LingoQA dataset for pytorch

How to use

from lingoqa_dataset.lingoqa_dataset import LingoQADataset, DatasetType
import torchvision.transforms as transforms
from torch.utils.data import DataLoader

dataset = LingoQADataset(
    DatasetType.EVALUATION, transforms=transforms.Resize((256, 512))
)
dataloader = DataLoader(dataset=dataset, batch_size=3, shuffle=True)
for data, question, answer in dataloader:
    pass

data

• type: torch.Tensor
• size : torch.Size([batch_size, 3 * number_of_images, height, width])
• description : Images in the target sequences.

question

• type: torch.Tuple(str)
• size: batch_size
• description : Questions in the batch.

answer

• type: torch.Tuple(str)
• size: batch_size
• description : Answers in the batch.

Special thanks

LingoQA project from wayve.