/NVTabular

A library that sits on top of RAPIDS cuDF library providing a range of benefits for processing extremely large tabular datasets, particularly those that do not fit in GPU or CPU memory. NVTabular has many capabilities including fast terabyte-scale data preparation and accelerated tabular data loading, all on GPU, which streamline the first step for both training and inference to any deep recommender system pipelines.

Primary LanguagePythonApache License 2.0Apache-2.0

Watchers