-
Notifications
You must be signed in to change notification settings - Fork 7
/
HTTN.py
248 lines (215 loc) · 9.48 KB
/
HTTN.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
import argparse
import os
import time
import torch
import torch.nn as nn
import torch.nn.parallel
import torch.backends.cudnn as cudnn
import torch.distributed as dist
import torch.nn.functional as F
import torch.optim
import torch.utils.data
import models
import data_got
import numpy as np
from utils import Bar, Logger, AverageMeter, precision_k, calc_acc, ndcg_k, calc_f1, mkdir_p
from sklearn.metrics import f1_score, precision_score, recall_score, accuracy_score
parser = argparse.ArgumentParser(description='PyTorch Training')
parser.add_argument('-j', '--workers', default=4, type=int, metavar='N',
help='number of data loading workers (default: 4)')
parser.add_argument('-b', '--batch-size', default=90, type=int,
metavar='N', help='mini-batch size (default: 64)')
parser.add_argument('-c', '--checkpoint', default='imprint_checkpoint', type=str, metavar='PATH',
help='path to save checkpoint (default: imprint_checkpoint)')
parser.add_argument('--model', default='samp36_pretrain_checkpoint/model_best.pth.tar', type=str, metavar='PATH',
help='path to model (default: none)')
parser.add_argument('--random', action='store_true', help='whether use random novel weights')
parser.add_argument('--num_sample', default=5, type=int,
metavar='N', help='number of novel sample (default: 1)')
parser.add_argument('--test-novel-only', action='store_true', help='whether only test on novel classes')
parser.add_argument('--aug', action='store_true', help='whether use data augmentation during training')
parser.add_argument('--lstm_hid_dim', default=150, type=int, metavar='N',
help='lstm_hid_dim')
parser.add_argument('--num_class', default=36, type=int, metavar='N',
help='the number of class')
parser.add_argument('--epochs', default=7 ,type=int, metavar='N',
help='number of total epochs to run')
parser.add_argument('--samp_freq', default=30, type=int, metavar='N',
help='number of total epochs to run')
parser.add_argument('--lr', '--learning-rate', default=0.00001, type=float,
metavar='LR', help='initial learning rate')
def main():
global args, best_micro
args = parser.parse_args()
base_transf, embed = data_got.one_sample_base2avg(batch_size=args.batch_size,sample_num=args.num_sample,samp_freq=args.samp_freq,num_class=args.num_class)
Ftest_loader, novel_loader,novelall_loader= data_got.Nload_data(batch_size=args.batch_size,sample_num=args.num_sample,num_class=args.num_class)
embed = torch.from_numpy(embed).float()
if not os.path.isdir(args.checkpoint):
mkdir_p(args.checkpoint)
model = models.Net(embed, args.lstm_hid_dim, num_classes=args.num_class).cuda()
print('==> Reading from model checkpoint..')
assert os.path.isfile(args.model), 'Error: no model checkpoint directory found!'
checkpoint = torch.load(args.model)
model.load_state_dict(checkpoint['state_dict'])
print("=> loaded model checkpoint '{}' (epoch {})"
.format(args.model, checkpoint['epoch']))
cudnn.benchmark = True
real_weight=model.classifier.fc.weight.data
criterion = nn.MSELoss()
trans_model = models.Transfer().cuda()
optimizer = torch.optim.Adam( trans_model.parameters(), lr=0.01, betas=(0.9, 0.99))
for epoch in range(args.epochs):
train_loss= train(base_transf, trans_model,model, criterion, optimizer,real_weight)
print("loss",train_loss)
imprint(novel_loader, model,trans_model)
# model_criterion= nn.BCELoss()
microF1 = validate(Ftest_loader, model)
# model_optimizer = torch.optim.Adam(model.parameters(), lr=args.lr, betas=(0.9, 0.99))
# for i in range(2):
# print("fine-tuning")
# train_loss, trn_micro, trn_macro = fine_tuning(novelall_loader, model, model_criterion, model_optimizer)
# microF1 = validate(Ftest_loader, model)
# print("microF1 and macroF1",microF1,macroF1)
def train(train_loader, trans_model,model, criterion,optimizer,real_weight):
trans_model.train()
base_rep=[]
losses=[]
with torch.no_grad():
for batch_idx, (input, target) in enumerate(train_loader):
input = input.cuda()
output = model.extract(input)
base_rep.extend(output.cpu().numpy())
base_rep =np.array(base_rep)
base_rep= torch.from_numpy(base_rep ).cuda()
