DACE/config.py at main · lvXiangwei/DACE · GitHub

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
data_statics = {
  "assist09":{
    'n_pid': 17737,
    'n_question': 167,
    'seqlen': 200,
    'data_dir': '../data/assist09',
    'data_name': 'assist09' + '_pid'
  },"ednet":{
    'n_pid': 12103,
    'n_question': 1598,
    'seqlen': 200,
    'data_dir': '../data/ednet',
    'data_name': 'ednet' + '_pid'
  },"assist12":{
    'n_pid': 53070,
    'n_question': 265,
    'seqlen': 200,
    'data_dir': '../data/assist12',
    'data_name': 'assist12' + '_pid'
  },"python":{
    'n_pid': 1149,
    'n_question': 73,
    'seqlen': 200,
    'data_dir': '../data/python',
    'data_name': 'python' + '_pid'
  }
}


class Config:
    dataset='assist09'
     ### dv_train
    bias_type = "None" # plag-by_pro, guess
    bias_p=0.0 # plag:0.5; plag-by_pro:0.5; guess:0.3
    inject_p=0.0


    file_name = f"{dataset}_bias_{bias_type}_bias_p_{bias_p}.log"
    fb_epoch=10
    max_iter=300


    disentangle=1
    model="DACE_pid"

    n_question=data_statics[dataset]['n_question']
    n_pid=data_statics[dataset]['n_pid']
    seqlen=data_statics[dataset]['seqlen']
    data_dir=data_statics[dataset]['data_dir']
    data_name=data_statics[dataset]['data_name']
    model_type="DACE"
    save=dataset


    ### common
    train_set=1
    seed=224
    optim='adam'
    batch_size=32
    lr=1e-5
    maxgradnorm=-1
    final_fc_dim=512
    l2=1e-5

    ### Knowledge state extractor
    d_model=256
    d_ff=1024
    dropout=0.1
    n_block=1
    n_head=8
    kq_same=1