Dataset and Dataloader

Dataset

PyTorch supports two different types of datasets.

map-style datasets

• Represents a map from (possibly non-integral) indices/keys to data samples.
• Implements the getitem() and len() protocols, and
• When accessed with dataset[idx], could read the idx-th image and its corresponding label from a folder on the disk.

iterable-style datasets

• Represents an iterable over data samples.
• Implements the iter() protocol, and
• This type of datasets is particularly suitable for cases where random reads are expensive or even improbable, and where the batch size depends on the fetched data.

Data Loading Order and Sampler Loading Batched and Non-Batched Data collate_fn Single- and Multi-process Data Loading

Reference link