Prompt
Utility class for storing and managing prompts for language models.
Overview
The Prompt class is an extension of the Prompt class from the promptfile library. It provides additional functionality for managing prompts, including response formats, few-shot examples, and cost tracking.
Attributes
The name of the language model to use for this prompt.
A list of messages that make up the prompt. Each message is a dictionary containing role and content.
Specifies the desired format for the model’s response. Can be a dictionary or a custom pydantic BaseModel.
A list of few-shot examples to be injected into the prompt template.
Controls the randomness of the model’s output. Lower values make the output more deterministic.
An optional name for the prompt, useful for identification and cost tracking.
A dictionary describing the expected input parameters for the prompt.
A dictionary describing the expected output format from the model.
Methods
Asynchronously calls the language model with the configured prompt.
Parameters:
lm_config: Optional dictionary for additional language model configuration.num_retries: Number of retries for API calls (default: 3).**kwargs: Additional keyword arguments for formatting the prompt.
Returns: The response from the language model, either as a string or a dictionary.
Loads a Prompt object from a string content.
Parameters:
content: The string content to load the prompt from.
Returns: A new Prompt object.
Loads a Prompt object from a file.
Parameters:
file_path: The path to the file to load the prompt from.
Returns: A new Prompt object.
Formats the prompt with given keyword arguments.
Parameters:
**kwargs: Keyword arguments to format the prompt with.
Returns: The formatted Prompt object.
Creates a reset copy of the Prompt object with empty few-shot examples.
Returns: A new Prompt object with reset few-shot examples.
Dumps the Prompt object to a string representation.
Returns: A string representation of the Prompt object.
Usage
import textwrap
from pydantic import BaseModel
from ape.common import Prompt
class Response(BaseModel):
reasoning: str
answer: str
base_prompt = Prompt(
messages=[
{
"role": "system",
"content": "You are a helpful assistant."
},
{
"role": "user",
"content": "Answer the following question: {question}"
}
],
model="gpt-4o",
response_format=Response
)
response: Response = await prompt(
question="What is the meaning of life?"
)
Was this page helpful?