-
Notifications
You must be signed in to change notification settings - Fork 0
/
train_search.py
275 lines (208 loc) · 10.1 KB
/
train_search.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
import os,sys,time, glob
import numpy as np
import torch
import utils, prune
import logging
import argparse
import torch.nn as nn
from torch import optim
import torchvision.datasets as dset
import torch.backends.cudnn as cudnn
from model_search import Network
from arch import Arch
parser = argparse.ArgumentParser("cifar")
parser.add_argument('--data', type=str, default='../data', help='location of the data corpus')
parser.add_argument('--batchsz', type=int, default=64, help='batch size')
parser.add_argument('--snipbatchsz', type=int, default=16, help='snip batch size')
parser.add_argument('--lr', type=float, default=0.025, help='init learning rate')
parser.add_argument('--lr_min', type=float, default=0.001, help='min learning rate')
parser.add_argument('--momentum', type=float, default=0.9, help='momentum')
parser.add_argument('--wd', type=float, default=3e-4, help='weight decay')
parser.add_argument('--report_freq', type=float, default=50, help='report frequency')
parser.add_argument('--gpu', type=int, default=0, help='gpu device id')
parser.add_argument('--epochs', type=int, default=50, help='num of training epochs')
parser.add_argument('--init_ch', type=int, default=16, help='num of init channels')
parser.add_argument('--layers', type=int, default=8, help='total number of layers')
parser.add_argument('--model_path', type=str, default='saved_models', help='path to save the model')
parser.add_argument('--cutout', action='store_true', default=False, help='use cutout')
parser.add_argument('--cutout_len', type=int, default=16, help='cutout length')
parser.add_argument('--drop_path_prob', type=float, default=0.3, help='drop path probability')
parser.add_argument('--exp_path', type=str, default='search', help='experiment name')
parser.add_argument('--seed', type=int, default=2, help='random seed')
parser.add_argument('--grad_clip', type=float, default=5, help='gradient clipping range')
parser.add_argument('--train_portion', type=float, default=0.5, help='portion of training/val splitting')
parser.add_argument('--unrolled', action='store_true', default=False, help='use one-step unrolled validation loss')
parser.add_argument('--arch_lr', type=float, default=3e-4, help='learning rate for arch encoding')
parser.add_argument('--arch_wd', type=float, default=1e-3, help='weight decay for arch encoding')
args = parser.parse_args()
args.exp_path += str(args.gpu)
utils.create_exp_dir(args.exp_path, scripts_to_save=glob.glob('*.py'))
log_format = '%(asctime)s %(message)s'
logging.basicConfig(stream=sys.stdout, level=logging.INFO,
format=log_format, datefmt='%m/%d %I:%M:%S %p')
fh = logging.FileHandler(os.path.join(args.exp_path, 'log.txt'))
fh.setFormatter(logging.Formatter(log_format))
logging.getLogger().addHandler(fh)
os.environ['CUDA_VISIBLE_DEVICES'] = str(args.gpu)
device = torch.device('cuda:0')
def main():
np.random.seed(args.seed)
cudnn.benchmark = True
cudnn.enabled = True
torch.manual_seed(args.seed)
# ================================================
# total, used = os.popen(
# 'nvidia-smi --query-gpu=memory.total,memory.used --format=csv,nounits,noheader'
# ).read().split('\n')[args.gpu].split(',')
# total = int(total)
# used = int(used)
#
# print('Total GPU mem:', total, 'used:', used)
# try:
# block_mem = 0.85 * (total - used)
# print(block_mem)
# x = torch.empty((256, 1024, int(block_mem))).cuda()
# del x
# except RuntimeError as err:
# print(err)
# block_mem = 0.8 * (total - used)
# print(block_mem)
# x = torch.empty((256, 1024, int(block_mem))).cuda()
# del x
#
#
# print('reuse mem now ...')
# ================================================
args.unrolled = True #when True optimize step is on alfa and w
# if False optimization is only on w, ordinary backprop, after pruning
logging.info('GPU device = %d' % args.gpu)
logging.info("args = %s", args)
criterion = nn.CrossEntropyLoss().to(device)
model = Network(args.init_ch, 10, args.layers, criterion)#.to(device)
logging.info("Total param size = %f MB", utils.count_parameters_in_MB(model))
# this is the optimizer to optimize
optimizer = optim.SGD(model.parameters(), args.lr, momentum=args.momentum, weight_decay=args.wd)
train_transform, valid_transform = utils._data_transforms_cifar10(args)
train_data = dset.CIFAR10(root=args.data, train=True, download=True, transform=train_transform)
num_train = len(train_data) # 50000
indices = list(range(num_train))
split = int(np.floor(args.train_portion * num_train)) # 25000
train_queue = torch.utils.data.DataLoader(
train_data, batch_size=args.batchsz,
sampler=torch.utils.data.sampler.SubsetRandomSampler(indices[:split]),
pin_memory=True, num_workers=2)
valid_queue = torch.utils.data.DataLoader(
train_data, batch_size=args.batchsz,
sampler=torch.utils.data.sampler.SubsetRandomSampler(indices[split:]),
pin_memory=True, num_workers=2)
scheduler = optim.lr_scheduler.CosineAnnealingLR(
optimizer, float(args.epochs), eta_min=args.lr_min)
#create similar queues for snip function
snip_train_queue = torch.utils.data.DataLoader(
train_data, batch_size=args.snipbatchsz,
sampler=torch.utils.data.sampler.SubsetRandomSampler(indices[:split]),
pin_memory=True, num_workers=2) #for runing on a pc should be one
snip_valid_queue = torch.utils.data.DataLoader(
train_data, batch_size=args.snipbatchsz,
sampler=torch.utils.data.sampler.SubsetRandomSampler(indices[split:]),
pin_memory=True, num_workers=2) #for runing on a pc should be one
#we don't need the validation set queue but I don't want to break anything
#TODO remove it later
snip_scheduler = optim.lr_scheduler.CosineAnnealingLR(
optimizer, float(args.epochs), eta_min=args.lr_min)
#I don't know if we need scheduler like this, maybe set all the fancy settings to zero?
