Getting Started

Welcome to Tiny-DL-Inference! This guide will help you get started with the high-performance WebGPU deep learning inference engine.

What is Tiny-DL-Inference?

Tiny-DL-Inference is a production-ready, zero-dependency deep learning inference engine built on WebGPU. It demonstrates a deep understanding of AI acceleration and heterogeneous computing by implementing neural network operators from scratch.

Quick Navigation

🚀 First Steps

New to Tiny-DL-Inference? Start here:

• Quick Start - Get up and running in 5 minutes
• Installation - Detailed setup instructions
• Architecture - Understand the system design

📚 Core Concepts

Learn the fundamentals:

• GPU Context - WebGPU resource management
• Tensor Operations - Multi-dimensional data structures
• Operators - Neural network layers
• Memory Layout - NCHW vs NHWC

💡 Examples

See it in action:

• MNIST Classification - Handwritten digit recognition
• Custom Model - Build models from scratch
• Web Integration - Browser-based applications
• Performance Tuning - Optimization tips

🔧 Advanced Topics

Deep dive into optimization:

• Kernel Fusion - Custom fused operators
• Custom Operators - Build your own WGSL operators
• Benchmarking - Performance measurement
• Optimization Guide - Best practices

Installation

npm

npm install tiny-dl-inference

pnpm

pnpm add tiny-dl-inference

yarn

yarn add tiny-dl-inference

First Inference

Here's a minimal example to get you started:

import { GPUContext, Tensor, ReLUOperator } from 'tiny-dl-inference';

// Initialize GPU context
const context = new GPUContext();
await context.init();

// Create input tensor
const input = Tensor.fromArray(context, 
  new Float32Array([1.0, -2.0, 3.0, -4.0]),
  [1, 4, 1, 1]
);

// Run ReLU activation
const relu = new ReLUOperator(context);
const output = await relu.forward([input]);

// Get results
const result = await output.download();
console.log(result); // Float32Array([1, 0, 3, 0])

// Cleanup
input.destroy();
output.destroy();
context.destroy();

Next Steps

1. Read the Quick Start Guide for a complete walkthrough
2. Explore the API Reference for detailed documentation
3. Try the MNIST Example for a real-world use case

Browser Support

Tiny-DL-Inference requires WebGPU support. See Browser Compatibility for details.

Getting Help

• 📖 Full Documentation
• 💻 GitHub Repository
• 🐛 Issue Tracker