Skip to content

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 

Repository files navigation

openai-function-calling

Is a simple library which adds convenience methods to query open ai's chat endpoint with support for function calling.

Installation

pip install openai-function-calling-aeolin

Basic Usage Example

from typing import Any
from function_calling import setup_api, get_chat_completion, ChatHistory, ChatRole, description
import random

def to_string(obj: Any) -> str:
    return str(obj) 

setup_api("openai api key", to_string) 

@description("Generates a password of the given length") 
def generate_password(length: int) -> str:
    return "".join([random.choice("123456789abcdefghijklmnopqrstuvwxyz") for x in range(length)])

functions = [generate_password] 
history = ChatHistory("gpt-3.5-turbo",4096,functions)
history.add_message(ChatRole.SYSTEM,"You're PasswordBot an ai that can generate passwords for people") 
history.add_message(ChatRole.USER, "can you generate me a password of length 16?")
response = get_chat_completion(history) 
print(response.content)

setup_api

The setup api function takes two parameters, the first one is your openai api key, the second one is a function which can be used to convert objects to the type string. The string function is used by the library when ever a callable function returns something which is not of type string

ChatHistory

The ChatHistory will keep track of all messages and function responses. With the last parameter of add_message a function can be marked as ephemeral. Per default all messages which are not of the role SYSTEM are ephemeral. Ephemeral messages can and will be removed by the library if the context runs out of tokens.

get_chat_completion

This function take a ChatHistory object as an argument and queries the openai chat endpoint with it. The resulting function responses and messages are automatically recorded in the history itself. Upon completion a ChatMessage object is returned with the assitant response. The message is stored in the field content inside the ChatMessage The second parameter is an optional function which receives a list of responses if N was greated than 1 and returns one of the responses. If this function is not supplied the first response will be taken by default The following optional parameters are passed through to the openai library (parameter name: default value)

  • function_call: auto
  • max_tokens: 1024
  • temperature: 0.9
  • top_p: 1.0
  • n: 1

About

No description, website, or topics provided.

Resources

Stars

2 stars

Watchers

1 watching

Forks

Releases

Packages

Used by

Contributors

Languages