autogen-watsonx-client

This is an autogen>=0.4 extension for watsonx client integration.

Disclaimer

• This is a community extension for the Autogen project, specifically for the new Autogen >= 0.4 architecture. The goal is to support IBM Watsonx.ai hosted LLMs in the Autogen framework.
• This project is still in a very early stage under development, please create issues in this github repo for bug reports.
• This project is a personal endeavor and is not affiliated with, endorsed by, or connected to any organization/employer in any way. The views, ideas, and opinions expressed in this project are solely my own and do not reflect those of others.

Announcements

• 2025-02-13: create_stream is now implemented!
• 2025-01-13: the version requirement on Autogen has been updated to >=0.4.1, the integration examples have been updated accordingly, see here.

Usage

Prerequisites

• create a python environment with version 3.10 or above
• pip install --upgrade autogen-watsonx-client
• pip install --upgrade autogen-agentchat>=0.4.1
• access to a watsonx.ai instance, setting up environment variables WATSONX_API_KEY, one of WATSONX_SPACE_ID or WATSONX_PROJECT_ID, optionally WATSONX_URL

code snippets

Importing dependencies:

import os

from autogen_agentchat.agents import AssistantAgent
from autogen_agentchat.ui import Console
from autogen_agentchat.conditions import TextMentionTermination
from autogen_agentchat.teams import RoundRobinGroupChat
from autogen_watsonx_client.config import WatsonxClientConfiguration
from autogen_watsonx_client.client import WatsonXChatCompletionClient

Create a watsonx client

wx_config = WatsonxClientConfiguration(
    model_id="meta-llama/llama-3-2-90b-vision-instruct",  # pick a model you have access to on wx.ai here
    api_key=os.environ.get("WATSONX_API_KEY"),
    url=os.environ.get("WATSONX_URL"),
    space_id=os.environ.get("WATSONX_SPACE_ID"),
    project_id=os.environ.get("WATSONX_PROJECT_ID"),
)

wx_client = WatsonXChatCompletionClient(**wx_config)

Define an agent using the watsonx client and register a dummy tool for querying weather

# Define a tool
async def get_weather(city: str) -> str:
    return f"The weather in {city} is 73 degrees and Sunny."


async def main() -> None:
    # Define an agent
    weather_agent = AssistantAgent(
        name="weather_agent",
        model_client=wx_client,
        tools=[get_weather],
    )

    # Define termination condition
    termination = TextMentionTermination("TERMINATE")

    # Define a team
    agent_team = RoundRobinGroupChat([weather_agent], termination_condition=termination)

    # Run the team and stream messages to the console
    stream = agent_team.run_stream(task="What is the weather in New York?")
    await Console(stream)


# NOTE: if running this inside a Python script you'll need to use asyncio.run(main()).
await main()

Refer to here for more detailed examples.