AdvDrop/parse.py at main · Arthurma71/AdvDrop · 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
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
import argparse


def parse_args():
    parser = argparse.ArgumentParser()
    parser.add_argument('--vis', nargs='?', default=-1,
                        help='we only want test value.')
    parser.add_argument('--test_only', nargs='?', default=False,
                        help='we only want test value.')
    parser.add_argument('--data_path', nargs='?', default='./data/',
                        help='Input data path.')
    parser.add_argument('--dataset', nargs='?', default='yahoo.new',
                        help='Choose a dataset')
    parser.add_argument('--embed_size', type=int, default=64,
                        help='Embedding size.')
    parser.add_argument('--batch_size', type=int, default=1024,
                        help='Batch size.')
    parser.add_argument('--lr', type=float, default=5e-4,
                        help='Learning rate.')
    parser.add_argument('--regs', type=float, default=1e-5,
                        help='Regularization.')
    parser.add_argument('--epoch', type=int, default=1600,
                        help='Number of epoch.')
    parser.add_argument('--Ks', nargs='?', default= [3],
                        help='Evaluate on Ks optimal items.')
    parser.add_argument('--log_interval', type=int, default=10,
                        help='log\'s interval epoch while training')
    parser.add_argument('--verbose', type=int, default=1,
                        help='Interval of evaluation.')
    parser.add_argument('--saveID', type=str, default="",
                        help='Specify model save path.')
    parser.add_argument('--patience', type=int, default=20,
                        help='Early stopping point.')
    parser.add_argument('--checkpoint', type=str, default='./',
                        help='Specify model save path.')
    parser.add_argument('--modeltype', type=str, default= 'BC_LOSS',
                        help='Specify model save path.')
    parser.add_argument('--cuda', type=int, default=0,
                        help='Specify which gpu to use.')
    parser.add_argument('--IPStype', type=str, default='cn',
                        help='Specify the mode of weighting')
    parser.add_argument('--n_layers', type=int, default=2,
                        help='Number of GCN layers')
    parser.add_argument('--codetype', type=str, default='train',
                        help='Calculate overlap with Item pop')
    parser.add_argument('--max2keep', type=int, default=10,
                        help='max checkpoints to keep')
    parser.add_argument('--infonce', type=int, default=1,
                        help='whether to use infonce loss or not')
    parser.add_argument('--num_workers', type=int, default=8,
                        help='number of workers in data loader')
    parser.add_argument('--neg_sample', type=int, default=1,
                        help='negative sample ratio.')
    parser.add_argument('--neg_test', type=int, default=1,
                        help='whether to use neg items for testing')

    # MACR
    parser.add_argument('--alpha', type=float, default=1e-3,
                        help='alpha')
    parser.add_argument('--beta', type=float, default=1e-3,
                        help='beta')
    parser.add_argument('--c', type=float, default=30.0,
                        help='Constant c.')
    #CausE
    parser.add_argument('--cf_pen', type=float, default=0.05,
                        help='Imbalance loss.')

    #SAM-REG

    parser.add_argument('--rweight', type=float, default=0.05)
    parser.add_argument('--sam',type=bool,default=True)
    parser.add_argument('--pop_test',type=bool,default=False)

    #SimpleX

    parser.add_argument('--w_neg', type=float, default=1)
    parser.add_argument('--neg_margin',type=float, default=0.4)

    #BC_LOSS
    parser.add_argument('--tau1', type=float, default=0.07,
                        help='temperature parameter for L1')
    parser.add_argument('--tau2', type=float, default=0.1,
                        help='temperature parameter for L2')
    parser.add_argument('--w_lambda', type=float, default=0.5,
                        help='weight for combining l1 and l2.')
    parser.add_argument('--freeze_epoch',type=int,default=5)

