vLLM, LMStudio, HuggingFace

We can leverage the OpenAI instrumentation to log calls from inference servers that use messages-based API, such as vLLM, LMStudio or HuggingFace’s TGI.

Create a new Python file named app.py in your project directory. This file will contain the main logic for your LLM application.

In app.py, import the necessary packages and define one function to handle messages incoming from the UI.

from openai import AsyncOpenAI
import chainlit as cl
client = AsyncOpenAI(base_url="http://localhost:1234/v1", api_key="lm-studio")
# Instrument the OpenAI client
cl.instrument_openai()

settings = {
    "model": "gpt-3.5-turbo",
    "temperature": 0,
    # ... more settings
}

@cl.on_message
async def on_message(message: cl.Message):
    response = await client.chat.completions.create(
        messages=[
            {
                "content": "You are a helpful bot, you always reply in Spanish",
                "role": "system"
            },
            {
                "content": input,
                "role": "user"
            }
        ],
        **settings
    )
    await cl.Message(content=response.choices[0].message.content).send()

Create a file named .env in the same folder as your app.py file. Add your OpenAI API key in the OPENAI_API_KEY variable. You can optionally add your Literal AI API key in the LITERAL_API_KEY.

To start your app, open a terminal and navigate to the directory containing app.py. Then run the following command:

chainlit run app.py -w

The -w flag tells Chainlit to enable auto-reloading, so you don’t need to restart the server every time you make changes to your application. Your chatbot UI should now be accessible at http://localhost:8000.