-
Notifications
You must be signed in to change notification settings - Fork 0
/
funcs.py
450 lines (389 loc) · 14.8 KB
/
funcs.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
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
import torch
import torch.nn.functional as F
from models import *
from collections import OrderedDict
import torch.optim as optim
import torchvision.transforms as transforms
import torchvision
import torch.nn as nn
import os
from copy import deepcopy
import glob
import numpy as np
import pandas as pd
import time
#### DICTIONARIES FOR CONVERTING BETWEEN STRING AND CLASS
string_to_conv = {
'Conv' : Conv,
'DConvA2' : DConvA2,
'DConvA4' : DConvA4,
'DConvA8' : DConvA8,
'DConvA16' : DConvA16,
'DConvG16' : DConvG16,
'DConvG8' : DConvG8,
'DConvG4' : DConvG4,
'DConvG2' : DConvG2,
'DConv' : DConv,
'ConvB2' : ConvB2,
'ConvB4' : ConvB4,
'A2B2' : A2B2,
'A4B2' : A4B2,
'A8B2' : A8B2,
'A16B2' : A16B2,
'G16B2' : G16B2,
'G8B2' : G8B2,
'G4B2' : G4B2,
'G2B2' : G2B2
}
conv_to_string = {
Conv : 'Conv',
DConvA2 : 'DConvA2',
DConvA4 : 'DConvA4',
DConvA8 : 'DConvA8',
DConvA16 : 'DConvA16',
DConvG16 : 'DConvG16',
DConvG8 : 'DConvG8',
DConvG4 : 'DConvG4',
DConvG2 : 'DConvG2',
DConv : 'DConv',
ConvB2 : 'ConvB2',
ConvB4 : 'ConvB4',
A2B2 : 'A2B2',
A4B2 : 'A4B2',
A8B2 : 'A8B2',
A16B2 : 'A16B2',
G16B2 : 'G16B2',
G8B2 : 'G8B2',
G4B2 : 'G4B2',
G2B2 : 'G2B2'
}
####
def distillation(y, teacher_scores, labels, T, alpha):
return F.kl_div(F.log_softmax(y/T, dim=1), F.softmax(teacher_scores/T, dim=1)) * (T*T * 2. * alpha)\
+ F.cross_entropy(y, labels) * (1. - alpha)
def at(x):
return F.normalize(x.pow(2).mean(1).view(x.size(0), -1))
def at_loss(x, y):
return (at(x) - at(y)).pow(2).mean()
def accuracy(output, target, topk=(1,)):
"""Computes the precision@k for the specified values of k"""
maxk = max(topk)
batch_size = target.size(0)
_, pred = output.topk(maxk, 1, True, True)
pred = pred.t()
correct = pred.eq(target.view(1, -1).expand_as(pred))
res = []
for k in topk:
correct_k = correct[:k].view(-1).float().sum(0, keepdim=True)
res.append(correct_k.mul_(100.0 / batch_size))
return res
def expand_model(model, layers=[]):
for layer in model.children():
if len(list(layer.children())) > 0:
expand_model(layer, layers)
else:
layers.append(layer)
return layers
def get_flops(net, x):
layers = expand_model(net, [])
flops = 0
for layer in layers:
if isinstance(layer, nn.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)
ops = layer.in_channels * layer.out_channels * layer.kernel_size[0] * \
layer.kernel_size[1] * out_h * out_w / layer.groups
flops += ops
return flops
def get_no_params(net, verbose=False):
params = net.state_dict()
tot= 0
conv_tot = 0
for p in params:
no = params[p].view(-1).__len__()
if ('num_batches_tracked' not in p) and ('running' not in p) and ('mask' not in p):
tot += no
if 'conv' in p:
conv_tot += no
if verbose:
print('Net has %d conv params' % conv_tot)
print('Net has %d params in total' % tot)
return tot
def get_inference_time(net, testloader, verbose=False):
start_time = time.time()
total_images = 0
total_correct = 0
k = 0
with torch.no_grad():
for data in testloader:
images, labels = data
outputs, _ = net(images)
_, predicted = torch.max(outputs.data, 1)
total_images += labels.size(0)
total_correct += (predicted == labels).sum().item()
break
param_time = round((time.time()-start_time) * 100)
return param_time
def get_imagenet_loaders(imagenet_loc, batch_size=128, workers=12):
num_classes = 1000
traindir = os.path.join(imagenet_loc, 'train')
valdir = os.path.join(imagenet_loc, 'val')
normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
transform_train = transforms.Compose([
transforms.RandomResizedCrop(224),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
normalize,
])
