r-chong/TSTorch
A PyTorch-style runtime library in TypeScript + WebGPU. Built to understand how ML frameworks and models work internally.
This is a machine learning library built from scratch in TypeScript, designed for those who want to understand the inner workings of ML frameworks. You can input data like numbers or arrays and define neural network structures using layers and activation functions. The output includes trained models capable of making predictions and insights into how automatic differentiation and tensor operations function. This tool is ideal for developers interested in learning about machine learning framework internals or building custom ML components in TypeScript.
Use this if you are a developer seeking to learn how machine learning frameworks like PyTorch work by implementing core components like autograd, tensors, and neural network layers from scratch in TypeScript.
Not ideal if you are looking for a high-performance, production-ready machine learning framework for large-scale applications or if you prefer a 'black-box' approach to ML development.
Stars
39
Forks
3
Language
TypeScript
License
MIT
Category
Last pushed
Mar 12, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/r-chong/TSTorch"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Compare
Higher-rated alternatives
tensorflow/tfjs
A WebGL accelerated JavaScript library for training and deploying ML models.
nnstreamer/api
Machine Learning API (Origin: C++: SNAP, C/C#: Tizen API, Java: Samsung-Research ML API). For...
microsoft/DMTK
Microsoft Distributed Machine Learning Toolkit
R-js/blasjs
Pure Javascript manually written :ok_hand: implementation of BLAS, Many numerical software...
mljs/matrix
Matrix manipulation and computation library