Function dataloader

pairwise/dataloader.py:28–68

should be starting from X, and Y X_{}_trainval, X_{}_test, Y_{}_trainval, Y_{}_test

(*args,**kwargs)

# def dataloader(X_train_val_set, X_test_set, Y_train_val_set, Y_test_set):
def dataloader(*args,**kwargs):
    """
        should be starting from X, and Y
        X_{}_trainval, X_{}_test, Y_{}_trainval, Y_{}_test
    """
    temp_loader = {}
    for name, input in kwargs.items():
    # First, format
        input_name = name.split('_')
        if input_name[0].startswith('Y'):
            input = input.astype('float32')
            input = torch.from_numpy(input)
            input = input.unsqueeze(1)
            temp_loader[name] = input

        elif input_name[0].startswith('X'):
            #input is tabular format
            input = input.astype('float32')
            input = torch.from_numpy(input)
            temp_loader[name] = input
        
        elif input_name[0].startswith('dummy'):
            input = input.astype('float32')
            input = torch.from_numpy(input)
            temp_loader[name] = input
    
    # Second, to tensordataset
    temp_loader_trainval,  temp_loader_test = [], []
    for key, val in temp_loader.items():
        load_name = key.split('_')
        # train_val_dataset
        if load_name[-1].endswith('trainval'):
            temp_loader_trainval.append(val)
        elif load_name[-1].endswith('test'):
            temp_loader_test.append(val)

    train_val_dataset = torch.utils.data.TensorDataset(*temp_loader_trainval)
    test_dataset = torch.utils.data.TensorDataset(*temp_loader_test)
    test_loader = DataLoader(test_dataset, batch_size=256,shuffle = False, sampler=sampler.SequentialSampler(test_dataset))
    #list(BatchSampler(SequentialSampler(range(10)), batch_size=3, drop_last=False))
    return train_val_dataset, test_loader


def dataloader_graph(*args,**kwargs):