docs

TinyLlama CLI - Documentation

Table of Contents

1. Installation

   • Quick Start
   • Manual Setup
   • Bootstrap Script
   • Environment Variables

2. Usage

   • Starting the CLI
   • Model Selection
   • Chat Commands
   • Settings

3. Model Download

   • Available Models
   • Custom Models
   • Download Options

4. Configuration

   • Environment Variables
   • HuggingFace Token
   • Model Directory

5. Advanced Features

   • Prompt Tuning
   • Training Data Export
   • Transcript Saving

6. API Reference

   • CLI Commands
   • Python API

7. Troubleshooting

   • Common Issues
   • Error Messages

---

Overview

TinyLlama CLI is a polished terminal chat application powered by local language models. It provides a rich terminal UI with markdown rendering, automatic prompt tuning, and training data export capabilities.

Key Features

• Local Inference: Run models locally on your machine
• Rich Terminal UI: Fancy panels, markdown rendering, status spinners
• Auto-Tuning: Runtime prompt-aware optimization for better responses
• Training Data Export: Automatic export to JSONL format
• Multiple Model Support: Download and use various models from HuggingFace