Skip to main content
A LoRA adapter specialises a base model without swapping it out. Apply one, generate, remove it. The base model stays loaded throughout, so switching behaviour costs far less than switching models.

The namespace

Scale

scale controls how strongly the adapter pulls the base model. Leave it unset for the adapter’s own default. Lower values blend the adapter with base behaviour; higher values commit to it. If output degrades after applying an adapter, scale is the first thing to turn down.

Removing

Removal is keyed by adapter id.

Switching adapters

Adapters stack. Remove before applying another, or two of them fight and the output is worse than either alone.

The base model has to support it

An adapter is trained against a specific base model. Applying one to a different base either fails or produces nonsense. Check the model’s supportsLora flag before offering adapters in your interface.

Catalog management

Swift carries the full catalog surface beyond the four short-form methods:
adaptersForModel is the one to build an interface around: it answers “what can I apply to what is loaded right now”.