MLflow Typescript SDK - OpenAI

Seamlessly integrate MLflow Tracing with OpenAI to automatically trace your OpenAI API calls.

| Package | NPM | Description | | -------------------- | --------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------- | | @mlflow/openai | | Auto-instrumentation integration for OpenAI. |

Installation

npm install @mlflow/openai

The package includes the @mlflow/core package and openai package as peer dependencies. Depending on your package manager, you may need to install these two packages separately.

Quickstart

Start MLflow Tracking Server. If you have a local Python environment, you can run the following command:

pip install mlflow
mlflow server --backend-store-uri sqlite:///mlruns.db --port 5000

If you don't have Python environment locally, MLflow also supports Docker deployment or managed services. See Self-Hosting Guide for getting started.

Instantiate MLflow SDK in your application:

import * as mlflow from '@mlflow/core';

mlflow.init({
  trackingUri: 'http://localhost:5000',
  experimentId: '<experiment-id>',
});

Create a trace:

import { OpenAI } from 'openai';
import { tracedOpenAI } from '@mlflow/openai';

// Wrap the OpenAI client with the tracedOpenAI function
const client = tracedOpenAI(new OpenAI());

// Invoke the client as usual
const response = await client.chat.completions.create({
  model: 'o4-mini',
  messages: [
    { role: 'system', content: 'You are a helpful weather assistant.' },
    { role: 'user', content: "What's the weather like in Seattle?" },
  ],
});

View traces in MLflow UI:

Documentation 📘

Official documentation for MLflow Typescript SDK can be found here.

License

This project is licensed under the Apache License 2.0.