OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Data & Analytics

Nvtabular

NVTabular is an open-source Python library for feature engineering and preprocessing tabular data, designed to manipulate terabyte-scale datasets for deep-learning recommender systems. It provides high-level operations and accelerates computation on GPUs through RAPIDS Dask-cuDF.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Nvtabular — OpenViST project profile in AI Data & Analytics.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use NVTabular when preparing large tabular datasets for deep-learning recommender systems, especially when GPU-accelerated feature engineering and preprocessing fit the surrounding NVIDIA Merlin workflow.

Who it is for

  • Data scientists preparing recommendation data for deep-learning models.
  • Machine-learning engineers building feature-engineering and preprocessing pipelines for large tabular datasets.

Documented uses

  • Transform tabular datasets into inputs for recommender-system training.
  • Create and refine feature-engineering pipelines for recommendation data.
  • Select and preprocess features in datasets that may exceed available CPU or GPU memory.
  • Reuse training-time preprocessing for incoming inference data when integrated with Triton Inference Server.
Before you choose

Requirements & limits

The repository identifies NVTabular as open source under the Apache-2.0 license. The README documents pip installation as CPU-only; GPU support has additional CUDA, NVIDIA GPU, driver and Linux or WSL prerequisites.

  • The library is focused on tabular feature engineering and preprocessing rather than serving as a recommender model by itself.
  • GPU acceleration depends on RAPIDS Dask-cuDF and requires CUDA 11.0+, a Pascal-or-newer NVIDIA GPU, driver 450.80.02+, and Linux or WSL.
  • The README documents Conda and Docker installation paths as well as pip; pip runs CPU-only and may require additional dependencies.
  • NVTabular is one part of NVIDIA Merlin and can be combined with Merlin model and systems components; Triton integration is described for applying training preprocessing during inference.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install NVTabular for CPU use with pip

Install the library using the README's documented pip command for CPU-only operation.

Before you start

  • Python 3.7 or later.
  • A Python environment with pip.

Steps

  1. Run the command below.

    Documented command
    pip install nvtabular

  2. Use the installed library in a Python workflow for tabular feature engineering and preprocessing.

Expected outcome

NVTabular is installed for CPU-only use.

Guide limitations

  • The README states that pip installation runs NVTabular on CPU only.
  • Additional dependencies may need to be installed manually.
  • This covers installation only, not a data-processing workflow.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
NVIDIA-Merlin/NVTabular · Official homepage title · retained in its source language
How we review sourcesSuggest a correction