-
Notifications
You must be signed in to change notification settings - Fork 1
/
train.py
108 lines (93 loc) · 2.94 KB
/
train.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
import sys
import getopt
from transformers import GPT2Tokenizer, GPT2LMHeadModel
from transformers import Trainer, TrainingArguments
from utils import load_dataset, load_data_collator
def train(
train_file_path,
model_name_or_type='gpt2',
output_dir='./results',
overwrite_output_dir=False,
per_device_train_batch_size=8,
num_train_epochs=3.0,
logging_dir='./logs',
loggint_steps=10
):
tokenizer = GPT2Tokenizer.from_pretrained(model_name_or_type)
train_dataset = load_dataset(train_file_path, tokenizer)
data_collator = load_data_collator(tokenizer)
training_args = TrainingArguments(
output_dir=output_dir,
overwrite_output_dir=overwrite_output_dir,
per_device_train_batch_size=per_device_train_batch_size,
num_train_epochs=num_train_epochs,
logging_dir=logging_dir,
logging_steps=loggint_steps
)
model = GPT2LMHeadModel.from_pretrained(model_name_or_type)
trainer = Trainer(
model=model,
args=training_args,
data_collator=data_collator,
train_dataset=train_dataset,
)
trainer.train()
trainer.save_model()
tokenizer.save_pretrained(output_dir)
model.save_pretrained(output_dir)
def main():
try:
opts, args = getopt.getopt(sys.argv[1:],
'',
[
'train_file_path=',
'model_name_or_type=',
'output_dir=',
'overwrite_output_dir',
'per_device_train_batch_size=',
'num_train_epochs=',
'logging_dir=',
'logging_steps=',
])
except getopt.GetoptError as err:
print(err)
sys.exit(2)
train_file_path = None
model_name_or_type='gpt2'
output_dir='./results'
overwrite_output_dir=False
per_device_train_batch_size=8
num_train_epochs=3.0
logging_dir='./logs'
logging_steps=10
for o, a in opts:
if o == '--train_file_path':
train_file_path = a
elif o == '--model_name_or_type':
model_name_or_type = a
elif o == '--output_dir':
output_dir = a
elif o == '--overwrite_output_dir':
overwrite_output_dir = True
elif o == '--per_device_train_batch_size':
per_device_train_batch_size = int(a)
elif o == '--num_train_epochs':
num_train_epochs = float(a)
elif o == '--logging_dir':
logging_dir = a
elif o == '--logging_steps':
logging_steps = int(a)
if train_file_path is None:
sys.exit(3)
train(
train_file_path=train_file_path,
model_name_or_type=model_name_or_type,
output_dir=output_dir,
overwrite_output_dir=overwrite_output_dir,
per_device_train_batch_size=per_device_train_batch_size,
num_train_epochs=num_train_epochs,
logging_dir=logging_dir,
loggint_steps=logging_steps
)
if __name__ == '__main__':
main()