Combine Train and Test data

vision
#1

What is the best way to combine train and test data in torchvision and how can we do it?

Can someone explain how can we do it using ConcatDataset ?

I tried the following:
train = datasets.MNIST(root=dirpath, train=True, download=True, transform=trans)
test = datasets.MNIST(root=dirpath, train=False, download=True, transform=trans)
data_list = list()
data_list.append(train)
data_list.append(test)
total = ConcatDataset(data_list)

But it did not work.

#2

Did you get an error?
Your code should work.
Alternatively try this:

train_dataset = datasets.MNIST(root=dirpath, train=True, transform=trans)
test_dataset = datasets.MNIST(root=dirpath, train=False, transform=trans)

dataset = ConcatDataset([train_dataset, test_dataset])

Both codes should be equivalent, so let me know, if mine works.

2 Likes