transform_validate = transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
normalize,
])
trainset = torchvision.datasets.ImageFolder(traindir, transform_train)
valset = torchvision.datasets.ImageFolder(valdir, transform_validate)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True,
num_workers=workers,
pin_memory = True)
valloader = torch.utils.data.DataLoader(valset, batch_size=100, shuffle=False,
num_workers=workers,
pin_memory=True)
return trainloader, valloader
def get_cifar_loaders(cifar_loc, batch_size=128, workers=0):
num_classes = 10
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
])
transform_validate = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
])
trainset = torchvision.datasets.CIFAR10(root=cifar_loc,
train=True, download=False, transform=transform_train)
valset = torchvision.datasets.CIFAR10(root=cifar_loc,
train=False, download=False, transform=transform_validate)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True,
num_workers=workers)
valloader = torch.utils.data.DataLoader(valset, batch_size=100, shuffle=False,
num_workers=workers)
return trainloader, valloader
def get_cifar100_loaders(cifar_loc='../cifar100', batch_size=128, workers=0):
num_classes = 100
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.5071, 0.4866, 0.4409), (0.2009, 0.1984, 0.2023)),
])
transform_validate = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.5071, 0.4866, 0.4409), (0.2009, 0.1984, 0.2023)),
])
trainset = torchvision.datasets.CIFAR100(root=cifar_loc,
train=True, download=False, transform=transform_train)
valset = torchvision.datasets.CIFAR100(root=cifar_loc,
train=False, download=False, transform=transform_validate)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True,
num_workers=workers)
valloader = torch.utils.data.DataLoader(valset, batch_size=100, shuffle=False,
num_workers=workers)
return trainloader, valloader
def get_imagenet_loaders(imagenet_loc, batch_size=128, workers=12):
num_classes = 1000
traindir = os.path.join(imagenet_loc, 'train')
valdir = os.path.join(imagenet_loc, 'val')
normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
transform_train = transforms.Compose([
transforms.RandomResizedCrop(224),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
normalize,
])
transform_validate = transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(),
normalize,
])
trainset = torchvision.datasets.ImageFolder(traindir, transform_train)
valset = torchvision.datasets.ImageFolder(valdir, transform_validate)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True,
num_workers=workers,
pin_memory = True)
valloader = torch.utils.data.DataLoader(valset, batch_size=100, shuffle=False,
num_workers=workers,
pin_memory=True)
return trainloader, valloader
def cifar_random_search(save_file, data_loc):
df = []
train, val = get_cifar_loaders(data_loc)
save_counter = 1
while(True):
random_blocks = np.random.choice(list(conv_to_string.keys()), 18)
net = WideResNet(40,2,Conv,BasicBlock,convs=random_blocks)
i_time = get_inference_time(net, val, verbose=False)
df.append([random_blocks, i_time])
if save_counter == 10000:
print('saving df ', save_counter)
df1 = df
df1 = pd.DataFrame(df1, columns=['convs','i_time'])
print(len(df1))
df1 = df1[~df1['convs'].apply(tuple).duplicated()]
print(len(df1))
df1.to_csv('archs/master_time' + str(save_counter) + '.csv' )
df1 = []
if save_counter == 20000:
print('saving df ', save_counter)
df2 = df
df2 = pd.DataFrame(df2, columns=['convs','i_time'])
print(len(df2))
df2 = df2[~df2['convs'].apply(tuple).duplicated()]
print(len(df2))
df2.to_csv('archs/master_time' + str(save_counter) + '.csv' )
df2 = []
if save_counter == 30000:
print('saving df ', save_counter)
df = pd.DataFrame(df, columns=['convs','i_time'])
print(len(df))
df = df[~df['convs'].apply(tuple).duplicated()]
print(len(df))
