-
Notifications
You must be signed in to change notification settings - Fork 0
/
part_3.py
171 lines (142 loc) · 5.52 KB
/
part_3.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
import keras
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential, Model
from keras.layers import Dense, Activation, Input, Lambda, Conv2D, MaxPooling2D, Flatten, Dropout
from keras.layers.normalization import BatchNormalization
from keras import backend as K
from keras.engine.topology import Layer
import numpy as np
from keras.datasets import cifar10
from keras.utils import np_utils
import tensorly.decomposition as td
import pickle
def get_model():
batch_norm_alpha=0.9
batch_norm_eps=1e-4
model=Sequential()
model.add(Conv2D(filters=64, kernel_size=3, strides=(1, 1), padding='valid',input_shape=[32,32,3]))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
model.add(Conv2D(filters=64, kernel_size=3, strides=(1, 1), padding='valid'))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
model.add(MaxPooling2D(pool_size=(2, 2),strides=(2,2)))
model.add(Conv2D(filters=128, kernel_size=3, strides=(1, 1), padding='valid'))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
model.add(Conv2D(filters=128, kernel_size=3, strides=(1, 1), padding='valid'))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
model.add(MaxPooling2D(pool_size=(2, 2),strides=(2,2)))
model.add(Conv2D(filters=256, kernel_size=3, strides=(1, 1), padding='valid'))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
model.add(Conv2D(filters=256, kernel_size=3, strides=(1, 1), padding='valid'))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
#model.add(MaxPooling2D(pool_size=(2, 2),strides=(2,2)))
model.add(Flatten())
model.add(Dense(512))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
model.add(Dense(512))
model.add(Activation('relu'))
model.add(BatchNormalization(axis=-1, momentum=batch_norm_alpha, epsilon=batch_norm_eps))
model.add(Dense(10))
model.add(Activation('softmax'))
return model
def decompose_tensor(X, R3, R4):
core, factors = td.partial_tucker(X.numpy(), modes = [2,3], ranks = [R3, R4])
return core, factors
def get_compressed_model(model, k = 1):
#iterate over the layers of the model:
ret_model = Sequential()
for i, layer in enumerate(model.layers):
if(isinstance(layer, Conv2D)) and i > 3:
X = layer.weights[0]
n_out = X.shape[-1]
b = layer.weights[1].numpy()
print(X.shape)
r3 = (X.shape[2] * k) // 8
r4 = (X.shape[3] * k) // 8
print(r3, r4)
# I, core, O = decompose_tensor(X, r3, r4)
core, factors = decompose_tensor(X, r3, r4)
I = factors[0]
I = np.expand_dims(I, axis = 0)
I = np.expand_dims(I, axis = 0)
O = factors[1].T
O = np.expand_dims(O, axis = 0)
O = np.expand_dims(O, axis = 0)
I_layer = Conv2D(filters=r3, kernel_size=1, strides=(1, 1), padding='valid',use_bias = False)
core_layer = Conv2D(filters=r4, kernel_size=3, strides=(1, 1), padding='valid', use_bias = False)
O_layer = Conv2D(filters=n_out, kernel_size=1, strides=(1, 1), padding='valid')
ret_model.add(I_layer)
ret_model.add(core_layer)
ret_model.add(O_layer)
I_layer.set_weights([I])
core_layer.set_weights([core])
O_layer.set_weights([O,b])
else:
ret_model.add(layer)
return ret_model
def expt_full_model(model):
losses = []
accs = []
opt = keras.optimizers.Adam(lr=0.001,decay=1e-6)
for k in range(1,8):
comp_model = get_compressed_model(model,k)
comp_model.compile(loss='categorical_crossentropy', optimizer=opt, metrics = ['accuracy'])
l, a = comp_model.evaluate(X_test, y = Y_test)
losses.append(l)
accs.append(a)
print(losses)
print(accs)
return losses, accs
def expt_reconstruction(model, rank_idx):
l2err_k = []
idx = 0
for i, layer in enumerate(model.layers):
if isinstance(layer, Conv2D) and i > 0:
idx = i
break
X = model.layers[idx].weights[0]
for k in range(1,65):
r3 = X.shape[2]
r4 = X.shape[3]
if rank_idx == 2:
r3 = k
elif rank_idx == 3:
r4 = k
core, factors = decompose_tensor(X, r3, r4)
I = factors[0]
O = factors[1].T
W_hat = np.transpose(np.dot(np.dot(I, core), O), (1,2,0,3))
print(r3, r4)
l2err_k.append((np.sum((W_hat - X)**2)))
print(l2err_k)
return np.array(l2err_k)
(X_train, y_train), (X_test, y_test) = cifar10.load_data()
X_train=X_train.astype(np.float32)
X_test=X_test.astype(np.float32)
Y_train = np_utils.to_categorical(y_train, 10)
Y_test = np_utils.to_categorical(y_test, 10)
X_train /= 255
X_test /= 255
X_train=2*X_train-1
X_test=2*X_test-1
print('X_train shape:', X_train.shape)
print(X_train.shape[0], 'train samples')
print(X_test.shape[0], 'test samples')
batch_size=100
lr=0.001
model=get_model()
opt = keras.optimizers.Adam(lr=0.001,decay=1e-6)
weights_path='pretrained_cifar10.h5'
model.load_weights(weights_path)
model.compile(loss='categorical_crossentropy', optimizer=opt, metrics = ['accuracy'])
model.evaluate(X_test, y = Y_test)
# with open('reconstr_err_r2fixed.pkl', 'wb+') as f:
# pickle.dump(expt_reconstruction(model, 2), f)
with open('reconstr_err_r2fixed.pkl', 'wb+') as f:
pickle.dump(expt_full_model(model), f)