TensorForgejs: Machine Learning framework for JavaScript

TensorForge is a modular machine learning framework for JavaScript and TypeScript. It provides essential mathematical structures (scalars, vectors, matrices, tensors) and a growing suite of machine learning models and utilities, making it easy to build, experiment, and learn about ML algorithms in a familiar language.

Documentation

Docs

Features

• Core Math Structures:
  • Scalar (single number)
  • Vector (1D array)
  • Matrix (2D array)
  • Tensor (ND array)
• Mathematical Operations:
  • Elementwise addition, multiplication, dot product, sum, average, reshape, transpose, and more
  • Random, zeros, and ones matrix generation
  • Activation functions (e.g., sigmoid, relu)
• Machine Learning Models:
  • K-Nearest Neighbors (KNN)
  • Linear Regression
  • Logistic Regression
• Error Metrics:
  • Mean Squared Error (MSE)
  • Custom error functions
• TypeScript Support:
  • Written in TypeScript for type safety and modern development

Installation

git clone https://github.com/philipszdavido/TensorForgejs.git
cd TensorForgejs
npm install

Usage Example

import { Matrix, Vector, KNN } from 'TensorForgejs';

// Create a vector
const v = new Vector(3);
v.set(0, 1);
v.set(1, 2);
v.set(2, 3);

// Create a matrix
const m = Matrix.from([
	[1, 2, 3],
	[4, 5, 6],
]);

// Use KNN (example)
const samples = [
	[1, 2],
	[2, 3],
	[3, 4],
];
const knn = new KNN(samples);
const prediction = knn.predict([2, 3]);
console.log('KNN Prediction:', prediction);

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests for new features, bug fixes, or documentation improvements.