seniorPyTorch

What is activation recomputation vs activation offloading?

Updated May 17, 2026

Short answer

Recomputation recalculates activations; offloading moves them to CPU memory.

Deep explanation

Checkpointing recomputes forward activations during backward pass, while offloading stores activations on CPU RAM and transfers them back when needed. Offloading reduces GPU memory but introduces transfer latency.

Unlock with a Pro subscription to view this section.

View pricing

Real-world example

No real-world example available yet.

Unlock with a Pro subscription to view this section.

Upgrade to Pro

Common mistakes

No common mistakes listed yet.

Unlock with a Pro subscription to view this section.

Upgrade to Pro

Follow-up questions

No follow-up questions available yet.

Unlock with a Pro subscription to view this section.

Upgrade to Pro

More PyTorch interview questions

View all →