trekhleb / micrograd-ts

🤖 A TypeScript version of karpathy/micrograd — a tiny scalar-valued autograd engine and a neural net on top of it

Hi there!

I'm Oleksii. I work as a full-time software engineer at Uber. In my spare time, I do open-sourcing.

Most of my open-source 🚀 projects below 👇🏻 are focused on one thing - to help people learn 📚. You may use these projects to learn about algorithms in JavaScript and get prepared for technical interviews, or to learn Python syntax and start experimenting with machine learning algorithms and the math behind them, etc.

There are also other projects that I've developed for fun but that you might find useful, like ✍🏻 okso.app online drawing app.

I also write 📝 articles about life, web development and machine learning.

Getting in touch

trekhleb.dev

   

micrograd-ts's Issues

Visualize Test Predictions after Training

Currently DemoMLPTraining visualizes the final predictions of the model based on the training data alongside a graph showing the loss over time. However, there is no visualization of how the model performs on a range of test data after it has been trained.