Summary
In this chapter, we learned what additional steps are required in moving a model to production. We learned what an API is and what its components are. After creating an API with the use of FastAPI, we glanced at the core steps of creating a Docker image of the API, creating a Docker container, and pushing the Docker container to cloud so that predictions can be made from any device. Then, we learned about ways to identify images that are out of distribution from the original dataset. Additionally, we also learned about leveraging FAISS to calculate the distance with similar vectors much faster.
To summarize, we have seen all the individual steps to deploy a model to production, including building a Docker container, deploying on the AWS cloud, identifying data drift and thereby understanding the scenario of when to re-train the model, and performing image similarity much faster so that we can identify data drift with less compute.
Images are fascinating. Storing them...
