/
IMDb_step1_hybrid.py
352 lines (227 loc) · 11.8 KB
/
IMDb_step1_hybrid.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
#!/usr/bin/env python
# coding: utf-8
# In[ ]:
import os
os.environ['PYTHONHASHSEED'] = str(2019)
os.environ['TRANSFORMERS_CACHE'] = 'D:\\python_pkg_data\\huggingface\\transformers'
import json
from tqdm import tqdm_notebook
import numpy as np
np.random.seed(2019)
import random
random.seed(2019)
import torch
torch.manual_seed(2019)
torch.cuda.manual_seed_all(2019)
torch.backends.cudnn.benchmark = False
torch.backends.cudnn.deterministic = True
from sklearn.utils import shuffle
import transformers
from datasets import load_metric,load_dataset,Value
import csv
import nltk
nltk.data.path.append('D:\\python_pkg_data\\nltk_data')
from nltk.tokenize import word_tokenize
from nltk.tokenize.treebank import TreebankWordDetokenizer
import ast
import glob
import shutil
import importlib
os.environ["WANDB_DISABLED"] = "true"
from optparse import OptionParser
# In[ ]:
args = {
'ori_train_dir':'./datasets/IMDb/orig/train.tsv',
'ori_dev_dir':'./datasets/IMDb/orig/dev.tsv',
'gpu_device':0,
'tokenizer': transformers.AutoTokenizer.from_pretrained('roberta-base'),
'dataset_cache_dir':"D:\\python_pkg_data\\huggingface\\Datasets", ## local directory for datasets
'num_per_class': 25, ## number of examples per class for initial training set
'save_dir': './SF_results/IMDb_step0_sf_trainer', ##directory for saving models
'new_save_dir': './Hybrid_results/IMDb_step0_hybrid_trainer',
}
# In[ ]:
metric = load_metric("accuracy")
def compute_metrics(eval_pred):
logits, labels = eval_pred
predictions = np.argmax(logits, axis=-1)
return metric.compute(predictions=predictions, references=labels)
class CustomerDataset(torch.utils.data.Dataset):
def __init__(self, encodings, labels):
self.encodings = encodings
self.labels = labels
def __getitem__(self, idx):
item = {key: torch.tensor(val[idx]) for key, val in self.encodings.items()}
item['labels'] = torch.tensor(self.labels[idx])
return item
def __len__(self):
return len(self.labels)
def import_data(directory):
data_pos = []
data_neg = []
with open(directory,errors='ignore') as file:
file = csv.reader(file, delimiter="\t")
for idx,row in enumerate(file):
if idx!=0:
if row[0] == 'Negative':
data_neg.append({'idx':idx,'text':row[1],'label':0})
else:
data_pos.append({'idx':idx,'text':row[1],'label':1})
return data_neg,data_pos
def import_paired_data(directory,original_texts):
paired_data = {}
with open(directory,errors='ignore') as file:
file = csv.reader(file, delimiter="\t")
for idx,row in enumerate(file):
if idx!=0:
if row[2] not in paired_data.keys():
paired_data[row[2]] = []
if row[1] in original_texts:
paired_data[row[2]].append({'text':row[1],'label':0 if row[0]=='Negative'else 1,'ori_flag':1})
else:
paired_data[row[2]].append({'text':row[1],'label':0 if row[0]=='Negative'else 1,'ori_flag':0})
else:
if row[1] in original_texts:
paired_data[row[2]].append({'text':row[1],'label':0 if row[0]=='Negative'else 1,'ori_flag':1})
else:
paired_data[row[2]].append({'text':row[1],'label':0 if row[0]=='Negative'else 1,'ori_flag':0})
return paired_data
# In[ ]:
def construct_training_eval(args,random_seed,num_per_example=7):
new_save_dir = f"{args['new_save_dir']}_{random_seed}_{args['num_per_class']}_{num_per_example}"
print(new_save_dir)
if not os.path.exists(new_save_dir):
os.mkdir(new_save_dir)
if not os.path.exists(f"{new_save_dir}/runs"):
os.mkdir(f"{new_save_dir}/runs")
## reading augmented data
augmented_data_dir = f"{args['save_dir']}_{random_seed}_{args['num_per_class']}_{num_per_example}/false_rationales_augmented_step1.json"
print(augmented_data_dir)
with open(augmented_data_dir, "r") as file_name:
augmented_data = json.load(file_name)
## reading missing-rationales augmented data
missing_augmented_data_dir = f"{args['save_dir']}_{random_seed}_{args['num_per_class']}_{num_per_example}/missing_rationales_augmented_step1.json"
print(missing_augmented_data_dir)
with open(missing_augmented_data_dir, "r") as file_name:
missing_augmented_data = json.load(file_name)
train_keys = list(augmented_data.keys())
print('Training example index',train_keys)
with open(f"{new_save_dir}/keys.txt", "w") as fp:
for k in train_keys:
fp.write(str(k) +"\n")
## import train data
IMDb_data = {}
with open(args['ori_train_dir'],errors='ignore') as file:
file = csv.reader(file, delimiter="\t")
for idx,row in enumerate(file):
if len(row)>0:
if row[0] == 'Negative':
IMDb_data[row[2]] = {'text':row[1],'label':0}
else:
IMDb_data[row[2]] = {'text':row[1],'label':1}
dev_data_neg,dev_data_pos = import_data(args['ori_dev_dir'])
dev_data = dev_data_neg +dev_data_pos
## magnify with the hybrid of false-rationales and missing rationales
train_texts = []
train_labels = []
for key in train_keys:
label = IMDb_data[key]['label']
train_texts.append(IMDb_data[key]['text'])
train_labels.append(label)
threshold = int(num_per_example+1/2)
if len(missing_augmented_data[key]['candidates']) == 0:
candidates = augmented_data[key]['candidates'][:num_per_example]
train_texts = train_texts + candidates
train_labels = train_labels + [label]*len(candidates)
elif len(missing_augmented_data[key]['candidates']) <= threshold:
missing_candidates = missing_augmented_data[key]['candidates']
train_texts = train_texts + missing_candidates
train_labels = train_labels + [label]*len(missing_candidates)
num_for_false =int(num_per_example - len(missing_candidates))
candidates = augmented_data[key]['candidates'][:num_for_false]
train_texts = train_texts + candidates
train_labels = train_labels + [label]*len(candidates)
elif len(missing_augmented_data[key]['candidates'])>threshold:
missing_candidates = missing_augmented_data[key]['candidates'][:threshold]
train_texts = train_texts + missing_candidates
train_labels = train_labels + [label]*len(missing_candidates)
num_for_false = int(num_per_example - len(missing_candidates))
candidates = augmented_data[key]['candidates'][:num_for_false]
train_texts = train_texts + candidates
train_labels = train_labels + [label]*len(candidates)
eval_texts = [doc['text'] for doc in dev_data]
eval_labels = [doc['label'] for doc in dev_data]
train_encodings = args['tokenizer'](train_texts, truncation=True, padding=True)
eval_encodings = args['tokenizer'](eval_texts, truncation=True, padding=True)
print('IMDb training data statistics -----------------------')
print(np.unique(train_labels),np.bincount(train_labels))
print('IMDb eval data statistics -----------------------')
print(np.unique(eval_labels),np.bincount(eval_labels))
train_dataset = CustomerDataset(train_encodings, train_labels)
eval_dataset = CustomerDataset(eval_encodings, eval_labels)
return train_dataset, eval_dataset
def fine_tune(args,pre_trained_model,train_dataset,eval_dataset,random_seed,num_per_example):
new_save_dir = f"{args['new_save_dir']}_{random_seed}_{args['num_per_class']}_{num_per_example}"
training_args = transformers.TrainingArguments(new_save_dir,per_device_train_batch_size=4,per_device_eval_batch_size=16,
evaluation_strategy="epoch",num_train_epochs = 20.0,
save_strategy='epoch',overwrite_output_dir=True,logging_strategy='epoch',load_best_model_at_end = True,
metric_for_best_model='loss',learning_rate=5e-6)
trainer = transformers.Trainer(
model=pre_trained_model,
args=training_args,
train_dataset=train_dataset,
eval_dataset=eval_dataset,
compute_metrics=compute_metrics,
callbacks = [transformers.EarlyStoppingCallback(early_stopping_patience=5)]
)
## remove logging file
if os.path.exists("lossoutput.txt"):
os.remove('lossoutput.txt')
trainer.train()
## record best performing model based on eval accuracy
loggings = {}
with open('./lossoutput.txt','r') as file:
for line in file.readlines():
line = ast.literal_eval(line)
if 'eval_accuracy' in line.keys():
loggings[line['step']] = {'eval_accuracy':line['eval_accuracy'],'eval_loss':line['eval_loss']}
## sort by eval_acc and eval_loss descending order
dicts = [{k: v} for (k,v) in loggings.items()]
dicts.sort(key=lambda d: (-list(d.values())[0]['eval_accuracy'], -list(d.values())[0]['eval_loss'],))
best_step = list(dicts[0].keys())[0]
best_acc =dicts[0][best_step]['eval_accuracy']
# remove the rest models
for idx,directory in enumerate(glob.glob(f"{new_save_dir}/*")):
if os.path.basename(directory)[11:] != str(best_step):
try:
shutil.rmtree(directory)
except Exception:
print(f"Exception {directory}")
with open(f"{new_save_dir}/loggings.json", "w") as file_name:
json.dump(loggings, file_name)
del pre_trained_model
del trainer
return best_acc, best_step
if __name__ == "__main__":
parser = OptionParser(usage='usage: -n num_per_example -r random_seed')
parser.add_option("-n","--num_per_example", action="store", type="int", dest="num_per_example", help="number of augmented example per review", default = '7')
parser.add_option("-r","--random_seed", action="store", type="int", dest="random_seed", help="random seed for initialisation", default = '2019')
(options, _) = parser.parse_args()
num_per_example = int(options.num_per_example)
random_seed = int(options.random_seed)
# learning_rate = float(options.learning_rate)
# print(f"learning rate is {learning_rate}")
## construct training/evaluation set
train_dataset, eval_dataset = construct_training_eval(args,num_per_example=num_per_example,random_seed=random_seed)
# In[ ]:
if 'model' in globals() or 'model' in locals():
del model
old_dir = f"{args['save_dir']}_{random_seed}_{args['num_per_class']}_{num_per_example}/checkpoint*"
model_dir = glob.glob(old_dir)[0]
print(model_dir)
model = transformers.AutoModelForSequenceClassification.from_pretrained(model_dir, num_labels=2).cuda(args['gpu_device'])
best_acc, best_step = fine_tune(args,model,train_dataset,eval_dataset,random_seed,num_per_example)
# del model
# In[ ]:
# In[ ]:
# In[ ]: