CloneMode

Enums · AiDotNet.Models

How much of the original’s storage a copy is allowed to share.

For Beginners

the difference is what happens when you train the copy. With Deep and CopyOnWrite, nothing happens to the original. With Shared, you train both, because there is only one set of weights.

How It Works

Orthogonal to the rest of CloneOptions. Those decide WHAT a copy carries — parameters, optimizer state, buffers; this decides whether what it carries is copied or shared.

Fields

Field Summary
CloneMode CopyOnWrite The copy shares each weight tensor’s storage until either side writes to it.
CloneMode Deep The copy owns its own storage.
CloneMode Shared The copy points at the SAME parameters as the original.
int value__