Baseten is a brand that provides a comprehensive platform and tools for deploying and managing machine learning models. With Baseten, you can easily package and deploy models, access model resources, view logs and metrics, and manage different versions of your models. To get started, you need to install the Baseten Python client and create an API key for your Baseten account.

Once authenticated, you can deploy a model from the model library or package your own model using the Truss format. Baseten supports both draft and published models. Draft models are mutable and allow for faster testing, while published models have access to autoscaling settings and can be marked as the primary version.

Baseten also offers live reload functionality for draft models, which reduces testing time by patching the running model server instead of redeploying the entire environment. However, it's important to note that draft models are not suitable for production workloads. When you're ready, you can publish your model, rebuild it onto production infrastructure, and receive an email notification once the process is complete.

Published models have full access to autoscaling settings and cannot be interrupted with patch requests. Whether you choose to deploy draft or published models, Baseten provides a powerful and user-friendly platform to streamline your machine learning workflow