How does one obtain indicies from a dataloader?

[ptrblck]

In the thread you posted is a valid solution:

How to retrieve the sample indices of a mini-batch

One way to do this is to implement a subclass of torch.utils.data.Dataset that returns a triple (data, target, index) from its __getitem__ method. Then your loop would be: for data, target, index in train_loader: ....

Here is a small example using CIFAR10 from the other thread:

How to retrieve the sample indices of a mini-batch

You could adapt this code: from torchvision import datasets, transforms from torch.utils.data import Dataset, DataLoader class MyDataset(Dataset): def __init__(self): self.cifar10 = datasets.CIFAR10(root='YOUR_PATH', download=False, train=True, transform=transforms.ToTensor()) def __getitem__(self, index): data, target = self.cifar10[index] …