Deploying 🤗 ViT on Vertex AI

@Sayak_Paul and I have been covering the deployment aspects of (TensorFlow) vision models from :hugs: Transformers, and we’re delighted to announce the final post on that series today!

In this post, we cover deployment with Vertex AI. You’ll learn how to deploy a ViT B/16 following best practices, consuming the deployed endpoint in different forms, conducting load-tests, and, more importantly, the pricing around the deployment.

When we started this series, there was a dearth of resources showing how to deploy TF models from :hugs: Transformers to GCP and TF ecosystem following good practices. We wanted to close that gap :hugs:

Here’s the latest post: Deploying :hugs: ViT on Vertex AI

For those of who are interested in the last two posts

  1. Deploying TensorFlow Vision Models in Hugging Face with TF Serving

  2. Deploying :hugs: ViT on Kubernetes with TF Serving

Thanks to all the reviewers from :hugs:: @merve , @osanseviero, João Gante, Matthew Carrigan, and Steven Liu! Thanks to the ML Developer Programs team at Google (Soonson Kwon) for all the support around GCP credits.