Deep Dive in Nx Backends - Lecture 1

Explore the inner workings of Elixir's Nx library in this conference talk from Code BEAM America 2022. Dive deep into Nx Backends, examining their role in performance and automatic differentiation. Learn how Nx compiles and executes code samples, and discover how numerical function definitions (defn) enable integration with Axon, a deep learning library. Gain insights into the technical aspects of Nx, including its configurable backend structure and runtime dispatch of inputs. Suitable for BEAM users interested in machine learning and those with an ML background curious about BEAM technologies. The talk covers Nx Backends, their use cases, comparison with numerical definitions, Nx.Defn.Expr usages, and future possibilities.

00:00 - - Intro & Table of Contents
01:11 - - What are the Nx Backends?
02:59 - - When to Use Different Backends?
04:30 - - Backends vs Numerical Definitions
09:26 - - Nx.Defn.Expr Usages
16:55 - - To +Inf and Beyond