What it is

🎨 NeMo Data Designer: Generate high-quality synthetic data from scratch or from seed data.

How it works

• Data Designer supports generating data using statistical samplers, LLMs, or existing seed datasets.
• It allows controlling relationships between fields with dependency-aware generation and validates outputs with built-in validators (Python, SQL, and remote validators).
• It includes a feature for scoring outputs using an LLM-as-a-judge for quality assessment and supports preview mode before full-scale generation.
• The project uses an async execution engine by default to overlap column generation and optimize concurrency.

Getting started

Install

"pip install data-designer"

Or install from source: "git clone https://github.com/NVIDIA-NeMo/DataDesigner.git" "cd DataDesigner" "make install"

Set your API key

"export NVIDIA_API_KEY="your-api-key-here"" "export OPENAI_API_KEY="your-openai-api-key-here"" "export OPENROUTER_API_KEY="your-openrouter-api-key-here""

Start generating data

"import data_designer.config as dd" "from data_designer.interface import DataDesigner" "data_designer = DataDesigner()" "config_builder = dd.DataDesignerConfigBuilder()"

Example: add a product category column

"config_builder.add_column(" " dd.SamplerColumnConfig(" " name="product_category"," " sampler_type=dd.SamplerType.CATEGORY," " params=dd.CategorySamplerParams(" " values=["Electronics", "Clothing", "Home & Kitchen", "Books"]," " )," " )" ")

Example: add an LLM-driven text column

"config_builder.add_column(" " dd.LLMTextColumnConfig(" " name="review"," " model_alias="nvidia-text"," " prompt="Write a brief product review for a {{ product_category }} item you recently purchased."," " )" ")

Preview

"preview = data_designer.preview(config_builder=config_builder)" "preview.display_sample_record()"