forked from zhongwen/predictron
/
predictron.py
196 lines (162 loc) · 7.62 KB
/
predictron.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
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
'''
A TensorFlow implementation of
The Predictron: End-To-End Learning and Planning
Silver et al.
https://arxiv.org/abs/1612.08810
'''
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import logging
import tensorflow as tf
import tensorflow.contrib.layers as layers
import tensorflow.contrib.losses as losses
import tensorflow.contrib.slim as slim
from six.moves import range
from util import predictron_arg_scope
logging.basicConfig()
logger = logging.getLogger('predictron')
logger.setLevel(logging.INFO)
class Predictron(object):
def __init__(self, maze_ims, maze_labels, config):
# self.inputs = tf.placeholder(tf.float32, shape=[None, config.maze_size, config.maze_size, 1])
# self.targets = tf.placeholder(tf.float32, shape=[None, 20])
self.inputs = maze_ims
self.targets = maze_labels
self.maze_size = config.maze_size
self.max_depth = config.max_depth
self.learning_rate = config.learning_rate
self.max_grad_norm = config.max_grad_norm
# Tensor rewards with shape [batch_size, max_depth + 1, maze_size]
self.rewards = None
# Tensor gammas with shape [batch_size, max_depth + 1, maze_size]
self.gammas = None
# Tensor lambdas with shape [batch_size, max_depth, maze_size]
self.lambdas = None
# Tensor values with shape [batch_size, max_depth + 1, maze_size]
self.values = None
# Tensor preturns with shape [batch_size, max_depth + 1, maze_size]
self.preturns = None
# Tensor lambda_preturns with shape [batch_size, maze_size]
self.lambda_preturns = None
self.sess = tf.Session()
self.graph = self.sess.graph
def build(self):
logger.info('Buidling Predictron.')
self.build_model()
self.build_loss()
logger.info('Trainable variables:')
logger.info('*' * 30)
for var in tf.trainable_variables():
logger.info(var.op.name)
logger.info('*' * 30)
def iter_func(self, state):
sc = predictron_arg_scope()
with tf.variable_scope('value'):
value_net = slim.fully_connected(slim.flatten(state), 32, scope='fc0')
value_net = layers.batch_norm(value_net, activation_fn=tf.nn.relu, scope='fc0/preact')
value_net = slim.fully_connected(value_net, self.maze_size, activation_fn=None, scope='fc1')
with slim.arg_scope(sc):
net = slim.conv2d(state, 32, [3, 3], scope='conv1')
net = layers.batch_norm(net, activation_fn=tf.nn.relu, scope='conv1/preact')
net_flatten = slim.flatten(net, scope='conv1/flatten')
with tf.variable_scope('reward'):
reward_net = slim.fully_connected(net_flatten, 32, scope='fc0')
reward_net = layers.batch_norm(reward_net, activation_fn=tf.nn.relu, scope='fc0/preact')
reward_net = slim.fully_connected(reward_net, self.maze_size, activation_fn=None, scope='fc1')
with tf.variable_scope('gamma'):
gamma_net = slim.fully_connected(net_flatten, 32, scope='fc0')
gamma_net = layers.batch_norm(gamma_net, activation_fn=tf.nn.relu, scope='fc0/preact')
gamma_net = slim.fully_connected(gamma_net, self.maze_size, activation_fn=tf.nn.sigmoid, scope='fc1')
with tf.variable_scope('lambda'):
lambda_net = slim.fully_connected(net_flatten, 32, scope='fc0')
lambda_net = layers.batch_norm(lambda_net, activation_fn=tf.nn.relu, scope='fc0/preact')
lambda_net = slim.fully_connected(lambda_net, self.maze_size, activation_fn=tf.nn.sigmoid, scope='fc1')
net = slim.conv2d(net, 32, [3, 3], scope='conv2')
net = layers.batch_norm(net, activation_fn=tf.nn.relu, scope='conv2/preact')
net = slim.conv2d(net, 32, [3, 3], scope='conv3')
net = layers.batch_norm(net, activation_fn=tf.nn.relu, scope='conv3/preact')
