hashbrown

ExperimentalWebLlmTransportOptions

Configuration for application-owned local WebLLM inference.

API

interface ExperimentalWebLlmTransportOptions  {
engine:  WebLlmEngine; 
media?: never;
model?: string;
}
engine

Loaded engine; Hashbrown never loads or unloads its model.

media
never

Image input requires the image-specific engine contract and overload.

model
string

Selects a model when the injected engine has multiple models loaded.