spin_sdk.llm API documentation (original) (raw)

Module spin_sdk.llm

Module for working with the Spin large language model API

def generate_embeddings(model: str, text: Sequence[str]) ‑> [EmbeddingsResult](wit/imports/llm.html#spin%5Fsdk.wit.imports.llm.EmbeddingsResult "spin_sdk.wit.imports.llm.EmbeddingsResult")

def infer(model: str, prompt: str) ‑> [InferencingResult](wit/imports/llm.html#spin%5Fsdk.wit.imports.llm.InferencingResult "spin_sdk.wit.imports.llm.InferencingResult")

def infer_with_options(model: str, prompt: str, options: [InferencingParams](#spin%5Fsdk.llm.InferencingParams "spin_sdk.llm.InferencingParams") | None) ‑> [InferencingResult](wit/imports/llm.html#spin%5Fsdk.wit.imports.llm.InferencingResult "spin_sdk.wit.imports.llm.InferencingResult")

class InferencingParams (max_tokens: int = 100, repeat_penalty: float = 1.1, repeat_penalty_last_n_token_count: int = 64, temperature: float = 0.8, top_k: int = 40, top_p: float = 0.9)

InferencingParams(max_tokens: int = 100, repeat_penalty: float = 1.1, repeat_penalty_last_n_token_count: int = 64, temperature: float = 0.8, top_k: int = 40, top_p: float = 0.9)

Expand source code

@dataclass
class InferencingParams:
    max_tokens: int = 100
    repeat_penalty: float = 1.1
    repeat_penalty_last_n_token_count: int = 64
    temperature: float = 0.8
    top_k: int = 40
    top_p: float = 0.9

Class variables

var max_tokens : int

var repeat_penalty : float

var repeat_penalty_last_n_token_count : int

var temperature : float

var top_k : int

var top_p : float