magic-decorator 0.0.8

A magic function decorator using OpenAI ChatCompletion

Simple demo function using OpenAI's ChatComplete

The only function implemented in this module is magic.

magic works as a decorator. It simply sends the function's signature and docstring to the GPT and expects to get a result.

It can be installed with pip install magic-decorator and used with from magic_decorator import magic.

from magic_decorator import magic
@magic(model="gpt-4")
def transpose(matrix: list[list[int]]) -> list[list[int]]:
    """
    This function finds the transpose of the given matrix.
    """
transpose([[1,2,3],[4,5,6]])

[[1, 4], [2, 5], [3, 6]]

from magic_decorator import magic
@magic(model="gpt-4")
def add(a: float, b: float) -> float:
    """
    This function adds a and b.
    """
add(1., 10.)

11.0

from magic_decorator import magic
@magic(model="gpt-4")
def friend(message: str) -> str:
    """
    This function responds like a friend.
    """
friend("Hi, how are you today?")

'Hey! I am doing great, thank you. How about you?'

Since the magic decorator works with OpenAI's API, you need to put the API_KEY in an environment variable or directly in the decorator.

import os
from magic_decorator import magic

os.environ["OPENAI_API_KEY"] = ...
# or
@magic(model="gpt-4", api_key=...)
def func():
    ...

Alternatively, you can pre-declare decorators that will be used repeatedly.

cold_magic = magic(model="gpt-4", temperature=0)
@cold_magic
def func():
    ...

Added a decorator for langchain. In this case, it becomes LLMChain.

llm = SomeLLMModels()
@magic_langchain(llm=llm)
def mychain(arg1: int, arg2: str):
    ...

mychain.predict(arg1=2, arg2="hi")

There is no preprocessing, no postprocessing, and no error handling in this module. It's just code to do a simple thing, so there are no prompts to encourage better answers.