-
Notifications
You must be signed in to change notification settings - Fork 2
/
train.py
66 lines (50 loc) · 2.34 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
import argparse
import datetime
import random
import numpy as np
import torch
from dataset import DatasetBuilder
from rmdt import RumorDetector
def train(opt):
DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
# get builder
dataset_builder = DatasetBuilder(opt.wv_index_path, opt.wv_path, DEVICE)
# build dataset
print("Building dataset...")
train_dataset = dataset_builder.build_dataset_from_file(opt.train_path)
valid_dataset = dataset_builder.build_dataset_from_file(opt.valid_path)
eval_dataset = dataset_builder.build_dataset_from_file(opt.eval_path)
rumor_detector = RumorDetector(DEVICE)
print("Start training...")
rumor_detector.train(train_dataset, valid_dataset, opt.epochs, opt.lr, opt.batch_size, opt.log_path)
rumor_detector.save(opt.save_path)
rumor_detector.load(opt.save_path)
rumor_detector.eval(eval_dataset, opt.log_path)
if __name__ == "__main__":
# 解析参数
parser = argparse.ArgumentParser()
parser.add_argument("--train_path", type=str, default="./data/dataset/train/")
parser.add_argument("--valid_path", type=str, default="./data/dataset/valid/")
parser.add_argument("--eval_path", type=str, default="./data/dataset/eval/")
parser.add_argument("--save_path", type=str, default="./data/model/rmdt.pt")
parser.add_argument("--log_path", type=str, default="./data/model/report.log")
parser.add_argument("--wv_index_path", type=str, default="./data/dict/pretrain_wv.index.json")
parser.add_argument("--wv_path", type=str, default="./data/dict/pretrain_wv.vec.dat")
parser.add_argument("--batch_size", type=int, default=8)
parser.add_argument("--lr", type=float, default=1e-3)
parser.add_argument("--epochs", type=int, default=50)
parser.add_argument("--seed", type=int, default=1)
opt = parser.parse_args()
print(opt)
# 固定随机种子
random.seed(opt.seed)
np.random.seed(opt.seed)
torch.manual_seed(opt.seed)
torch.cuda.manual_seed(opt.seed)
with open(opt.log_path, "a", encoding="utf8") as f:
print("*"*10+"Start: "+datetime.datetime.now().strftime("%F %T")+"*"*10, file=f)
print(opt, file=f)
# 开始训练
train(opt)
with open(opt.log_path, "a", encoding="utf8") as f:
print("*"*10+"End: "+datetime.datetime.now().strftime("%F %T")+"*"*10+"\n", file=f)