Главная
Study mode:
on
1
Intro - Flax is performant and reproducible
2
Deepnote walk-through sponsored
3
Flax basics
4
Flax vs Haiku
5
Benchmarking Flax
6
Linear regression toy example
7
Introducing Optax Adam state example
8
Creating custom models
9
self.param example
10
self.variable example
11
Handling dropout, BatchNorm, etc.
12
CNN on MNIST example
13
TrainState source code
14
CNN dropout modification
15
Outro and summary
Description:
Dive into a comprehensive tutorial on Flax, a JAX-based machine learning library, covering everything from basics to advanced concepts. Learn how to build performant and reproducible ML models, explore Flax's advantages over Haiku, and master key concepts like linear regression, custom model creation, and CNN implementation. Gain hands-on experience with practical examples, including a linear regression toy example and a CNN on MNIST dataset. Discover how to handle dropout, BatchNorm, and other essential techniques for building robust neural networks. Follow along with the provided Jupyter notebook and leverage additional resources to deepen your understanding of Flax and its ecosystem.

Machine Learning with Flax - From Zero to Hero

Aleksa Gordić - The AI Epiphany
Add to list