_model.parameters(), local_model.parameters()):
target_param.data.copy_(tau*local_param.data + (1.0-tau)*target_param.data)
def dqn(n_episodes=2000, max_t=500, eps_start=1.0, eps_end=0.01, eps_decay=0.995):
"""Deep Q-Learning.
Params
======
n_episodes (int): maximum number of training episodes
max_t (int): maximum number of timesteps per episode
eps_start (float): starting value of epsilon, for epsilon-greedy action selection
eps_end (float): minimum value of epsilon
eps_decay (float): multiplicative factor (per episode) for decreasing epsilon
"""
playtimes=[]
scores = [] # list containing scores from each episode
scores_window = deque(maxlen=100) # last 100 scores
eps = eps_start # initialize epsilon
for i_episode in range(1, n_episodes+1):
state = env.reset()
score = 0
for t in range(max_t):
## STEP1:
action = agent.act(state, eps)
## STEP2:
next_state, reward, done, _ = env.step(action)
## STEP3:
agent.step(state, action, reward, next_state, done)
## STEP4:
state = next_state
## STEP5:
score += reward
if done:
playtimes.append(t)
break
scores_window.append(score) # save most recent score
scores.append(score) # save most recent score
eps = max(eps_end, eps_decay*eps) # decrease epsilon
print('\rEpisode {}\tAverage Score: {:.2f}\tPlaytime: {:.2f}'.format(i_episode, np.mean(scores_window), np.mean(playtimes)), end="")
if i_episode % 100 == 0:
print('\rEpisode {}\tAverage Score: {:.2f}\tPlaytime: {:.2f}'.format(i_episode, np.mean(scores_window), np.mean(playtimes)))
torch.save(agent.qnetwork_local.state_dict(), 'checkpoint.pth')
if np.mean(scores_window)>=200.0:
print('\nEnvironment solved in {:d} episodes!\tAverage Score: {:.2f}'.format(i_episode, np.mean(scores_window)))
torch.save(agent.qnetwork_local.state_dict(), 'checkpoint.pth')
break
return scores
agent = Agent(state_size=8, action_size=4, seed=0)
scores = dqn()
fig = plt.figure()
ax = fig.add_subplot(111)
plt.plot(np.arange(len(scores)), scores)
plt.ylabel('Score')
plt.xlabel('Episode #')
plt.show()
def show_video(env_name):
mp4list = glob.glob('video/*.mp4')
if len(mp4list) > 0:
mp4 = mp4list[0]
video = io.open(mp4, 'r+b').read()
encoded = base64.b64encode(video)
display.display(HTML(data='''<video alt="test" autoplay
loop controls style="height: 400px;">
<source src="data:video/mp4;base64,{0}" type="video/mp4" />
</video>'''.format(encoded.decode('ascii'))))
else:
print("Could not find video")
def wrap_env(env):
env = Monitor(env, './video', force=True)
return env
def gen_wrapped_env(env_name):
return wrap_env(gym.make(env_name))
def show_video_of_model(agent, env_name):
env = gen_wrapped_env(env_name)
agent.qnetwork_local.load_state_dict(torch.load('checkpoint.pth'))
state = env.reset()
done = False
while not done:
frame = env.render(mode='rgb_array')
action = agent.act(state)
state, reward, done, _ = env.step(action)
env.close()
agent = Agent(state_size=8, action_size=4, seed=0)
show_video_of_model(agent, 'LunarLander-v2')
show_video('LunarLander-v2')