-
Notifications
You must be signed in to change notification settings - Fork 8
/
train-net.py
236 lines (193 loc) · 8.4 KB
/
train-net.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
import torch
from torch import nn
import utils
import numpy as np
import os
import time
from models.lenet import FConvMNIST
from models.cifarnet import CIFARNet, CIFARNetNew
import utils
from logger import Logger
from utils import tonp
import sys
from torch import distributions as dist
import models
from models.bayes import _Bayes, BayesConv2d, FFGLinear
from torch.optim.lr_scheduler import ReduceLROnPlateau, MultiStepLR
import myexman
from torch.nn import functional as F
def adjust_learning_rate(optimizer, lr):
for param_group in optimizer.param_groups:
param_group['lr'] = lr
def lr_linear(epoch):
lr = args.lr * np.minimum((args.decrease_from - epoch) * 1. / (args.epochs - args.decrease_from) + 1, 1.)
return max(0, lr)
def predict(data, net):
pred = []
l = []
for x, y in data:
l.append(y.numpy())
x = x.to(device)
p = F.log_softmax(net(x), dim=1)
pred.append(p.data.cpu().numpy())
return np.concatenate(pred), np.concatenate(l)
parser = myexman.ExParser(file=__file__)
parser.add_argument('--name', default='')
parser.add_argument('--data', default='cifar')
parser.add_argument('--num_examples', default=None, type=int)
parser.add_argument('--data_split_seed', default=42, type=int)
parser.add_argument('--resume', default='')
parser.add_argument('--lr', default=0.001, type=float, help='Initial learning rate')
parser.add_argument('--gamma', default=0.5, type=float)
parser.add_argument('--do', default=[], type=float, nargs='*')
parser.add_argument('--epochs', default=100, type=int, help='Number of epochs')
parser.add_argument('--decrease_from', default=0, type=int)
parser.add_argument('--bs', default=256, type=int, help='Batch size')
parser.add_argument('--test_bs', default=500, type=int, help='Batch size for test dataloader')
parser.add_argument('--model', default='vgg')
parser.add_argument('--model_size', default=1., type=float)
parser.add_argument('--pretrained', default='')
parser.add_argument('--filters_list', default=[], nargs='*', type=str)
parser.add_argument('--seed', default=42, type=int)
parser.add_argument('--init', default='xavier')
parser.add_argument('--init_list', type=str, nargs='*', default=[])
parser.add_argument('--vae', default='')
parser.add_argument('--vae_list', type=str, nargs='*', default=[])
parser.add_argument('--milestones', type=int, nargs='*', default=[])
parser.add_argument('--net_cfg', default='E')
parser.add_argument('--hid_dim', default=[32, 64], type=int, nargs='+')
parser.add_argument('--n_classes', default=10, type=int)
parser.add_argument('--l2', default=0., type=float)
parser.add_argument('--eval_freq', default=1, type=int)
parser.add_argument('--dwp_reg', default=0., type=float)
parser.add_argument('--dwp_samples', default=1, type=int)
parser.add_argument('--rfe', default=0, type=int)
parser.add_argument('--fastconv', default=0, type=int)
parser.add_argument('--aug', default=0, type=int)
args = parser.parse_args()
use_cuda = torch.cuda.is_available()
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
torch.cuda.manual_seed_all(args.seed)
torch.manual_seed(args.seed)
np.random.seed(args.seed)
fmt = {
'lr': '.5f',
'sec': '.0f',
}
logger = Logger('logs', base=args.root, fmt=fmt)
# Load Datasets
trainloader, testloader = utils.load_dataset(data=args.data, train_bs=args.bs, test_bs=args.test_bs,
num_examples=args.num_examples, seed=args.data_split_seed,
augmentation=(args.aug == 1))
if args.model == 'fconv':
net = FConvMNIST(args.net_cfg, device=device, hid_dim=args.hid_dim, do=args.do)
elif args.model == 'cifarnet':
net = CIFARNet(args.net_cfg, device=device, n_classes=args.n_classes, do=args.do, k=args.model_size)
elif args.model == 'cifarnetnew':
