checkLoad abstract method
Throws LlamaUnsupportedException when loading a model in format with
params would ignore an explicitly set field, or needs a runtime or
device this platform lacks; the message names the missing capability
and the platform or runtime condition.
format is null when only the model file can tell it; the backend then
checks again when it loads the file, and the engine ignores the result.
Otherwise returns null when the backend checked the load, or why it
cannot tell before loading, such as a runtime it did not build. Must not
start a runtime or read a file.
Implementation
String? checkLoad(ModelParams params, ModelFormat? format);