Machine Learning in Elixir

Wrapping Up

In this chapter, you implemented a semantic search application from scratch using Phoenix, Bumblebee, and Elixir. You went through the process of deciding when to use and not use machine learning in an application. You learned a bit about some of the important questions to ask when designing machine learning applications. You used Nx.Serving to wrap machine learning pipelines into a performant, fault-tolerant inference pipeline. You learned about the significance of dynamic batching to machine learning performance, and you saw how Nx.Serving can help you achieve dynamic batching.

In addition, you implemented your own vector index and search functionality using pgvector. In doing so, you learned a bit about vector search.

Finally, ...