Module spin_sdk.wit.imports.llm

A WASI interface dedicated to performing inferencing for Large Language Models.

Global variables

var Error

The set of errors which may be raised by functions in this interface

Functions

def generate_embeddings(model: str, text: List[str]) ‑> EmbeddingsResult

Expand source code

def generate_embeddings(model: str, text: List[str]) -> EmbeddingsResult:
    """
    Generate embeddings for the supplied list of text
    
    Raises: `spin_sdk.wit.types.Err(spin_sdk.wit.imports.llm.Error)`
    """
    raise NotImplementedError

Generate embeddings for the supplied list of text

Raises: Err(Error)

def infer(model: str,
prompt: str,
params: InferencingParams | None) ‑> InferencingResult

Expand source code

def infer(model: str, prompt: str, params: Optional[InferencingParams]) -> InferencingResult:
    """
    Perform inferencing using the provided model and prompt with the given optional params
    
    Raises: `spin_sdk.wit.types.Err(spin_sdk.wit.imports.llm.Error)`
    """
    raise NotImplementedError

Perform inferencing using the provided model and prompt with the given optional params

Raises: Err(Error)

Classes

class EmbeddingsResult (embeddings: List[List[float]],
usage: EmbeddingsUsage)

Expand source code

@dataclass
class EmbeddingsResult:
    """
    Result of generating embeddings
    """
    embeddings: List[List[float]]
    usage: EmbeddingsUsage

Result of generating embeddings

Instance variables

var embeddings : List[List[float]]

var usage : EmbeddingsUsage

class EmbeddingsUsage (prompt_token_count: int)

Expand source code

@dataclass
class EmbeddingsUsage:
    """
    Usage related to an embeddings generation request
    """
    prompt_token_count: int

Usage related to an embeddings generation request

Instance variables

var prompt_token_count : int

class Error_InvalidInput (value: str)

Expand source code

@dataclass
class Error_InvalidInput:
    value: str

Error_InvalidInput(value: str)

Instance variables

var value : str

class Error_ModelNotSupported

Expand source code

@dataclass
class Error_ModelNotSupported:
    pass

Error_ModelNotSupported()

class Error_RuntimeError (value: str)

Expand source code

@dataclass
class Error_RuntimeError:
    value: str

Error_RuntimeError(value: str)

Instance variables

var value : str

class InferencingParams (max_tokens: int,
repeat_penalty: float,
repeat_penalty_last_n_token_count: int,
temperature: float,
top_k: int,
top_p: float)

Expand source code

@dataclass
class InferencingParams:
    """
    Inference request parameters
    """
    max_tokens: int
    repeat_penalty: float
    repeat_penalty_last_n_token_count: int
    temperature: float
    top_k: int
    top_p: float

Inference request parameters

Instance 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

class InferencingResult (text: str,
usage: InferencingUsage)

Expand source code

@dataclass
class InferencingResult:
    """
    An inferencing result
    """
    text: str
    usage: InferencingUsage

An inferencing result

Instance variables

var text : str

var usage : InferencingUsage

class InferencingUsage (prompt_token_count: int, generated_token_count: int)

Expand source code

@dataclass
class InferencingUsage:
    """
    Usage information related to the inferencing result
    """
    prompt_token_count: int
    generated_token_count: int

Usage information related to the inferencing result

Instance variables

var generated_token_count : int

var prompt_token_count : int