net = CIFARNetNew(args.net_cfg, device=device, n_classes=args.n_classes, do=args.do, k=args.model_size,
vae_list=args.vae_list)
else:
raise NotImplementedError
# Initialization
if hasattr(net, 'weights_init'):
net.weights_init(args.init_list, args.vae_list, pretrained=args.pretrained, filters_list=args.filters_list)
else:
utils.net_init(net, args.init, args.vae)
if args.dwp_reg != 0:
net.set_dwp_regularizer(args.vae_list)
# Optimizer
train_params = []
if args.rfe == 0:
train_params = net.parameters()
elif args.rfe == 1:
tr_modeules = [net.classifier]
train_params = list(net.classifier.parameters())
for m in net.features.modules():
if isinstance(m, nn.BatchNorm2d):
train_params += list(m.parameters())
tr_modeules += [m]
print('==> Random Feature Extration mode')
print(*tr_modeules)
else:
raise NotImplementedError
opt = torch.optim.Adam(train_params, lr=args.lr)
lrscheduler = MultiStepLR(opt, args.milestones, gamma=args.gamma)
# Load params if fine-tuning
if args.resume:
net.load_state_dict(torch.load(os.path.join(args.resume, 'model.torch')))
opt.load_state_dict(torch.load(os.path.join(args.resume, 'opt.torch')))
N = len(trainloader.dataset)
t0 = time.time()
it = 0
for e in range(1, args.epochs + 1):
if args.milestones:
lrscheduler.step()
else:
adjust_learning_rate(opt, lr_linear(e - 1))
net.train()
train_acc = utils.MovingMetric()
train_nll = utils.MovingMetric()
train_loss = utils.MovingMetric()
opt.zero_grad()
for x, y in trainloader:
opt.zero_grad()
it += 1
x = x.to(device)
y = y.to(device)
p = net(x)
data_term = F.cross_entropy(p, y)
l2_norm = torch.FloatTensor([0.]).to(device)
if args.l2 != 0:
l2_norm = torch.sum(torch.stack([torch.sum(p**2) for p in net.features.parameters()]))
dwp_reg = 0.
if args.dwp_reg != 0.:
dwp_reg = net.get_dwp_reg(backward=True, n_tries=args.dwp_samples, weight=args.dwp_reg)
loss = data_term + args.l2 * l2_norm
loss.backward()
opt.step()
loss += args.dwp_reg * dwp_reg
acc = torch.sum(p.max(1)[1] == y)
train_acc.add(acc.item(), p.size(0))
train_nll.add(data_term.item() * x.size(0), x.size(0))
train_loss.add(loss.item() * x.size(0), x.size(0))
if args.fastconv == 1:
if (it % args.eval_freq) == 0 or it == 1:
net.eval()
logp_test, labels = predict(testloader, net)
test_acc = np.mean(logp_test.argmax(1) == labels)
test_nll = -logp_test[np.arange(len(labels)), labels].mean()
logger.add_scalar(it, 'loss', train_loss.get_val())
logger.add_scalar(it, 'train_nll', train_nll.get_val())
logger.add_scalar(it, 'test_nll', test_nll)
logger.add_scalar(it, 'train_acc', train_acc.get_val())
logger.add_scalar(it, 'test_acc', test_acc)
logger.add_scalar(it, 'lr', opt.param_groups[0]['lr'])
logger.add_scalar(it, 'sec', time.time() - t0)
logger.add_scalar(it, 'l2_norm', l2_norm.item())
logger.iter_info()
logger.save()
torch.save(net.state_dict(), os.path.join(args.root, 'model.torch'))
torch.save(opt.state_dict(), os.path.join(args.root, 'opt.torch'))
t0 = time.time()
net.train()
if ((e % args.eval_freq) == 0 or e == 1) and (args.fastconv == 0):
net.eval()
logp_test, labels = predict(testloader, net)
test_acc = np.mean(logp_test.argmax(1) == labels)
test_nll = -logp_test[np.arange(len(labels)), labels].mean()
logger.add_scalar(e, 'loss', train_loss.get_val())
logger.add_scalar(e, 'train_nll', train_nll.get_val())
logger.add_scalar(e, 'test_nll', test_nll)
logger.add_scalar(e, 'train_acc', train_acc.get_val())
logger.add_scalar(e, 'test_acc', test_acc)
logger.add_scalar(e, 'lr', opt.param_groups[0]['lr'])
logger.add_scalar(e, 'sec', time.time() - t0)
logger.add_scalar(e, 'l2_norm', l2_norm.item())
logger.add_scalar(e, 'dwp_reg', dwp_reg)
logger.iter_info()
logger.save()
torch.save(net.state_dict(), os.path.join(args.root, 'model.torch'))
torch.save(opt.state_dict(), os.path.join(args.root, 'opt.torch'))
t0 = time.time()
torch.save(net.state_dict(), os.path.join(args.root, 'model.torch'))
torch.save(opt.state_dict(), os.path.join(args.root, 'opt.torch'))