yAwareContrastiveLearning/config.py at main · Junebeomstics/yAwareContrastiveLearning · 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
class Config:

    def __init__(self, args):

        #args에 있는 속성들을 config에 등록
        for name, value in vars(args).items():
            setattr(self, name, value)

        # self.mode = mode
        # self.framework = args.framework
        # self.world_size = args.world_size
        # self.rank = args.rank
        # self.dist_backend = args.dist_backend
        # self.local_rank = args.local_rank
        # self.batch_size = args.batch_size
        # self.lr_policy = args.lr_policy
        # self.lr_decay_iters = args.lr_decay_iters
        # self.gamma = args.gamma


        if self.mode == 'pretraining':
            self.nb_epochs_per_saving = 1 # 몇 에포크에 한 번씩 저장할 것인가
            self.pin_mem = True
            self.num_cpu_workers = 8
            self.cuda = True

            # Optimizer
            self.weight_decay = 5e-5

            # Hyperparameters for our y-Aware InfoNCE Loss
            self.temperature = 0.1

            self.model = "DenseNet"

            self.checkpoint_dir = args.ckpt_dir
            self.tb_dir = args.tb_dir
            self.tf = args.tf
            self.label_name = args.label_name

            self.data = '/global/cfs/cdirs/m3898/2.UKB/1.sMRI_fs_cropped'
            self.label = '/global/cfs/cdirs/m3898/2.UKB/2.demo_qc/UKB_phenotype_gps_included.csv' #'/global/cfs/cdirs/m3898/2.UKB/2.demo_qc/UKB_phenotype.csv'

            self.val_size = 0.1
            # Paths to the data
#             self.data_train = "/path/to/your/training/data.npy"
#             self.label_train = "/path/to/your/training/metadata.csv"

#             self.data_val = "/path/to/your/validation/data.npy"
#             self.label_val = "/path/to/your/validation/metadata.csv"

            self.input_size = (1, 80, 80, 80) #(1, 121, 145, 121)

            self.checkpoint_dir = args.ckpt_dir

            self.train_continue = "on"

        elif self.mode == 'finetuning':
            ## We assume a classification task here
            self.batch_size = 8
            self.nb_epochs_per_saving = 10
            self.pin_mem = True
            self.num_cpu_workers = 1
            self.nb_epochs = 100
            self.cuda = True
            # Optimizer
            self.lr = 1e-4
            self.weight_decay = 5e-5

            self.pretrained_path = "/path/to/model.pth"
            self.num_classes = 2
            self.model = "DenseNet"