3.3 KiB

Raw Blame History

Zephyr Documentation

Introduction

Welcome to the documentation for Zephyr, a language model by Hugging Face designed for text generation tasks. Zephyr is capable of generating text in response to prompts and is highly customizable using various parameters. This document will provide you with a detailed understanding of Zephyr, its purpose, and how to effectively use it in your projects.

Overview

Zephyr is a text generation model that can be used to generate human-like text based on a given prompt. It utilizes the power of transformers and fine-tuning to create coherent and contextually relevant text. Users can control the generated text's characteristics through parameters such as temperature, top_k, top_p, and max_new_tokens.

Class Definition

class Zephyr:
    def __init__(
        self,
        max_new_tokens: int = 300,
        temperature: float = 0.5,
        top_k: float = 50,
        top_p: float = 0.95,
    ):
        """
        Initialize the Zephyr model.

        Args:
            max_new_tokens (int): The maximum number of tokens in the generated text.
            temperature (float): The temperature parameter, controlling the randomness of the output.
            top_k (float): The top-k parameter, limiting the vocabulary used in generation.
            top_p (float): The top-p parameter, controlling the diversity of the output.
        """

Parameters

max_new_tokens (int): The maximum number of tokens in the generated text.
temperature (float): The temperature parameter, controlling the randomness of the output.
top_k (float): The top-k parameter, limiting the vocabulary used in generation.
top_p (float): The top-p parameter, controlling the diversity of the output.

Usage

To use the Zephyr model, follow these steps:

Initialize the Zephyr model with your desired parameters:

from swarms.models import Zephyr
model = Zephyr(max_new_tokens=300, temperature=0.7, top_k=50, top_p=0.95)

Generate text by providing a prompt:

output = model("Generate a funny joke about cats")
print(output)

Example 1 - Generating a Joke

model = Zephyr(max_new_tokens=100)
output = model("Tell me a joke about programmers")
print(output)

Example 2 - Writing Poetry

model = Zephyr(temperature=0.2, top_k=30)
output = model("Write a short poem about the moon")
print(output)

Example 3 - Asking for Advice

model = Zephyr(temperature=0.8, top_p=0.9)
output = model("Give me advice on starting a healthy lifestyle")
print(output)

Additional Information

Zephyr is based on the Hugging Face Transformers library and uses the "HuggingFaceH4/zephyr-7b-alpha" model.
The generated text can vary based on the values of temperature, top_k, and top_p. Experiment with these parameters to achieve the desired output.
The max_new_tokens parameter can be adjusted to control the length of the generated text.
You can integrate Zephyr into chat applications, creative writing projects, or any task that involves generating human-like text.

That concludes the documentation for Zephyr. We hope you find this model useful for your text generation needs! If you have any questions or encounter any issues, please refer to the Hugging Face Transformers documentation for further assistance. Happy text generation!

3.3 KiB Raw Blame History