Interface: LLMProps

Defined in: types/llm.ts:69

Properties for initializing and configuring a Large Language Model (LLM) instance.

Properties

model

model: object

Defined in: types/llm.ts:70

capabilities?

optional capabilities: readonly "vision"[]

Optional list of modality capabilities the model supports. Determines the type of the media argument in sendMessage. Example: ['vision'] enables sendMessage(text, { imagePath }).

generationConfig?

optional generationConfig: GenerationConfig

Recommended default generation settings, typically copied from the upstream generation_config.json or the model card. Applied automatically after the native module loads and before any user configure() call, so callers only need to override the values they want to change.

modelName

modelName: LLMModelName

The built-in model name (e.g. 'llama-3.2-3b'). Used for telemetry and hook reload triggers. Pass one of the pre-built LLM constants (e.g. LLAMA3_2_3B) to populate all required fields.

modelSource

modelSource: ResourceSource

ResourceSource that specifies the location of the model binary.

tokenizerConfigSource

tokenizerConfigSource: ResourceSource

ResourceSource pointing to the JSON file which contains the tokenizer config.

tokenizerSource

tokenizerSource: ResourceSource

ResourceSource pointing to the JSON file which contains the tokenizer.

---

preventLoad?

optional preventLoad: boolean

Defined in: types/llm.ts:105

Boolean that can prevent automatic model loading (and downloading the data if you load it for the first time) after running the hook.