-
Notifications
You must be signed in to change notification settings - Fork 0
/
utils.py
234 lines (198 loc) · 8.1 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
232
233
234
from decimal import MAX_EMAX
import torch
import random
import numpy as np
import time
import sys
import logging
import os
import time
from sklearn.metrics import classification_report
from sklearn.metrics import f1_score
import random
import sys
from dataclasses import dataclass, field
from typing import Optional
from copy import deepcopy
from tqdm import tqdm
import hashlib
import json
from sklearn.cluster import KMeans
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.distributions.beta import Beta
import torch.optim as optim
from data_loader import get_data_loader
from sampler import data_sampler
from texttable import Texttable
from sklearn.metrics import confusion_matrix
from model.RE_model import BertForRE
def confusion_matrix_view(true_label, pred_label, labels, logger):
cf_matrix = confusion_matrix(true_label, pred_label)
for2later = np.triu(cf_matrix, 1).sum()
later2for = np.tril(cf_matrix, -1).sum()
right = np.tril(cf_matrix).sum() - later2for
logger.info(f'Total: {cf_matrix.sum()}; Right: {right}; for2later False: {for2later}; later2for False: {later2for}')
table = Texttable()
table.add_row([" "] + [i[:8] for i in labels])
table.set_max_width(2000)
for idx, r in enumerate(cf_matrix):
table.add_row([labels[idx][:8]] + [str(i) for i in cf_matrix[idx]])
return table.draw()
def set_seed(seed):
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
torch.cuda.manual_seed(seed)
np.random.seed(seed)
random.seed(seed)
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
def select_data(config, encoder, sample_set, select_num=None):
data_loader = get_data_loader(config, sample_set, shuffle=False, drop_last=False, batch_size=1)
features = []
encoder.eval()
for step, (_, _, tokens, _, _) in enumerate(data_loader):
tokens = tokens.to(config.device)
with torch.no_grad():
try:
feature = encoder(tokens).rel_hidden_states.cpu()
except:
feature = encoder(tokens).cpu()
features.append(feature)
features = np.concatenate(features)
if select_num is None:
num_clusters = min(config.memory_size, len(sample_set))
else:
num_clusters = min(select_num, len(sample_set))
distances = KMeans(n_clusters=num_clusters, random_state=0).fit_transform(features)
mem_set = []
for k in range(num_clusters):
sel_index = np.argmin(distances[:, k])
instance = sample_set[sel_index]
mem_set.append(instance)
return mem_set
def get_proto(config, encoder, mem_set, r=None):
# aggregate the prototype set for further use.
data_loader = get_data_loader(config, mem_set, False, False, 1)
features = []
encoder.eval()
for _, (_, _, tokens, _, _) in enumerate(data_loader):
tokens = torch.stack([x.to(config.device) for x in tokens], dim=0)
with torch.no_grad():
feature = encoder(tokens)
features.append(feature)
if r is not None:
features.append(r.unsqueeze(0))
features = [x / x.norm() for x in features]
features = torch.cat(features, dim=0)
proto = torch.mean(features, dim=0, keepdim=True)
return proto
class Save:
def __init__(self, model, args):
self.model = model
self.args = args
def __call__(self, score, name):
torch.save({'param': self.model.state_dict(),
'score': score, 'args': self.args},
name)
def write_select_data_to_file(file, select_data, id2rel, pid2name):
with open(file, 'w') as f:
for rel in select_data:
if len(select_data[rel]) == 0:
continue
rel2data = select_data[rel]
if type(rel) == int:
rel = id2rel[rel]
f.write(rel + " " + str(pid2name.get(rel, rel)) + '\n')
for data_tmp in rel2data:
f.write(data_tmp['string'].replace('\n','') + '\n')
f.write('\n\n')
def get_aca_data(config, training_data, current_relations, tokenizer):
rel_id = config.num_of_relation
aca_data = []
for rel1, rel2 in zip(current_relations[:config.rel_per_task // 2], current_relations[config.rel_per_task // 2:]):
datas1 = training_data[rel1]
datas2 = training_data[rel2]
L = 5
for data1, data2 in zip(datas1, datas2):
token1 = data1['tokens'][1:-1][:]
e11 = token1.index(30522); e12 = token1.index(30523)
e21 = token1.index(30524); e22 = token1.index(30525)
if e21 <= e11:
continue
token1_sub = token1[max(0, e11-L): min(e12+L+1, e21)]
token2 = data2['tokens'][1:-1][:]
e11 = token2.index(30522); e12 = token2.index(30523)
e21 = token2.index(30524); e22 = token2.index(30525)
if e21 <= e11:
continue
token2_sub = token2[max(e12+1, e21-L): min(e22+L+1, len(token2))]
token = [101] + token1_sub + token2_sub + [102]
aca_data.append({
'relation': rel_id,
'tokens': token,
'string': tokenizer.decode(token)
})
for index in [30522, 30523, 30524, 30525]:
assert index in token and token.count(index) == 1
rel_id += 1
for rel in current_relations:
if rel in ['P26', 'P3373', 'per:siblings', 'org:alternate_names', 'per:spous', 'per:alternate_names', 'per:other_family']:
continue
for data in training_data[rel]:
token = data['tokens'][:]
e11 = token.index(30522); e12 = token.index(30523)
e21 = token.index(30524); e22 = token.index(30525)
token[e11] = 30524; token[e12] = 30525
token[e21] = 30522; token[e22] = 30523
aca_data.append({
'relation': rel_id,
'tokens': token,
'string': tokenizer.decode(token)
})
for index in [30522, 30523, 30524, 30525]:
assert index in token and token.count(index) == 1
rel_id += 1
return aca_data
def save_representation_to_file(config, model, sampler, save_relations, save_file, memory_datas):
datas = []
rel2id = sampler.rel2id
id2rel = sampler.id2rel
for relation in save_relations:
datas.extend(sampler.test_dataset[rel2id[relation]])
data_loader = get_data_loader(config, datas, batch_size=256)
model.eval()
linear_param = model.fc.weight.data.cpu().clone()
save_datas = {'rep':[], 'label':[], 'is_memory': [], 'classifier':{}}
for relation in save_relations:
save_datas['classifier'][relation] = linear_param[rel2id[relation]]
# for _, (_, labels, tokens, _, _) in enumerate(tqdm(data_loader, desc="Evaluate {}".format(mode))):
with torch.no_grad():
for _, (_, labels, tokens, _, _) in enumerate(data_loader):
labels = labels.tolist()
tokens = tokens.to(config.device)
_, rep = model(tokens)
features = rep.cpu()
labels = [id2rel[idx] for idx in labels]
save_datas['label'].extend(labels)
save_datas['rep'].extend(features)
save_datas['is_memory'].extend([0 for _ in range(len(labels))])
if memory_datas is not None:
datas = []
for relation in save_relations:
datas.extend(memory_datas.get(relation, []))
data_loader = get_data_loader(config, datas, batch_size=256)
model.eval()
with torch.no_grad():
for _, (_, labels, tokens, _, _) in enumerate(data_loader):
labels = labels.tolist()
tokens = tokens.to(config.device)
_, rep = model(tokens)
features = rep.cpu()
labels = [id2rel[idx] for idx in labels]
save_datas['label'].extend(labels)
save_datas['rep'].extend(features)
save_datas['is_memory'].extend([1 for _ in range(len(labels))])
torch.save(obj=save_datas, f=save_file)