Developer Guide

Developer Guide

Overview

Welcome to the TokenPrint Developer Guide. This section is intended for engineers, researchers, and open-source contributors who want to modify the source code, add new features, or fix bugs.

Why it matters

TokenPrint is a complex application spanning multiple languages (Python, TypeScript) and paradigms (REST, WebSockets, React, WebGL). Without a clear map of how the codebase is structured and how to compile it, contributing can be daunting.

How TokenPrint implements it

The repository is strictly divided into two root folders: backend/ and frontend/. They do not share dependencies, build systems, or languages.

Before you write any code, we highly recommend reading the Architecture and Visual Mapping documentation to understand the project's core philosophy: Never fabricate data.

Section Contents

• Repository Structure: Where files live.
• Building From Source: How to compile and run in dev mode.
• Code Style: Linters, formatters, and architectural rules.
• Adding a New Visualization: How to tie new 3D geometry to tensor data.
• Adding a New Model: How to support new architectures like DeepSeek.
• Creating UI Components: How to add buttons and panels to the AppShell.
• Debugging: How to use TokenPrint's own debug tools on itself.
• Performance Tips: How to keep WebGL running at 60fps.

Diagram

flowchart LR
    Dev([Developer]) --> Backend[Python Backend]
    Dev --> Frontend[TypeScript Frontend]
    
    subgraph Backend Setup
        Backend --> Venv[Virtual Environment]
        Venv --> PyTorch[PyTorch Engine]
    end
    
    subgraph Frontend Setup
        Frontend --> NPM[Node Package Manager]
        NPM --> Next[Next.js Server]
    end

Loading

Related pages

• Contributing
• Architecture

Further reading

• Project README

Navigation

Previous	Home	Next
Event System	Home	Repository Structure