    #INV-LGN
    parser.add_argument('--inv_tau', type=float, default=1,
                        help='temperature parameter for inv loss in train step')
    parser.add_argument('--adaptive_tau', type=float, default=1,
                        help='temperature parameter for rec loss in adaptive step')
    parser.add_argument('--gumble_tau', type=float, default=10,
                        help='temperature parameter for rec loss in adaptive step')
    parser.add_argument('--mask', type=int, default=0,
                        help='indicator for mask type')
    parser.add_argument('--att_dim', type=int, default=10,
                        help='attention dim')
    parser.add_argument('--pre_epochs', type=int, default=1,
                        help='warmup epochs')
    parser.add_argument('--interval', type=int, default=7,
                        help='normal training epoch before entering adversarial training in each cycle')
    parser.add_argument('--adv_epochs', type=int, default=5,
                        help='adversarial epochs')
    parser.add_argument('--adv_patience', type=int, default=3,
                        help='adversarial patience')
    parser.add_argument('--num_samples', type=int, default=100,
                        help='No. samples in inv loss')
    parser.add_argument('--is_geometric', type=int, default=0,
                        help='using geometric package or not')
    parser.add_argument('--keep_prob', type=float, default=0.8,
                        help='keep prob for dropout')
    parser.add_argument('--embed_tau', type=float, default=0.1,
                        help='tau for embed-level infonce')
    parser.add_argument('--dropout_type', type=int, default=1,
                        help='dropout type for embed-level infonce 0 for random 1 for attention')
    parser.add_argument('--grad_coeff', type=float, default=5,
                        help='coefficient for ARM gradient')
    parser.add_argument('--remove_inv', type=float, default=0,
                        help='0: not remove. 1: remove')
    parser.add_argument('--adv_lr',type=float,default=1e-2, help='lr for adv')

    parser.add_argument('--draw_graph',type=int,default=0, help='draw graph or not')
    parser.add_argument('--use_attribute',type=int,default=0, help='use attribute or not')
    parser.add_argument('--use_mask_inv', type=int, default=0, help='constrain mask or not')
    parser.add_argument('--cluster_coe', type=float, default=10.0, help='cluster coefficient')
    parser.add_argument('--use_new_mask_inv', type=int, default=0, help='constrain mask or not')
    parser.add_argument('--draw_t_sne', type=int, default=0, help='Draw t sne or not')


    # CVIB
    parser.add_argument('--cvib_alpha', type=float, default=0.1,
                        help='alpha for CVIB')
    parser.add_argument('--cvib_gamma', type=float, default=0.01,
                        help='gamma for CVIB')

    # DR
    parser.add_argument('--y_ips_D', type=float, default=0.1,
                        help='y_ips denominator coefficient')


    # CFC
    parser.add_argument('--sample_mask', type=int, default=1,
                        help='use sample mask or not')

    # CDAN
    parser.add_argument('--tau', type=float, default=0.15,
                        help='temparature of infonce loss')
    parser.add_argument('--hidden_size',type=int, default=8)
    parser.add_argument('--bias',type=int, default=0)
    parser.add_argument('--lambda1', type=float, default=1e-5,
                        help='weight for popularity embedding loss')
    parser.add_argument('--lambda2', type=float, default=0,
                        help='weight for dicor loss')
    parser.add_argument('--lambda3', type=float, default=1e-4,
                        help='weight for concat loss')

    #sDRO
    parser.add_argument('--thres1', type = float, default=0.6,
                        help='threshold of percent of popular items in total interactions for niche users')
    parser.add_argument('--thres2', type = float, default=0.8,
                        help='threshold of percent of popular items in total interactions for blockbuster users')
    parser.add_argument('--dro_temperature', type = float, default=0.001,
                        help='A float, temperature of the group re-weighting in DRO. A\
                                suggested range is between [0.001,0.1]')
    parser.add_argument('--str_lr', type = float, default=0.01,
                        help='between [0,1], larger value will let the\
                                estimations of group loss focus more on the current batch.')


    return parser.parse_args()