df.to_csv('archs/master_time' + str(save_counter) + '.csv' )
df = []
break;
save_counter = save_counter + 1
def imagenet_random_search():
df = []
save_counter = 0
while(True):
random_blocks = np.random.choice(list(conv_to_string.keys()), 16)
net = ResNet(Conv, Block, [3, 4, 6, 3], convs=random_blocks)
params = get_no_params(net, verbose=False)
df.append([random_blocks, params])
if save_counter % 10000 == 0:
print('saving df ', save_counter)
df = pd.DataFrame(df, columns=['convs','params'])
df.to_csv(str('imagenet_models' + str(save_counter) + '.csv'))
df = []
save_counter = save_counter + 1
class AverageMeter(object):
"""Computes and stores the average and current value"""
def __init__(self):
self.reset()
def reset(self):
self.val = 0
self.avg = 0
self.sum = 0
self.count = 0
def update(self, val, n=1):
self.val = val
self.sum += val * n
self.count += n
self.avg = self.sum / self.count
class Pruner:
def __init__(self, module_name='MaskBlock'):
self.module_name = module_name
self.masks = []
self.prune_history= []
def go_fish(self, model):
self._get_fisher(model)
tot_loss = self.fisher.div(1) + 1e6 * (1 - self.masks) #giga
return tot_loss
def _get_fisher(self, model):
masks=[]
fisher=[]
flops=[]
self._update_flops(model)
for m in model.modules():
if m._get_name() == 'MaskBlock' or m._get_name() == 'MaskBottleBlock':
masks.append(m.mask.detach())
fisher.append(m.running_fisher.detach())
flops.append(m.flops_vector)
m.reset_fisher()
self.masks = self.concat(masks)
self.fisher = self.concat(fisher)
self.flops = self.concat(flops)
def _get_masks(self, model):
masks=[]
for m in model.modules():
if m._get_name() == 'MaskBlock' or m._get_name() == 'MaskBottleBlock':
masks.append(m.mask.detach())
self.masks = self.concat(masks)
def _update_flops(self, model):
for m in model.modules():
if m._get_name() == 'MaskBlock' or m._get_name() == 'MaskBottleBlock':
m.cost()
@staticmethod
def concat(input):
return torch.cat([item for item in input])
def get_convs(model):
blocks = []
for m in model.modules():
if 'Mask' in m._get_name():
if isinstance(m.conv1, Conv):
blocks.append('S')
else:
blocks.append('G')
return blocks
def concat_archs():
archs = glob.glob("arch/*.csv")
# get all of our archs in a row
master = pd.concat([pd.read_csv(arch) for arch in archs])
master.to_csv('arch/master_arch.csv')
def one_shot_fisher(net, trainloader, val, n_steps=1):
inference_time = get_inference_time(net, val, verbose=False)
criterion = nn.CrossEntropyLoss()
# switch to train mode
net.train()
optimizer = optim.SGD(net.parameters(), lr=0.1, momentum=0.9, weight_decay=0.0005)
dataiter = iter(trainloader)
pruner = Pruner()
data = torch.rand(net.input_spatial_dims)
#if cuda:
# data = data.cuda()
net(data)
pruner._get_masks(net)
NO_STEPS = n_steps # single minibatch
for i in range(0, NO_STEPS):
try:
input, target = dataiter.next()
except StopIteration:
dataiter = iter(trainloader)
input, target = dataiter.next()
# if cuda:
# input, target = input.cuda(), target.cuda()
# compute output
output, act = net(input)
loss = criterion(output, target)
# compute gradient and do SGD step
optimizer.zero_grad()
loss.backward()
optimizer.step()
fisher_inf = pruner.go_fish(net)
features = fisher_inf.size()
fish_market = dict()
running_index = 0
block_count = 0
data = torch.rand(1,16,32,32)
#if cuda:
# data = data.cuda()
# partition fisher_inf into blocks blocks blocks
for m in net.modules():
if m._get_name() == 'MaskBlock' or m._get_name() == 'MaskBottleBlock':
mask_indices = range(running_index, running_index + len(m.mask))
fishies = [fisher_inf[j] for j in mask_indices]
running_index += len(m.mask)
fish = sum(fishies)
data = m(data)
flops = m.flops * data.size()[2]
fish = fish / flops
fish_market[block_count] = fish
block_count +=1
return inference_time, fish_market