Skip to content

Latest commit

 

History

History
58 lines (36 loc) · 1.39 KB

README.md

File metadata and controls

58 lines (36 loc) · 1.39 KB
Raw

Lazer

Lazer is a Python library that provides a convenient way to expose Python functions as schemas for OpenAI chat models.

For those not in the know: OpenAI - Function Calling

What it Does

Allows you to more easily inform GPT about functions in your code 😊

Warning don't RCE yourself lol

Installation

To install Lazer, simply run:

pip install lazer

Usage

Here is an example of how to use Lazer:

from lazer import Lazer, LazerConversation

lazer = Lazer()

# GPT is now made aware of your function `qux`
@lazer.use
def qux(num: int, name: str) -> str:
    """Retrieve a number and a name from the user and compute the qux of it"""
    return str(num + len(name) * 13)

conversation = LazerConversation(lazer, {"model": "gpt-3.5-turbo-0613"})
response = conversation.talk("What is the qux of 3 and steven")
print(response)
# ... 81 ...

Note You do not have to use the LazerConversation GPT frontend, you can simply use the Lazer functions .use (decorator) and .get_functions() as well as .dispatch() to build your own GPT frontend utilizing Lazer.

Note If you are using the LazerConversation frontend be sure to set an OPENAI_API_KEY in a .env file somewhere in your directory.

Demo

Go to demo/ for some demo code.

Authors:

  • @JustinStitt
  • @diamondburned