> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/sdgm/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/sdgm/_mcp/server.

# Overview

> Get started with the Structured Data Models library for local inference on tabular, relational, and time series data.

[**Structured Data Models**](https://github.com/NVIDIA/structured-data-models)
(`sdm`) is a Python library for pretrained model inference on tabular,
relational, and time series data. It provides model implementations,
PyTorch-compatible tensor types, and preprocessing and postprocessing tools.

## Get Started

1. Choose a model from the
   [NVIDIA Kumo collection on Hugging Face](https://huggingface.co/collections/nvidia/nvidia-kumo-structured).
   Check its supported tasks, input requirements, and weight license.
2. Follow the [User Guide](/structured-data-models/user-guide) to install SDM,
   load weights, and run inference.
3. Use the [repository examples](https://github.com/NVIDIA/structured-data-models/tree/main/examples)
   for model-specific data preparation and inference code.

## Choose How to Run Predictions

| Your goal                                                                    | Start with                                 |
| ---------------------------------------------------------------------------- | ------------------------------------------ |
| Load pretrained models and run inference directly in your Python environment | **Structured Data Models** — this tab      |
| Send prediction requests to a deployed NVIDIA NIM endpoint                   | [Kumo Relational Client](/client/overview) |

## Library Documentation

The [SDM documentation](https://nvidia.github.io/structured-data-models/) covers
model parameters, tensor semantics, data processing, and APIs.

* [Supported models and APIs](https://nvidia.github.io/structured-data-models/api/models.html)
* [In-context learning](https://nvidia.github.io/structured-data-models/icl.html)
* [Source code and issues](https://github.com/NVIDIA/structured-data-models)