oai.client_utils

Utilities for client classes

validate_parameter

def validate_parameter(params: Dict[str, Any], param_name: str,
                       allowed_types: Tuple, allow_None: bool,
                       default_value: Any, numerical_bound: Tuple,
                       allowed_values: list) -> Any

Validates a given config parameter, checking its type, values, and setting defaults

Arguments:

• params Dict[str, Any] - Dictionary containing parameters to validate.
• param_name str - The name of the parameter to validate.
• allowed_types Tuple - Tuple of acceptable types for the parameter.
• allow_None bool - Whether the parameter can be None.
• default_value Any - The default value to use if the parameter is invalid or missing. numerical_bound (Optional[Tuple[Optional[float], Optional[float]]]): A tuple specifying the lower and upper bounds for numerical parameters. Each bound can be None if not applicable.
• allowed_values Optional[List[Any]] - A list of acceptable values for the parameter. Can be None if no specific values are required.

Returns:

• Any - The validated parameter value or the default value if validation fails.

Raises:

• TypeError - If allowed_values is provided but is not a list.

  Example Usage:

    # Validating a numerical parameter within specific bounds
    params = {"temperature": 0.5, "safety_model": "Meta-Llama/Llama-Guard-7b"}
    temperature = validate_parameter(params, "temperature", (int, float), True, 0.7, (0, 1), None)
    # Result: 0.5

    # Validating a parameter that can be one of a list of allowed values
    model = validate_parameter(
    params, "safety_model", str, True, None, None, ["Meta-Llama/Llama-Guard-7b", "Meta-Llama/Llama-Guard-13b"]
    )
    # If "safety_model" is missing or invalid in params, defaults to "default"

should_hide_tools

def should_hide_tools(messages: List[Dict[str, Any]], tools: List[Dict[str,
                                                                       Any]],
                      hide_tools_param: str) -> bool

Determines if tools should be hidden. This function is used to hide tools when they have been run, minimising the chance of the LLM choosing them when they shouldn't.

Arguments:

• messages List[Dict[str, Any]] - List of messages
• tools List[Dict[str, Any]] - List of tools
• hide_tools_param str - "hide_tools" parameter value. Can be "if_all_run" (hide tools if all tools have been run), "if_any_run" (hide tools if any of the tools have been run), "never" (never hide tools). Default is "never".

Returns:

• bool - Indicates whether the tools should be excluded from the response create request

  Example Usage:

    # Validating a numerical parameter within specific bounds
    messages = params.get("messages", [])
    tools = params.get("tools", None)
    hide_tools = should_hide_tools(messages, tools, params["hide_tools"])