Skip to main content

🐔 PromptRefinerNode Module

The PromptRefinerNode module refines the user's initial prompt using the output schema and any additional provided context. It aims to create a more precise prompt that establishes explicit links between elements in the user's original input and their corresponding representations in the JSON schema.

Classes​

PromptRefinerNode​

PromptRefinerNode refines the user's prompt using the output schema and any additional provided context.

Attributes​

  • llm_model: An instance of a language model client, configured for generating answers.
  • verbose (bool): A flag indicating whether to show print statements during execution.
  • additional_info: Additional context to help the language model better understand the request.
  • output_schema: The output schema to be used for prompt refinement.

Methods​

  • __init__(self, input: str, output: List[str], node_config: Optional[dict] = None, node_name: str = "PromptRefiner")

    • Initializes the PromptRefinerNode with necessary configurations.
    • Args:
      • input (str): Boolean expression defining the input keys needed from the state.
      • output (List[str]): List of output keys to be updated in the state.
      • node_config (dict, optional): Additional configuration for the node.
      • node_name (str, optional): The unique identifier name for the node. Defaults to "PromptRefiner".

  • execute(self, state: dict) -> dict

    • Generates a refined prompt using the user's prompt, the schema, and additional context.
    • Args:
      • state (dict): The current state of the graph. The input keys will be used to fetch the correct data from the state.
    • Returns:
      • dict: The updated state with the output key containing the generated refined prompt.
    • Raises:
      • KeyError: If the input keys are not found in the state, indicating that the necessary information for generating an answer is missing.

Example Usage​

Here is an example of how to use the PromptRefinerNode class:

from prompt_refiner_node import PromptRefinerNode

# Define a prompt refiner node
prompt_refier_node = PromptRefinerNode(
    input="user_prompt",
    output=["refined_prompt"]
)

# Define the state
state = {"user_input": "What are the attractions in Chioggia?"}

# Execute the generate answer node
state = prompt_refier_node.execute(state)

# Retrieve the generated answer from the state
refined_prompt = state["refined_prompt"]

print(f"Refined Prompt: {refined_prompt}")