Weekly round-up: Explore pre-trained models

The new pre-trained model exploration page showcases more than twenty pre-trained models that can be deployed and immediately used in your Baseten applications or called via API requests.

The new pre-trained model exploration page

Find a model that works for your use case by filtering by tags and libraries. And if there is an open-source model that you’d like to see us add to our pre-trained model library, let us know at support@baseten.co

Select model version in worklets

When you add a model block to a worklet, you can now select which version of the model is invoked. By default, the primary version of the model will be selected.

Select any version of a deployed model to invoke in a worklet

This is especially useful for keeping production behavior consistent as you deploy new model versions until you have a chance to update the applications that depend on these models.

🎃 The pumpkin patch

A handful of small-but-mighty changes to make Baseten more joyful to use:

  • Added documentation around specific versions of model frameworks that are officially supported by the Baseten Python client

  • Fixed issue with AWS SageMaker input format selector when deploying a model from SageMaker

  • Performance fixes and bug improvements