-
Notifications
You must be signed in to change notification settings - Fork 12
/
utils.py
234 lines (195 loc) · 6.75 KB
/
utils.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
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
'''Some helper functions for PyTorch, including:
- get_mean_and_std: calculate the mean and std value of dataset.
- msr_init: net parameter initialization.
- progress_bar: progress bar mimic xlua.progress.
'''
import os
import sys
import time
import math
import operator
from functools import reduce
import torch.nn as nn
import torch
import torch.nn.init as init
def get_mean_and_std(dataset):
'''Compute the mean and std value of dataset.'''
dataloader = torch.utils.data.DataLoader(dataset, batch_size=1, shuffle=True, num_workers=2)
mean = torch.zeros(3)
std = torch.zeros(3)
print('==> Computing mean and std..')
for inputs, targets in dataloader:
for i in range(3):
mean[i] += inputs[:,i,:,:].mean()
std[i] += inputs[:,i,:,:].std()
mean.div_(len(dataset))
std.div_(len(dataset))
return mean, std
def init_params(net):
'''Init layer parameters.'''
for m in net.modules():
if isinstance(m, nn.Conv2d):
init.kaiming_normal(m.weight, mode='fan_out')
if m.bias:
init.constant(m.bias, 0)
elif isinstance(m, nn.BatchNorm2d):
init.constant(m.weight, 1)
init.constant(m.bias, 0)
elif isinstance(m, nn.Linear):
init.normal(m.weight, std=1e-3)
if m.bias:
init.constant(m.bias, 0)
_, term_width = os.popen('stty size', 'r').read().split()
term_width = int(term_width)
TOTAL_BAR_LENGTH = 65.
last_time = time.time()
begin_time = last_time
def progress_bar(current, total, msg=None):
global last_time, begin_time
if current == 0:
begin_time = time.time() # Reset for new bar.
cur_len = int(TOTAL_BAR_LENGTH*current/total)
rest_len = int(TOTAL_BAR_LENGTH - cur_len) - 1
sys.stdout.write(' [')
for i in range(cur_len):
sys.stdout.write('=')
sys.stdout.write('>')
for i in range(rest_len):
sys.stdout.write('.')
sys.stdout.write(']')
cur_time = time.time()
step_time = cur_time - last_time
last_time = cur_time
tot_time = cur_time - begin_time
L = []
L.append(' Step: %s' % format_time(step_time))
L.append(' | Tot: %s' % format_time(tot_time))
if msg:
L.append(' | ' + msg)
msg = ''.join(L)
sys.stdout.write(msg)
for i in range(term_width-int(TOTAL_BAR_LENGTH)-len(msg)-3):
sys.stdout.write(' ')
# Go back to the center of the bar.
for i in range(term_width-int(TOTAL_BAR_LENGTH/2)+2):
sys.stdout.write('\b')
sys.stdout.write(' %d/%d ' % (current+1, total))
if current < total-1:
sys.stdout.write('\r')
else:
sys.stdout.write('\n')
sys.stdout.flush()
def format_time(seconds):
days = int(seconds / 3600/24)
seconds = seconds - days*3600*24
hours = int(seconds / 3600)
seconds = seconds - hours*3600
minutes = int(seconds / 60)
seconds = seconds - minutes*60
secondsf = int(seconds)
seconds = seconds - secondsf
millis = int(seconds*1000)
f = ''
i = 1
if days > 0:
f += str(days) + 'D'
i += 1
if hours > 0 and i <= 2:
f += str(hours) + 'h'
i += 1
if minutes > 0 and i <= 2:
f += str(minutes) + 'm'
i += 1
if secondsf > 0 and i <= 2:
f += str(secondsf) + 's'
i += 1
if millis > 0 and i <= 2:
f += str(millis) + 'ms'
i += 1
if f == '':
f = '0ms'
return f
def cal_param_size(model):
return sum([reduce(operator.mul, i.size(), 1) for i in model.parameters()])
count_ops = 0
def measure_layer(layer, x, multi_add=1):
type_name = str(layer)[:str(layer).find('(')].strip()
# print(type_name)
if type_name in ['Conv2d']:
out_h = int((x.size()[2] + 2 * layer.padding[0] - layer.kernel_size[0]) //
layer.stride[0] + 1)
out_w = int((x.size()[3] + 2 * layer.padding[1] - layer.kernel_size[1]) //
layer.stride[1] + 1)
delta_ops = layer.in_channels * layer.out_channels * layer.kernel_size[0] * \
layer.kernel_size[1] * out_h * out_w // layer.groups * multi_add
### ops_nonlinearity
elif type_name in ['ReLU']:
delta_ops = x.numel()// x.size(0)
### ops_pooling
elif type_name in ['AvgPool2d']:
in_w = x.size()[2]
kernel_ops = layer.kernel_size * layer.kernel_size
out_w = int((in_w + 2 * layer.padding - layer.kernel_size) // layer.stride + 1)
out_h = int((in_w + 2 * layer.padding - layer.kernel_size) // layer.stride + 1)
delta_ops = x.size()[1] * out_w * out_h * kernel_ops
elif type_name in ['AdaptiveAvgPool2d']:
delta_ops = x.numel() // x.size(0)
### ops_linear
elif type_name in ['Linear']:
weight_ops = layer.weight.numel() * multi_add
bias_ops = layer.bias.numel()
delta_ops = weight_ops + bias_ops
elif type_name in ['BatchNorm2d']:
normalize_ops = x.numel() // x.size(0)
scale_shift = normalize_ops
delta_ops = normalize_ops + scale_shift
### ops_nothing
elif type_name.startswith('Drop') or type_name.startswith('MaxPool'):
delta_ops = 0
### unknown layer type
else:
#print('unknown layer type: %s !' % type_name)
return
global count_ops
count_ops += delta_ops
return
def is_leaf(module):
return sum(1 for x in module.children()) == 0
# 判断是否为需要计算flops的结点模块
def should_measure(module):
if str(module).startswith('Sequential'):
return False
if is_leaf(module):
return True
return False
def cal_multi_adds(model, shape=(2,3,32,32)):
global count_ops
data = torch.zeros(shape)
# 将计算flops的操作集成到forward函数
def new_forward(m):
def lambda_forward(x):
measure_layer(m, x)
return m.old_forward(x)
return lambda_forward
def modify_forward(model):
for child in model.children():
if should_measure(child):
# 新增一个old_forward属性保存默认的forward函数
# 便于计算flops结束后forward函数的恢复
child.old_forward = child.forward
child.forward = new_forward(child)
else:
modify_forward(child)
def restore_forward(model):
for child in model.children():
# 对修改后的forward函数进行恢复
if is_leaf(child) and hasattr(child, 'old_forward'):
child.forward = child.old_forward
child.old_forward = None
else:
restore_forward(child)
modify_forward(model)
# forward过程中对全局的变量count_ops进行更新
model.forward(data)
restore_forward(model)
return count_ops