Progressive Shrinking
Definition
Stepwise model shrinking trains a supernet for various smaller variants.
A training method in which a supernet is gradually made robust for smaller depths, widths, kernels, or resolutions.
Simply explained
A large model toolkit learns sequentially to cope with fewer or smaller parts.
Why it matters in practice
The term helps to select, develop, evaluate AI systems in a well-founded manner and realistically assess their technical risks.
Example from business
A supernet is trained sequentially for different depths and channel widths.
Limits and pitfalls
- Confusing with ordinary pruning or using subnets without calibration.
Origin & history
Progressive Shrinking is a core method of the Once-for-All network from 2019.
Use in mid-sized companies
- Possible application: A supernet is trained sequentially for different depths and channel widths.
Frequently asked questions
What does Stepwise Model Shrinking mean?
A training method in which a supernet is gradually made robust for smaller depths, widths, kernels, or resolutions. What mistakes should be avoided? Confusing with ordinary pruning or using subnets without calibration.
Sources
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: