Interface: LLMModel

Defined in: types/llm.ts:81

Properties defining LLMModel.

Properties

audioConfig?

optional audioConfig: AudioConfig

Defined in: types/llm.ts:116

Defines config for audio input modality for multimodal LLMs. capabilities must include 'audio'.

---

capabilities?

optional capabilities: readonly LLMCapability[]

Defined in: types/llm.ts:104

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

Defined in: types/llm.ts:111

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

Defined in: types/llm.ts:86

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

Defined in: types/llm.ts:90

ResourceSource that specifies the location of the model binary.

---

tokenizerConfigSource

tokenizerConfigSource: ResourceSource

Defined in: types/llm.ts:98

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

---

tokenizerSource

tokenizerSource: ResourceSource

Defined in: types/llm.ts:94

ResourceSource pointing to the JSON file which contains the tokenizer.