NVIDIA releases Kumo Relational, a pre-trained model that predicts directly on relational databases without training
NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required
NVIDIA released Kumo Relational, a pre-trained relational foundation model that answers predictive queries on enterprise databases in a single forward pass with no task-specific training or feature engineering. Across 41 tasks on 15 databases from RelBench V1/V2, SAP SALT and 4DBInfer, it beats supervised relational deep learning models and expert-engineered gradient-boosted tree pipelines, and on SAP SALT it surpasses AutoGluon ensembles by 8% in MRR. It reads up to 10,000 context examples at inference, scales to 500B+ rows, and is available through NVIDIA's open-source structured-data-models library or as an NVIDIA NIM, with code under Apache 2.0 and weights under the OpenMDW License 1.1.
The release reports benchmark margins over supervised pipelines and the conditions where in-context prediction still loses.
Source: Hugging Face Blog · huggingface.co