arch = Arch(model, args)
# TODO: how to call minibatches on snip ? do we need one minibatch or more? if so how many?
for (inputs_snip_batch, labels_snip_batch) in enumerate(train_queue):
inputs_snip_batch, labels_snip_batch = inputs_snip_batch.to(device), labels_snip_batch.cuda(non_blocking=True)
model = prune.snip(model, inputs_snip_batch, labels_snip_batch)
#TODO learning: what is epoch?
for epoch in range(args.epochs):
scheduler.step()
lr = scheduler.get_lr()[0]
logging.info('\nEpoch: %d lr: %e', epoch, lr)
genotype = model.genotype()
logging.info('Genotype: %s', genotype)
# print(F.softmax(model.alphas_normal, dim=-1))
# print(F.softmax(model.alphas_reduce, dim=-1))
# training
train_acc, train_obj = train(train_queue, valid_queue, model, arch, criterion, optimizer, lr)
logging.info('train acc: %f', train_acc)
# validation
valid_acc, valid_obj = infer(valid_queue, model, criterion)
logging.info('valid acc: %f', valid_acc)
utils.save(model, os.path.join(args.exp_path, 'search.pt'))
def train(train_queue, valid_queue, model, arch, criterion, optimizer, lr):
"""
:param train_queue: train loader
:param valid_queue: validate loader
:param model: network
:param arch: Arch class
:param criterion:
:param optimizer:
:param lr:
:return:
"""
losses = utils.AverageMeter()
top1 = utils.AverageMeter()
top5 = utils.AverageMeter()
valid_iter = iter(valid_queue)
# step is a counter which starts from 0 to the number of tuples in train_quit other than that has no relation to train_queue
# (x, target) are the tuple
#TODO: is (x, target) a single data point of a minibatch or is it all the data points of a minibatch
for step, (x, target) in enumerate(train_queue):
batchsz = x.size(0)
model.train()
# [b, 3, 32, 32], [40]
x, target = x.to(device), target.cuda(non_blocking=True)
x_search, target_search = next(valid_iter) # [b, 3, 32, 32], [b]
x_search, target_search = x_search.to(device), target_search.cuda(non_blocking=True)
# x : train_data , target : train_labels
# x_search : validation_data, target_search : validation_labels
# 1. update alpha
arch.step(x, target, x_search, target_search, lr, optimizer, unrolled=args.unrolled)
logits = model(x)
loss = criterion(logits, target)
# 2. update weight
optimizer.zero_grad()
loss.backward()
nn.utils.clip_grad_norm_(model.parameters(), args.grad_clip)
optimizer.step()
prec1, prec5 = utils.accuracy(logits, target, topk=(1, 5))
losses.update(loss.item(), batchsz)
top1.update(prec1.item(), batchsz)
top5.update(prec5.item(), batchsz)
if step % args.report_freq == 0:
logging.info('Step:%03d loss:%f acc1:%f acc5:%f', step, losses.avg, top1.avg, top5.avg)
return top1.avg, losses.avg
def infer(valid_queue, model, criterion):
"""
:param valid_queue:
:param model:
:param criterion:
:return:
"""
losses = utils.AverageMeter()
top1 = utils.AverageMeter()
top5 = utils.AverageMeter()
model.eval()
with torch.no_grad():
for step, (x, target) in enumerate(valid_queue):
x, target = x.to(device), target.cuda(non_blocking=True)
batchsz = x.size(0)
logits = model(x)
loss = criterion(logits, target)
prec1, prec5 = utils.accuracy(logits, target, topk=(1, 5))
losses.update(loss.item(), batchsz)
top1.update(prec1.item(), batchsz)
top5.update(prec5.item(), batchsz)
if step % args.report_freq == 0:
logging.info('>> Validation: %3d %e %f %f', step, losses.avg, top1.avg, top5.avg)
return top1.avg, losses.avg
if __name__ == '__main__': #TODO learning: why is this always used
main()
# git status
# cd path-to-git-repository
# git add .
# git commit -m "explain what did you add"
# git push origin master(branch name)