-
Notifications
You must be signed in to change notification settings - Fork 0
/
Tensorflow Basic.py
155 lines (96 loc) · 4.27 KB
/
Tensorflow Basic.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
import tensorflow as tf
import numpy as np
print(tf.__version__)
tf.compat.v1.disable_eager_execution()
tf.compat.v1.reset_default_graph()
a = tf.compat.v1.placeholder(np.float32, (2, 2))
b = tf.compat.v1.Variable(tf.ones((2, 2)))
c = a @ b
print(c)
s = tf.compat.v1.InteractiveSession()
s.run(tf.compat.v1.global_variables_initializer())
s.run(c, feed_dict={a: np.ones((2, 2))})
s.close()
tf.compat.v1.reset_default_graph()
x = tf.compat.v1.get_variable("x", shape=(), dtype=tf.float32, trainable=True)
f = x ** 2
optimizer = tf.compat.v1.train.GradientDescentOptimizer(0.1)
step = optimizer.minimize(f, var_list=[x])
tf.compat.v1.trainable_variables()
with tf.compat.v1.Session() as s: # in this way session will be closed automatically
s.run(tf.compat.v1.global_variables_initializer())
for i in range(10):
_, curr_x, curr_f = s.run([step, x, f])
print(curr_x, curr_f)
# ### Simple optimization (with tf.Print)
tf.compat.v1.reset_default_graph()
x = tf.compat.v1.get_variable("x", shape=(), dtype=tf.float32)
f = x ** 2
f = tf.compat.v1.Print(f, [x, f], "x, f:")
optimizer = tf.compat.v1.train.GradientDescentOptimizer(0.1)
step = optimizer.minimize(f)
with tf.compat.v1.Session() as s:
s.run(tf.compat.v1.global_variables_initializer())
for i in range(10):
s.run([step, f])
# Prints to jupyter server stdout (not available in Coursera Notebooks):
# 2018-07-21 18:01:27.308270: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-1.0670249][1.1385423]
# 2018-07-21 18:01:27.308809: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.85361993][0.72866696]
# 2018-07-21 18:01:27.309116: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.68289596][0.46634689]
# 2018-07-21 18:01:27.309388: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.54631674][0.29846197]
# 2018-07-21 18:01:27.309678: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.43705338][0.19101566]
# 2018-07-21 18:01:27.309889: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.34964269][0.12225001]
# 2018-07-21 18:01:27.310213: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.27971417][0.078240015]
# 2018-07-21 18:01:27.310475: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.22377133][0.050073609]
# 2018-07-21 18:01:27.310751: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.17901707][0.032047112]
# 2018-07-21 18:01:27.310963: I tensorflow/core/kernels/logging_ops.cc:79] x, f:[-0.14321366][0.020510152]
# ### Simple optimization (with TensorBoard logging)
tf.compat.v1.reset_default_graph()
x = tf.compat.v1.get_variable("x", shape=(), dtype=tf.float32)
f = x ** 2
optimizer = tf.compat.v1.train.GradientDescentOptimizer(0.1)
step = optimizer.minimize(f)
tf.summary.scalar('curr_x', x)
tf.summary.scalar('curr_f', f)
s = tf.compat.v1.InteractiveSession()
summary_writer = tf.compat.v1.summary.FileWriter("logs/1", s.graph)
s.run(tf.compat.v1.global_variables_initializer())
summaries = tf.compat.v1.summary.merge_all()
for i in range(1, 10):
_, curr_summaries = s.run([step, summaries])
summary_writer.add_summary(curr_summaries, i)
summary_writer.flush()
# Run `tensorboard --logdir=./logs` in bash
import os
os.system("tensorboard --logdir=./logs --host 0.0.0.0 --port 6006 &")
s.close()
# ### Training a linear model
# generate model data
N = 1000
D = 3
x = np.random.random((N, D))
w = np.random.random((D, 1))
y = x @ w + np.random.randn(N, 1) * 0.20
print(x.shape, y.shape)
print(w.T)
# In[22]:
tf.compat.v1.reset_default_graph()
features = tf.compat.v1.placeholder(tf.float32, shape=(None, D))
target = tf.compat.v1.placeholder(tf.float32, shape=(None, 1))
weights = tf.compat.v1.get_variable("weights", shape=(D, 1), dtype=tf.float32)
predictions = features @ weights
loss = tf.reduce_mean((target - predictions) ** 2)
print(target.shape, predictions.shape, loss.shape)
optimizer = tf.compat.v1.train.GradientDescentOptimizer(0.1)
step = optimizer.minimize(loss)
with tf.compat.v1.Session() as s:
s.run(tf.compat.v1.global_variables_initializer())
for i in range(300):
_, curr_loss, curr_weights = s.run([step, loss, weights],
feed_dict={features: x, target: y})
if i % 50 == 0:
print(curr_loss)
# found weights
print(curr_weights.T)
# true weights
print(w.T)