Using OpenAI function calling

The Poe API allows you to use OpenAI function calling when accessing OpenAI models. In order to use this feature, you will simply need to provide a tools list which contains objects describing your function and an executables list which contains functions that correspond to the tools list. The following is an example.

def get_current_weather(location, unit="fahrenheit"):
    """Get the current weather in a given location"""
    if "tokyo" in location.lower():
        return json.dumps({"location": "Tokyo", "temperature": "11", "unit": unit})
    elif "san francisco" in location.lower():
        return json.dumps(
            {"location": "San Francisco", "temperature": "72", "unit": unit}
        )
    elif "paris" in location.lower():
        return json.dumps({"location": "Paris", "temperature": "22", "unit": unit})
    else:
        return json.dumps({"location": location, "temperature": "unknown"})


tools_executables = [get_current_weather]

tools_dict_list = [
    {
        "type": "function",
        "function": {
            "name": "get_current_weather",
            "description": "Get the current weather in a given location",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {
                        "type": "string",
                        "description": "The city and state, e.g. San Francisco, CA",
                    },
                    "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]},
                },
                "required": ["location"],
            },
        },
    }
]
tools = [fp.ToolDefinition(**tools_dict) for tools_dict in tools_dict_list]

Additionally, you will need to define a dependency of two calls on an OpenAI model of your choice (in this case, the GPT-3.5-Turbo). You need a dependency of two because as part of the OpenAI function calling flow, you need to call OpenAI twice. Adjust this dependency limit if you want to make more than one function calling request while computing your response.

async def get_settings(self, setting: fp.SettingsRequest) -> fp.SettingsResponse:
    return fp.SettingsResponse(server_bot_dependencies={"GPT-3.5-Turbo": 2})

The final code (including the setup code you need to host this on Modal) that goes into your main.py is as follows:

from __future__ import annotations

import json
from typing import AsyncIterable

import fastapi_poe as fp
from modal import Image, Stub, asgi_app


def get_current_weather(location, unit="fahrenheit"):
    """Get the current weather in a given location"""
    if "tokyo" in location.lower():
        return json.dumps({"location": "Tokyo", "temperature": "11", "unit": unit})
    elif "san francisco" in location.lower():
        return json.dumps(
            {"location": "San Francisco", "temperature": "72", "unit": unit}
        )
    elif "paris" in location.lower():
        return json.dumps({"location": "Paris", "temperature": "22", "unit": unit})
    else:
        return json.dumps({"location": location, "temperature": "unknown"})


tools_executables = [get_current_weather]

tools_dict_list = [
    {
        "type": "function",
        "function": {
            "name": "get_current_weather",
            "description": "Get the current weather in a given location",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {
                        "type": "string",
                        "description": "The city and state, e.g. San Francisco, CA",
                    },
                    "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]},
                },
                "required": ["location"],
            },
        },
    }
]
tools = [fp.ToolDefinition(**tools_dict) for tools_dict in tools_dict_list]


class GPT35FunctionCallingBot(fp.PoeBot):
    async def get_response(
        self, request: fp.QueryRequest
    ) -> AsyncIterable[fp.PartialResponse]:
        async for msg in fp.stream_request(
            request,
            "GPT-3.5-Turbo",
            request.access_key,
            tools=tools,
            tool_executables=tools_executables,
        ):
            yield msg

    async def get_settings(self, setting: fp.SettingsRequest) -> fp.SettingsResponse:
        return fp.SettingsResponse(server_bot_dependencies={"GPT-3.5-Turbo": 2})


REQUIREMENTS = ["fastapi-poe==0.0.36"]
image = Image.debian_slim().pip_install(*REQUIREMENTS)
stub = Stub("function-calling-poe")


@stub.function(image=image)
@asgi_app()
def fastapi_app():
    bot = GPT35FunctionCallingBot()
    # Optionally, provide your Poe access key here:
    # 1. You can go to https://poe.com/create_bot?server=1 to generate an access key.
    # 2. We strongly recommend using a key for a production bot to prevent abuse,
    # but the starter examples disable the key check for convenience.
    # 3. You can also store your access key on modal.com and retrieve it in this function
    # by following the instructions at: https://modal.com/docs/guide/secrets
    # POE_ACCESS_KEY = ""
    # app = make_app(bot, access_key=POE_ACCESS_KEY)
    app = fp.make_app(bot, allow_without_key=True)
    return app

To learn how to setup Modal, please follow Steps 1 and 2 in our Quick start. If you already have Modal set up, simply run modal deploy main.py. Modal will then deploy your bot server to the cloud and output the server url. Use that url when creating a server bot on Poe. Once your bot is up, update your bot's settings (one time only after you override get_settings) by following this guide. That's it, your bot is now ready.