new_weight = torch.zeros(1350, 128).cuda()
j = 0
for i in range(1350):
tmp =base_rep[j:j + args.num_sample]
tmp = torch.sum(tmp, 0) / args.num_sample
new_weight[i] = tmp / tmp.norm(p=2)
j = j + args.num_sample
e=0
for h in range(args.samp_freq):
doc_avg=new_weight[e:e+args.num_class,:]
e=e+args.num_class
output = trans_model(doc_avg)
loss = criterion(output, real_weight)
losses.append(float(loss))
optimizer.zero_grad()
loss.backward()
optimizer.step()
avg_loss = np.mean(losses)
return avg_loss
def imprint(novel_loader, model, trans_model):
batch_time = AverageMeter()
data_time = AverageMeter()
end = time.time()
with torch.no_grad():
for batch_idx, (input, target) in enumerate(novel_loader):
data_time.update(time.time() - end)
input = input.cuda()
output = model.extract(input)
if batch_idx == 0:
output_stack = output
target_stack = target
else:
output_stack = torch.cat((output_stack, output), 0)
target_stack = torch.cat((target_stack, target), 0)
batch_time.update(time.time() - end)
end = time.time()
new_weight = torch.zeros(9, 128).cuda()
j=0
for i in range(9):
tmp=output_stack[j:j+args.num_sample]
tmp=torch.sum(tmp,0)/args.num_sample
new_weight[i] = tmp / tmp.norm(p=2)
j=j+args.num_sample
tail_real=trans_model.transfor(new_weight)
weight = torch.cat([model.classifier.fc.weight.data, tail_real])
model.classifier.fc = nn.Linear(128, 54, bias=False)
model.classifier.fc.weight.data = weight
print("the time cost",data_time.val)
print(format(data_time.val, '.8f'))
print('imprint done')
def validate(val_loader, model):
data_time = AverageMeter()
microF1 = AverageMeter()
test_p1, test_p3, test_p5 = 0, 0, 0
test_ndcg1, test_ndcg3, test_ndcg5 = 0, 0, 0
model.eval()
with torch.no_grad():
end = time.time()
for batch_idx, (input, target) in enumerate(val_loader):
data_time.update(time.time() - end)
input = input.cuda()
target = target.cuda()
output = model(input)
target = target.data.cpu().float()
output = output.data.cpu()
_p1, _p3, _p5 = precision_k(output.topk(k=5)[1].numpy(), target.numpy(), k=[1, 3, 5])
test_p1 += _p1
test_p3 += _p3
test_p5 += _p5
_ndcg1, _ndcg3, _ndcg5 = ndcg_k(output.topk(k=5)[1].numpy(), target.numpy(), k=[1, 3, 5])
test_ndcg1 += _ndcg1
test_ndcg3 += _ndcg3
test_ndcg5 += _ndcg5
output[output > 0.5] = 1
output[output <= 0.5] = 0
micro, macro = calc_f1(target, output)
microF1.update(micro.item(), input.size(0))
np.set_printoptions(formatter={'float': '{: 0.4}'.format})
print('the result of micro: \n',microF1.avg)
test_p1 /= len(val_loader)
test_p3 /= len(val_loader)
test_p5 /= len(val_loader)
test_ndcg1 /= len(val_loader)
test_ndcg3 /= len(val_loader)
test_ndcg5 /= len(val_loader)
print("precision@1 : %.4f , precision@3 : %.4f , precision@5 : %.4f " % (test_p1, test_p3, test_p5))
print("ndcg@1 : %.4f , ndcg@3 : %.4f , ndcg@5 : %.4f " % (test_ndcg1, test_ndcg3, test_ndcg5))
return (microF1.avg)
def fine_tuning(train_loader, model, criterion, optimizer):
batch_time = AverageMeter()
data_time = AverageMeter()
losses = AverageMeter()
microF1 = AverageMeter()
macroF1 = AverageMeter()
model.train()
end = time.time()
bar = Bar('Training', max=len(train_loader))
for batch_idx, (input, target) in enumerate(train_loader):
# measure data loading time
data_time.update(time.time() - end)
input = input.cuda()
target = target.cuda()
output = model(input)
loss = criterion(output, target.float())
target = target.data.cpu().float()
output=output.data.cpu()
micro,macro = calc_f1( target, output)
losses.update(loss.item(), input.size(0))
microF1.update(micro.item(), input.size(0))
macroF1.update(macro.item(), input.size(0))
optimizer.zero_grad()
loss.backward()
optimizer.step()
batch_time.update(time.time() - end)
end = time.time()
model.weight_norm()
bar.suffix = '({batch}/{size}) Data: {data:.3f}s | Batch: {bt:.3f}s | Total: {total:} | ETA: {eta:} | Loss: {loss:.4f} | Micro-f1: {microF1: .4f} |Macro-f1: {macroF1: .4f}'.format(
batch=batch_idx + 1,
size=len(train_loader),
data=data_time.val,
bt=batch_time.val,
total=bar.elapsed_td,
eta=bar.eta_td,
loss=losses.avg,
microF1=microF1.avg,
macroF1=macroF1.avg,
)
bar.next()
bar.finish()
return (losses.avg, microF1.avg, macroF1.avg)
if __name__ == '__main__':
main()