DataLoader is used to efficiently load data in batches during training. Why it is needed: Loads data in batches (not one-by-one) Speeds up training using parallel workers Supports shuffling for better generalization Handles large datasets that don’t fit in memory Without DataLoader, training would be slow and manual.
