-
Notifications
You must be signed in to change notification settings - Fork 1
/
agent.py
147 lines (120 loc) · 6.02 KB
/
agent.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
import numpy as np
import random
from collections import namedtuple, deque
import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
# Define the QNetwork architecture
class QNetwork(nn.Module):
def __init__(self, state_size, action_size, seed, fc1_units=1024, fc2_units=512, fc3_units=256):
super(QNetwork, self).__init__()
self.seed = torch.manual_seed(seed)
self.fc1 = nn.Linear(state_size, fc1_units)
self.fc2 = nn.Linear(fc1_units, fc2_units)
self.fc3 = nn.Linear(fc2_units, fc3_units)
self.fc4 = nn.Linear(fc3_units, action_size)
def forward(self, state):
x = F.relu(self.fc1(state))
x = F.relu(self.fc2(x))
x = F.relu(self.fc3(x))
return self.fc4(x)
class Agent():
def __init__(self, state_size, action_size, seed, buffer_size=int(1e5), batch_size=64, gamma=0.99, tau=1e-3, lr=5e-4, update_every=4, eps_start=1.0, eps_end=0.01, eps_decay=0.995, gradient_clip=1.0):
self.state_size = state_size
self.action_size = action_size
self.seed = random.seed(seed)
self.buffer_size = buffer_size
self.batch_size = batch_size
self.gamma = gamma
self.tau = tau
self.lr = lr
self.update_every = update_every
self.eps = eps_start
self.eps_end = eps_end
self.eps_decay = eps_decay
self.gradient_clip = gradient_clip
# Q-Networks
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)
self.t_step = 0
def save_model(self, file_path):
torch.save({
'qnetwork_local_state_dict': self.qnetwork_local.state_dict(),
'qnetwork_target_state_dict': self.qnetwork_target.state_dict(),
'optimizer_state_dict': self.optimizer.state_dict(),
'eps': self.eps
}, file_path)
def load_model(self, file_path):
checkpoint = torch.load(file_path)
self.qnetwork_local.load_state_dict(checkpoint['qnetwork_local_state_dict'])
self.qnetwork_target.load_state_dict(checkpoint['qnetwork_target_state_dict'])
self.optimizer.load_state_dict(checkpoint['optimizer_state_dict'])
self.eps = checkpoint['eps']
self.qnetwork_local.eval()
self.qnetwork_target.eval()
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) % self.update_every
if self.t_step == 0:
if len(self.memory) > self.batch_size:
experiences = self.memory.sample()
self.learn(experiences, self.gamma)
def act(self, state, eps=None):
if eps is None:
eps = self.eps
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):
states, actions, rewards, next_states, dones = experiences
# Double DQN
Q_targets_next_local = self.qnetwork_local(next_states).detach().max(1)[1].unsqueeze(1)
Q_targets_next = self.qnetwork_target(next_states).detach().gather(1, Q_targets_next_local)
Q_targets = rewards + (gamma * Q_targets_next * (1 - dones))
Q_expected = self.qnetwork_local(states).gather(1, actions)
loss = F.mse_loss(Q_expected, Q_targets)
self.optimizer.zero_grad()
loss.backward()
# Gradient clipping
torch.nn.utils.clip_grad_norm_(self.qnetwork_local.parameters(), self.gradient_clip)
self.optimizer.step()
self.soft_update(self.qnetwork_local, self.qnetwork_target, self.tau)
# Update epsilon
self.eps = max(self.eps_end, self.eps_decay * self.eps)
def soft_update(self, local_model, target_model, tau):
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)
class ReplayBuffer:
def __init__(self, action_size, buffer_size, batch_size, 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):
e = self.experience(state, action, reward, next_state, done)
self.memory.append(e)
def sample(self):
experiences = random.sample(self.memory, k=self.batch_size)
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)
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 len(self.memory)
# Set the device to be used (cuda if available, else cpu)
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")