DataDesigner/docs/notebook_source/_README.md

Overview

Welcome to the Data Designer tutorial series! These hands-on notebooks will guide you through the core concepts and features of Data Designer, from basic synthetic data generation to advanced techniques like structured outputs and dataset seeding.

🚀 Setting Up Your Environment

Local Setup Best Practices

First, download the tutorial from the release assets. To run the tutorial notebooks locally, we recommend using a virtual environment to manage dependencies:

=== "uv (Recommended)"

```bash
# Extract tutorial notebooks
unzip data_designer_tutorial.zip
cd data_designer_tutorial

# Launch Jupyter
uv run jupyter notebook
```

=== "pip + venv"

```bash
# Extract tutorial notebooks
unzip data_designer_tutorial.zip
cd data_designer_tutorial

# Create Python virtual environment and install required packages
python -m venv venv
source venv/bin/activate
pip install data-designer jupyter

# Launch Jupyter
jupyter notebook
```

API Keys and Authentication

Data Designer is able to interface with various LLM providers. You'll need to set up API keys for the models you want to use:

# For NVIDIA API Catalog (build.nvidia.com)
export NVIDIA_API_KEY="your-api-key-here"

# For OpenAI
export OPENAI_API_KEY="your-api-key-here"

# For OpenRouter
export OPENROUTER_API_KEY="your-api-key-here"

For more information, check the Welcome, Default Model Settings and how to Configure Model Settings Using The CLI.

📚 Tutorial Series

The tutorials are designed to be completed in sequence, building upon concepts introduced in previous notebooks:

1. The Basics

Learn the fundamentals of Data Designer by generating a simple product review dataset. This notebook covers:

• Setting up the DataDesigner interface
• Configuring models and inference parameters
• Using built-in samplers (Category, Person, Uniform)
• Generating LLM text columns with dependencies
• Understanding the generation workflow

Start here if you're new to Data Designer!

2. Structured Outputs, Jinja Expressions, and Conditional Generation

Explore more advanced data generation capabilities:

• Creating structured JSON outputs with schemas
• Using Jinja expressions for derived columns
• Combining samplers with structured data
• Building complex data dependencies
• Working with nested data structures
• Conditional generation with skip.when

3. Seeding with an External Dataset

Learn how to leverage existing datasets to guide synthetic data generation:

• Loading and using seed datasets
• Sampling from real data distributions
• Combining seed data with LLM generation
• Creating realistic synthetic data based on existing patterns

4. Providing Images as Context

Learn how to use vision-language models to generate text descriptions from images:

• Processing and converting images to base64 format for model consumption
• Using vision-language models (VLMs) to analyze visual documents
• Understanding how image, audio, and video context share the same multi_modal_context field, while still requiring model support for each modality
• Generating detailed summaries from document images
• Inspecting and validating vision-based generation results

5. Generating Images

Generate synthetic image data with Data Designer:

• Configuring image-generation models with ImageInferenceParams
• Adding image columns with Jinja2 prompts and sampler-driven diversity
• Preview (base64 in dataframe) vs create (images saved to disk, paths in dataframe)
• Displaying generated images in the notebook

6. Image-to-Image Editing

Chain image generation columns to generate and then edit images:

• Generating images from text and then editing them in a follow-up column
• Using ImageContext with auto-detection to pass generated images to an editing model
• Combining sampled accessories and settings for varied edits
• Comparing generated vs edited images in preview and create modes

📖 Important Documentation Sections

Before diving into the tutorials, familiarize yourself with these key documentation sections:

Getting Started

• Welcome & Installation - Overview of Data Designer capabilities and installation instructions

Core Concepts

Understanding these concepts will help you make the most of the tutorials:

• Columns - Learn about different column types (Sampler, LLM, Expression, Validation, etc.)
• Validators - Understand how to validate generated data with Python, SQL, and remote validators
• Person Sampling - Learn how to sample realistic person data with demographic attributes