return net, reward_net, gamma_net, lambda_net, value_net
def build_model(self):
sc = predictron_arg_scope()
with tf.variable_scope('state'):
with slim.arg_scope(sc):
state = slim.conv2d(self.inputs, 32, [3, 3], scope='conv1')
state = layers.batch_norm(state, activation_fn=tf.nn.relu, scope='conv1/preact')
state = slim.conv2d(state, 32, [3, 3], scope='conv2')
state = layers.batch_norm(state, activation_fn=tf.nn.relu, scope='conv2/preact')
iter_template = tf.make_template('iter', self.iter_func, unique_name_='iter')
rewards_arr = []
gammas_arr = []
lambdas_arr = []
values_arr = []
for k in range(self.max_depth):
state, reward, gamma, lambda_, value = iter_template(state)
rewards_arr.append(reward)
gammas_arr.append(gamma)
lambdas_arr.append(lambda_)
values_arr.append(value)
_, _, _, _, value = iter_template(state)
# K + 1 elements
values_arr.append(value)
bs = tf.shape(self.inputs)[0]
# [batch_size, K * maze_size]
self.rewards = tf.pack(rewards_arr, axis=1)
# [batch_size, K, maze_size]
self.rewards = tf.reshape(self.rewards, [bs, self.max_depth, self.maze_size])
# [batch_size, K + 1, maze_size]
self.rewards = tf.concat_v2(values=[tf.zeros(shape=[bs, 1, self.maze_size], dtype=tf.float32), self.rewards],
axis=1, name='rewards')
# [batch_size, K * maze_size]
self.gammas = tf.pack(gammas_arr, axis=1)
# [batch_size, K, maze_size]
self.gammas = tf.reshape(self.gammas, [bs, self.max_depth, self.maze_size])
# [batch_size, K + 1, maze_size]
self.gammas = tf.concat_v2(values=[tf.ones(shape=[bs, 1, self.maze_size], dtype=tf.float32), self.gammas],
axis=1, name='gammas')
# [batch_size, K * maze_size]
self.lambdas = tf.pack(lambdas_arr, axis=1)
# [batch_size, K, maze_size]
self.lambdas = tf.reshape(self.lambdas, [-1, self.max_depth, self.maze_size])
# [batch_size, (K + 1) * maze_size]
self.values = tf.pack(values_arr, axis=1)
# [batch_size, K + 1, maze_size]
self.values = tf.reshape(self.values, [-1, (self.max_depth + 1), self.maze_size])
self.build_preturns()
self.build_lambda_preturns()
def build_preturns(self):
''' Eqn (2) '''
g_preturns = []
# for k = 0, g_0 = v[0], still fits.
for k in range(self.max_depth, -1, -1):
g_k = self.values[:, k, :]
for kk in range(k, 0, -1):
g_k = self.rewards[:, kk, :] + self.gammas[:, kk, :] * g_k
g_preturns.append(g_k)
# reverse to make 0...K from K...0
g_preturns = g_preturns[::-1]
self.g_preturns = tf.pack(g_preturns, axis=1, name='preturns')
self.g_preturns = tf.reshape(self.g_preturns, [-1, self.max_depth + 1, self.maze_size])
def build_lambda_preturns(self):
''' Eqn (4) '''
g_k = self.values[:, -1, :]
for k in range(self.max_depth - 1, -1, -1):
g_k = (1 - self.lambdas[:, k, :]) * self.values[:, k, :] + \
self.lambdas[:, k, :] * (self.rewards[:, k + 1, :] + self.gammas[:, k + 1, :] * g_k)
self.g_lambda_preturns = g_k
def build_loss(self):
with tf.variable_scope('loss'):
# Loss Eqn (5)
# [batch_size, 1, maze_size]
self.targets_tiled = tf.expand_dims(self.targets, 1)
# [batch_size, K + 1, maze_size]
self.targets_tiled = tf.tile(self.targets_tiled, [1, self.max_depth + 1, 1])
self.loss_preturns = losses.mean_squared_error(self.g_preturns, self.targets_tiled, scope='preturns')
losses.add_loss(self.loss_preturns)
tf.summary.scalar('loss_preturns', self.loss_preturns)
# Loss Eqn (7)
self.loss_lambda_preturns = losses.mean_squared_error(
self.g_lambda_preturns, self.targets, scope='lambda_preturns')
losses.add_loss(self.loss_lambda_preturns)
tf.summary.scalar('loss_lambda_preturns', self.loss_lambda_preturns)
self.total_loss = losses.get_total_loss(name='total_loss')