Repository navigation
Expand file tree
/
Copy pathmain.py
More file actions
159 lines (144 loc) · 6.67 KB
/
Copy pathmain.py
File metadata and controls
159 lines (144 loc) · 6.67 KB
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
import numpy as np
import pandas as pd
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.utils.data as data
import argparse
from datetime import datetime
from model import DiffuQKT
from run import run_epoch
mp2path = {
'assist09': {
'ques_skill_path': './data/ASSIST09/ques_skill.csv',
'train_path': './data/ASSIST09/train_question.txt',
'test_path': './data/ASSIST09/test_question.txt',
'train_skill_path': './data/ASSIST09/train_skill.txt',
'test_skill_path': './data/ASSIST09/test_skill.txt',
'skill_max': None
},
'assist17': {
'ques_skill_path': './data/ASSIST17/ques_skill.csv',
'train_path': './data/ASSIST17/train_question.txt',
'test_path': './data/ASSIST17/test_question.txt',
'train_skill_path': './data/ASSIST17/train_skill.txt',
'test_skill_path': './data/ASSIST17/test_skill.txt',
'skill_max': None # 将在运行时从数据计算
},
'xes3g5m': {
'ques_skill_path': './data/XES3G5M/ques_skill.csv',
'train_path': './data/XES3G5M/train_question.txt',
'test_path': './data/XES3G5M/test_question.txt',
'train_skill_path': './data/XES3G5M/train_skill.txt',
'test_skill_path': './data/XES3G5M/test_skill.txt',
'skill_max': None
},
'statics2011': {
'ques_skill_path': './data/STATICS2011/ques_skill.csv',
'train_path': './data/STATICS2011/train_question.txt',
'test_path': './data/STATICS2011/test_question.txt',
'train_skill_path': './data/STATICS2011/train_skill.txt',
'test_skill_path': './data/STATICS2011/test_skill.txt',
'skill_max': None
}
}
def main(dataset):
if dataset not in mp2path:
raise ValueError(f"未知数据集: {dataset}，可选: {list(mp2path.keys())}")
date_tag = datetime.now().strftime('%Y%m%d')
with open(f'./result/{dataset}_output_{date_tag}.txt', 'w') as file:
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
if 'ques_skill_path' in mp2path[dataset]:
ques_skill_path = mp2path[dataset]['ques_skill_path']
train_path = mp2path[dataset]['train_path']
test_path = mp2path[dataset]['test_path']
train_skill_path = mp2path[dataset]['train_skill_path']
test_skill_path = mp2path[dataset]['test_skill_path']
skill_max = mp2path[dataset]['skill_max']
if skill_max is None:
# 从数据文件读取skill_max
skill_data = pd.read_csv(mp2path[dataset]['ques_skill_path'])
skill_max = int(skill_data['skill_id'].max()) + 2
if 'ques_skill_path' in mp2path[dataset]:
pro_max = int(pd.read_csv(ques_skill_path)['problem_id'].max()) + 2
else:
pro_max = skill_max
p = 0.4
d = 128
learning_rate = 0.002
epochs = 500
batch_size = 512
min_seq = 3
max_seq = 200
grad_clip = 15.0
head = 2
beta_start = 0.0001
beta_end = 0.02
diff_num_step = 100
lamda_1 = 0.01
lamda_2 = 0.001
patience = 30
avg_auc = 0
avg_acc = 0
sublist = []
# 只训练1次，而不是5次
for now_step in range(1):
best_acc = 0
best_auc = 0
best_epoch = -1
state = {'auc': 0, 'acc': 0, 'loss': 0, 'best_epoch': -1}
model = DiffuQKT(pro_max, skill_max, d, p, head, beta_start, beta_end, diff_num_step, lamda_1, lamda_2)
model = model.to(device)
criterion = nn.BCELoss()
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate, weight_decay=1e-5)
one_p = 0
for epoch in range(epochs):
# 训练
train_loss, train_acc, train_auc = run_epoch(pro_max, train_path, train_skill_path, batch_size,
True, min_seq, max_seq, model, optimizer, criterion,
device,
grad_clip)
# 评估测试集
test_loss, test_acc, test_auc = run_epoch(pro_max, test_path, test_skill_path, batch_size, False,
min_seq, max_seq, model, optimizer, criterion, device,
grad_clip)
print(
f'epoch: {epoch}, train_loss: {train_loss:.4f}, train_acc: {train_acc:.4f}, train_auc: {train_auc:.4f}, test_auc: {test_auc:.4f}')
# 保存模型
# Save latest model each epoch.
torch.save(model.state_dict(), f"./result/DiffuQKT_{dataset}_{now_step}_{date_tag}_last_model.pkl")
state['auc'] = test_auc
state['acc'] = test_acc
state['loss'] = test_loss
torch.save(state, f'./result/DiffuQKT_{dataset}_{now_step}_{date_tag}_state.ckpt')
sublist.append(test_auc)
# 早停机制（基于测试集AUC）
if test_auc > best_auc:
one_p = 0
best_auc = test_auc
best_acc = test_acc
best_epoch = epoch
# Save best-by-test-AUC model.
torch.save(model.state_dict(), f"./result/DiffuQKT_{dataset}_{now_step}_{date_tag}_model.pkl")
torch.save(model.state_dict(), f"./result/DiffuQKT_{dataset}_{now_step}_{date_tag}_best_model.pkl")
state['best_epoch'] = best_epoch
state['best_auc'] = best_auc
state['best_acc'] = best_acc
torch.save(state, f'./result/DiffuQKT_{dataset}_{now_step}_{date_tag}_state.ckpt')
else:
one_p += 1
if one_p >= patience:
print(f'早停：{patience}个epoch没有改进，停止训练')
break
avg_auc += best_auc
avg_acc += best_acc
print(f'*******************************************************************************')
print(f'最终结果 - best_acc: {avg_acc:.4f}, best_auc: {avg_auc:.4f}')
print(f'*******************************************************************************')
line = '\n'.join(str(item) for item in sublist)
file.write(line)
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--dataset', type=str, required=True, choices=list(mp2path.keys()), help='选择要训练的数据集')
args = parser.parse_args()
main(args.dataset)