Livecoding Madness: Let's Build a Deep Learning Library

video of the livecoding.

slides

Joel Grus

@joelgrus

research engineer, Allen Institute for AI

author, Data Science from Scratch

co-host, "Adversarial Learning" podcast

Livecoding Madness: Let's Build a Deep Learning Library

In this talk we'll build a deep learning library from scratch (well, from NumPy) in 45 minutes.

How is that possible? By typing really fast.

Along the way we'll talk about library design, type hinting, Python 3.6, and of course deep learning.

Here's the plan:

  1. Tensors
  2. Loss Functions
  3. Layers
  4. Neural Nets
  5. Optimizers
  6. Data
  7. Training
  8. XOR Example
  9. FizzBuzz Example