---
title: "Discussion on \"Optax: Optimizers You Can Compose Like LEGO\" | Hashnode"
description: "Discussion on \"Optax: Optimizers You Can Compose Like LEGO\". In our last article on this blog in March, we built a complete training loop. We used optax.adamw() to create an optimizer, wrapped it in nnx.Optimizer, and watched our model learn.\nBut we barely scra"
url: "https://hashnode.com/posts/optax-optimizers/6a44cf56dcce9d94f5e9173e"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Optax: Optimizers You Can Compose Like LEGO](https://kambale.dev/optax-optimizers?utm_source=hashnode&utm_medium=feed)

In our last article on this blog in March, we built a complete training loop. We used optax.adamw() to create an optimizer, wrapped it in nnx.Optimizer, and watched our model learn. But we barely scra

Jul 1 · 11 min read · kambale.dev  
Updated Jul 29  
Read the article

-   [#jax](https://hashnode.com/tag/jax)
-   [#optimizers](https://hashnode.com/tag/optimizers)

## Written by

[Wesley Kambale](https://hashnode.com/@kambale)

@kambale · Mbarara, Uganda

  
[View profile](https://hashnode.com/@kambale)

Machine Learning Engineer | AI Community Leader | Technical Educator

I'm a Machine Learning Engineer passionate about building production-ready ML systems for the African market. With experience in TensorFlow, Keras, and Python-based workflows, I help teams bridge the gap between machine learning research and real-world deployment—especially on resource-constrained devices. I'm also a Google Developer Expert in AI. I regularly speak at tech conferences including PyCon Africa, DevFest Kampala, DevFest Nairobi and more and also write technical articles on AI/ML here.

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
