出现 ValueError: expected sequence of length 61 at dim 1 (got 16) ，怎么解决？

源程序是在 “把数据转换成torch的tensor” 那里出错的，
错误提示如下

Traceback (most recent call last):
  File "C:/Users/12254/Desktop/User-Difference-Attention-main - 1/uda_main.py", line 143, in <module>
    args.batch_size), epoch, args, 'group')
  File "C:\Users\12254\Desktop\User-Difference-Attention-main - 1\dataset.py", line 128, in get_group_dataloader
    train_data = TensorDataset(torch.tensor(group).to(device), torch.tensor(group_members).to(device),
ValueError: expected sequence of length 61 at dim 1 (got 16)

错误代码如下：

 for epoch in range(args.epoch):
        # training
        agree.train()
        t1 = time()
        training(agree, dataset.get_user_dataloader(
            args.batch_size), epoch, args, 'user')
        training(agree, dataset.get_group_dataloader(
            args.batch_size), epoch, args, 'group')
        print("user and group training time is: [%.1f s]" % (time() - t1))
        t2 = time()

    def get_group_dataloader(self, batch_size):
        group, positem_negitem_at_g = self.get_train_instances(self.group_trainMatrix)
        group_members = []
        for gid in group:
            group_members.append(self.g_m_d[gid])
        train_data = TensorDataset(torch.tensor(group).to(device), torch.tensor(group_members).to(device),
                                   torch.tensor(positem_negitem_at_g).to(device))
        group_train_loader = DataLoader(train_data, batch_size=batch_size, shuffle=True)
        return group_train_loader