forked from ehp/udacity-drl-navigation
/
agent.py
173 lines (140 loc) · 6.79 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
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
import numpy as np
import random
from model import NoisyDuelingQNetwork, DuelingQNetwork, QNetwork
from buffer import PrioritizedReplayBuffer, ReplayBuffer
import torch
import torch.nn.functional as F
import torch.optim as optim
class Agent():
"""Interacts with and learns from the environment."""
def __init__(self, state_size, action_size, seed, training, args):
"""Initialize an Agent object.
Params
======
state_size (int): dimension of each state
action_size (int): dimension of each action
seed (int): random seed
training (bool): Prepare for training
args (object): Command line arguments
"""
self.state_size = state_size
self.action_size = action_size
random.seed(seed)
self.seed = seed
nn_type = args.type.lower()
self._sample_noise = False
self._update_buffer_priorities = False
if args.cuda:
self.device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
else:
self.device = "cpu"
# NN
if training:
self.batch_size = args.batch_size
self.gamma = args.gamma
self.tau = args.tau
self.update_every = args.update_every
self.qnetwork_local = self._create_nn(nn_type, state_size, action_size, self.seed, self.device)
self.qnetwork_target = self._create_nn(nn_type, state_size, action_size, self.seed, self.device)
self.optimizer = optim.Adam(self.qnetwork_local.parameters(), lr=args.learning_rate)
# Replay memory
self.memory = self._create_buffer(args.buffer.lower(), action_size, args.buffer_size,
self.batch_size, args.alpha, args.beta, self.seed, self.device)
# Initialize time step (for updating every UPDATE_EVERY steps)
self.t_step = 0
else:
self.qnetwork_local = self._create_nn(nn_type, state_size, action_size, self.seed, self.device)
def _create_buffer(self, buffer_type, action_size, buffer_size, batch_size, alpha, beta, seed, device):
if buffer_type == 'prioritized':
self._update_buffer_priorities = True
return PrioritizedReplayBuffer(action_size, buffer_size, batch_size, seed, alpha=alpha, beta=beta, device=device)
elif buffer_type == 'sample':
return ReplayBuffer(action_size, buffer_size, batch_size, seed, device=device)
else:
raise Exception('Unknown buffer type - must be one of prioritized or sample')
def _create_nn(self, nn_type, state_size, action_size, seed, device):
if nn_type == 'noisydueling':
self._sample_noise = True
return NoisyDuelingQNetwork(state_size, action_size, seed, device=device).to(device)
elif nn_type == 'dueling':
return DuelingQNetwork(state_size, action_size, seed).to(device)
elif nn_type == 'q':
return QNetwork(state_size, action_size, seed).to(device)
else:
raise Exception('Unknown NN type - must be one of NoisyDueling, Dueling or Q')
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 enough samples are available in memory, get random subset and learn
if len(self.memory) > self.batch_size:
experiences = self.memory.sample()
self.learn(experiences, self.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(self.device)
self.qnetwork_local.eval()
with torch.no_grad():
if self._sample_noise:
self.qnetwork_local.sample_noise()
action_values = self.qnetwork_local(state)
self.qnetwork_local.train()
# Epsilon-greedy action selection
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.Tensor]): tuple of (s, a, r, s', done) tuples
gamma (float): discount factor
"""
if self._update_buffer_priorities:
states, actions, rewards, next_states, dones, indexes, weights = experiences
else:
states, actions, rewards, next_states, dones = experiences
if self._sample_noise:
self.qnetwork_target.sample_noise()
self.qnetwork_local.sample_noise()
# Get max predicted Q values (for next states) from target model
Q_targets_next = self.qnetwork_target(next_states).detach().max(1)[0].unsqueeze(1)
# Compute Q targets for current states
Q_targets = rewards + (gamma * Q_targets_next * (1 - dones))
# Get expected Q values from local model
Q_expected = self.qnetwork_local(states).gather(1, actions)
# Update memory priorities
if self._update_buffer_priorities:
self.memory.update_priorities(indexes, (Q_expected - Q_targets).detach().squeeze().abs().cpu().numpy().tolist())
# Compute loss
if self._update_buffer_priorities:
loss = (F.mse_loss(Q_expected, Q_targets) * weights).mean()
else:
loss = F.mse_loss(Q_expected, Q_targets)
# Minimize the loss
self.optimizer.zero_grad()
loss.backward()
for param in self.qnetwork_local.parameters():
param.grad.data.clamp_(-1, 1)
self.optimizer.step()
# ------------------- update target network ------------------- #
self.soft_update(self.qnetwork_local, self.qnetwork_target, self.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)