1

Is it possible to split a dataloader object of training dataset into training and validation dataloader?

from torch.utils.data import DataLoader
from torchvision import datasets, transforms

train_dataset = datasets.ImageFolder(train_data_directory, transform=transforms.ToTensor())
# Data loader
train_loader = DataLoader(train_dataset, batch_size=100, shuffle=True)

Now I would like to split a train_loader to train and validation dataloader.

Rajendra Sapkota
  • 13
  • 1
  • 5

1 Answers1

3

Look at random_split in torch.utils.data. It will handle a random Dataset split (you have to split before creating the DataLoader, not after).

Adam Kern
  • 566
  • 4
  • 14
  • @RajendraSapkota If this answers your question then please mark the question as accepted. – jodag Nov 03 '20 at 21:11