749438fd8b7ffc29e78d042473cb00fda595d937,dataloader.py,,load_dataset,#Any#Any#Any#Any#Any#Any#Any#Any#Any#,13

Before Change

            data_num = len(trainset) // data_split
            ind_start = data_num*split_idx
            ind_end = min(data_num*(split_idx + 1), len(trainset))
            train_indices = indices[ind_start:ind_end]
            train_sampler = torch.utils.data.sampler.SubsetRandomSampler(train_indices)
            train_loader = torch.utils.data.DataLoader(trainset, batch_size=batch_size,
                                                       sampler=train_sampler,
                                                       shuffle=False, num_workers=threads)

After Change

            // Randomly sample indices. Use seed=0 in the generator to make this reproducible
            state = np.random.get_state()
            np.random.seed(0)
            indices = np.random.choice(indices, data_num, replace=False)
            np.random.set_state(state)

            train_sampler = torch.utils.data.sampler.SubsetRandomSampler(indices)
            train_loader = torch.utils.data.DataLoader(trainset, batch_size=batch_size,

Instances