What is torch.stack vs torch.cat?

Updated May 17, 2026

Short answer

stack adds a new dimension; cat concatenates along existing dimension.

Deep explanation

stack increases rank, cat preserves rank but extends size.

Real-world example

Used in batching tensors or merging features.

Common mistakes

Confusing output shapes.

Follow-up questions

When to use stack?
Does cat require same shape?

More PyTorch interview questions