-
Notifications
You must be signed in to change notification settings - Fork 4
/
input_data.py
executable file
·192 lines (156 loc) · 8.46 KB
/
input_data.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
from __future__ import print_function
import numpy as np
import os
import SharedArray as sa
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
class InputData:
def __init__(self, model, batch_size=64):
self.model = model # to get endpoint
self.batch_size = batch_size
self.z = dict()
self.x = dict()
def add_data(self, path_new, key='train'):
self.x[key] = np.load(path_new)
print('data size:', self.x[key].shape)
def add_data_sa(self, path_new, key='train'):
self.x[key] = sa.attach(path_new)
print('data size:', self.x[key].shape)
def add_data_np(self, data, key='train'):
self.x[key] = data
print('data size:', self.x[key].shape)
def get_batch_num(self, key='train'):
return len(self.x[key]) // self.batch_size
def get_batch(self, idx=0, data_size=None, key='train'):
data_size = self.batch_size if data_size is None else data_size
st = self.batch_size*idx
##condition on chord/mel in testing part
##return [self.x[key][st+8]* 2. - 1.]*data_size ##
return self.x[key][st:st+data_size] * 2. - 1.
def get_rand_smaples(self, sample_size=64, key='train'):
random_idx = np.random.choice(len(self.x[key]), sample_size, replace=False)
return self.x[key][random_idx]*2. - 1.
def gen_feed_dict(self, idx=0, data_size=None, key='train', z=None):
batch_size = self.batch_size if data_size is None else data_size
feed_dict = self.gen_z_dict(data_size=data_size, z=z)
if key is not None:
x = self.get_batch(idx, data_size, key)
feed_dict[self.model.x] = x
return feed_dict
#######################################################################################################################
# Image
#######################################################################################################################
class InputDataMNIST(InputData):
dataset_dir = 'dataset/mnist/original'
def __init__(self, model, batch_size=64):
self.model = model # to get endpoint
self.batch_size = batch_size
self.x = dict()
mnist = input_data.read_data_sets(self.dataset_dir, one_hot = True)
self.add_data_np(mnist.train.images.reshape((-1,28,28,1)), 'train')
self.add_data_np(mnist.test.images.reshape((-1,28,28,1)), 'test')
def gen_feed_dict(self, idx=0, data_size=None, key='train'):
batch_size = self.batch_size if data_size is None else data_size
z = np.random.uniform(-1., 1., size=(self.batch_size, self.model.z_dim)).astype(np.float32)
x = self.get_batch(idx, data_size, key)
feed_dict = {self.model.z: z, self.model.x: x}
return feed_dict
#######################################################################################################################
# Music
#######################################################################################################################
# Nowbar
class InputDataNowBarHybrid(InputData):
def gen_z_dict(self, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['inter']= np.random.normal(0, 0.1, [batch_size, self.model.z_inter_dim]).astype(np.float32)
self.z['intra'] = np.random.normal(0, 0.1, [batch_size, self.model.z_intra_dim, self.model.track_dim]).astype(np.float32)
z_dict = {self.model.z_intra: self.z['intra'], self.model.z_inter:self.z['inter']}
return z_dict
class InputDataNowBarJamming(InputData):
def gen_z_dict(self, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['intra'] = np.random.normal(0, 0.1, [batch_size, self.model.z_intra_dim, self.model.track_dim]).astype(np.float32)
z_dict = {self.model.z_intra: self.z['intra']}
return z_dict
class InputDataNowBarComposer(InputData):
def gen_z_dict(self, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['inter'] = np.random.normal(0, 0.1, [batch_size, self.model.z_inter_dim]).astype(np.float32)
z_dict = {self.model.z_inter: self.z['inter']}
return z_dict
# temporal
class InputDataTemporalHybrid(InputData):
def gen_z_dict(self, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['z_intra_v'] = np.random.normal(0, 0.1, [batch_size, self.model.z_intra_dim, self.model.track_dim]).astype(np.float32)
self.z['z_intra_i'] = np.random.normal(0, 0.1, [batch_size, self.model.z_intra_dim, self.model.track_dim]).astype(np.float32)
self.z['z_inter_v'] = np.random.normal(0, 0.1, [batch_size, self.model.z_inter_dim]).astype(np.float32)
self.z['z_inter_i'] = np.random.normal(0, 0.1, [batch_size, self.model.z_inter_dim]).astype(np.float32)
feed_dict = {self.model.z_intra_v: self.z['z_intra_v'], self.model.z_intra_i: self.z['z_intra_i'],
self.model.z_inter_v: self.z['z_inter_v'], self.model.z_inter_i: self.z['z_inter_i']}
return feed_dict
class InputDataTemporalJamming(InputData):
def gen_z_dict(self, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['z_intra_v'] = np.random.normal(0, 0.1, [batch_size, self.model.z_intra_dim, self.model.track_dim]).astype(np.float32)
self.z['z_intra_i'] = np.random.normal(0, 0.1, [batch_size, self.model.z_intra_dim, self.model.track_dim]).astype(np.float32)
feed_dict = {self.model.z_intra_v: self.z['z_intra_v'], self.model.z_intra_i: self.z['z_intra_i']}
return feed_dict
class InputDataTemporalComposer(InputData):
def gen_z_dict(self, idx=0, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['z_inter_v'] = np.random.normal(0, 0.1, [batch_size, self.model.z_inter_dim]).astype(np.float32)
self.z['z_inter_i'] = np.random.normal(0, 0.1, [batch_size, self.model.z_inter_dim]).astype(np.float32)
feed_dict = {self.model.z_inter_v: self.z['z_inter_v'], self.model.z_inter_i: self.z['z_inter_i']}
return feed_dict
class InputDataRNNComposer(InputData):
def gen_z_dict(self, idx=0, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['z_inter'] = np.random.normal(0, 0.1, [batch_size, self.model.output_bar, self.model.z_inter_dim]).astype(np.float32)
feed_dict = {self.model.z_inter: self.z['z_inter']}
return feed_dict
class InputDataRNNHybrid(InputData):
def gen_z_dict(self, idx=0, data_size=None, z=None):
batch_size = self.batch_size if data_size is None else data_size
if z is not None:
self.z = z
else:
self.z = dict()
self.z['z_inter_v'] = np.random.normal(0, 0.1, [batch_size, self.model.output_bar,
self.model.z_inter_dim]).astype(np.float32)
self.z['z_inter_i'] = np.random.normal(0, 0.1, [batch_size, self.model.z_inter_dim]).astype(np.float32)
self.z['z_intra_v'] = np.random.normal(0, 0.1, [batch_size, self.model.output_bar, self.model.z_intra_dim,
self.model.track_dim]).astype(np.float32)
self.z['z_intra_i'] = np.random.normal(0, 0.1, [batch_size, self.model.z_intra_dim,
self.model.track_dim]).astype(np.float32)
feed_dict = {self.model.z_inter_v: self.z['z_inter_v'], self.model.z_inter_i: self.z['z_inter_i'],
self.model.z_intra_v: self.z['z_intra_v'], self.model.z_intra_i: self.z['z_intra_i']}
return feed_dict