​
Overview

The BaseGenerator class provides an abstract base for implementing generation tasks. It defines a common interface for generating responses from language models using prompts.

​
Methods

generate
method

Abstract method to generate a response from the model.

Parameters:

  • prompt (Prompt): The prompt to generate a response for.
  • inputs (Dict[str, Any], optional): The inputs to the prompt. Defaults to .

Returns: Union[str, Dict[str, Any], Awaitable[Union[str, Dict[str, Any]]]]: The generated response.

__call__
method

Unified method to compute the metric, handling both sync and async implementations.

Parameters:

  • prompt (Prompt): The prompt to generate a response for.
  • inputs (Dict[str, Any], optional): The inputs to the prompt. Defaults to .

Returns: Union[str, Dict[str, Any], Awaitable[Union[str, Dict[str, Any]]]]: The generated response.

​
Usage

To use the BaseGenerator class, create a subclass and implement the generate method:

from ape.common import BaseGenerator

class MyGenerate(BaseGenerator):
    def generate(self, prompt: Prompt, inputs: Dict[str, Any] = {}) -> Union[str, Dict[str, Any]]:
        # Implement your generation logic here
        return "Hello, world!"

Was this page helpful?

DatasetItemGenerator