Story Generator

A Python application that generates creative story continuations with configurable parameters using a language model.

Capabilities

Initialize model for text generation

• Creates a text generation service that can accept prompts and return generated text @test

Generate story with default parameters

• Given the prompt "Once upon a time in a magical forest,", the service generates a continuation of at least 50 characters @test
• Given an empty prompt "", the service raises a ValueError @test

Generate story with temperature control

• Given the prompt "The robot said:" with temperature 0.0, generates a deterministic response (same output on repeated calls) @test
• Given the prompt "The robot said:" with temperature 0.8, generates more creative responses (different outputs on repeated calls) @test

Generate multiple story variations

• Given a prompt and requesting 3 variations, returns a list of exactly 3 distinct story continuations @test

Implementation

@generates

API

class StoryGenerator:
    """
    A service for generating story continuations using an LLM.
    """

    def __init__(self, model_name: str = "facebook/opt-125m"):
        """
        Initialize the story generator with a specified model.

        Args:
            model_name: The name or path of the language model to use
        """
        pass

    def generate(
        self,
        prompt: str,
        max_tokens: int = 100,
        temperature: float = 0.7,
        num_variations: int = 1
    ) -> list[str]:
        """
        Generate story continuation(s) based on the given prompt.

        Args:
            prompt: The starting text for the story
            max_tokens: Maximum number of tokens to generate
            temperature: Controls randomness (0.0 = deterministic, 1.0 = very random)
            num_variations: Number of different continuations to generate

        Returns:
            A list of generated story continuations

        Raises:
            ValueError: If prompt is empty
        """
        pass

Dependencies { .dependencies }

vllm { .dependency }

Provides high-performance inference and text generation capabilities.

vllm>=0.6.0