signalHugging Face Blog2026-09-30
NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction
NVIDIA Kumo Tabular, an open foundation model for tabular data, is now available on Hugging Face. It predicts labels in a single forward pass without training, tuning, or feature engineering, for both classification and regression. Pretrained only on artificial data, it comes in three sizes from 28M to 215M parameters and ranks first on TabArena, BeyondArena, TALENT, and ScoringBench.
- for who
- Data scientists and ML engineers working with tabular prediction tasks.
- why now
- NVIDIA's Kumo Tabular open model now available, offering zero-training tabular prediction, a timely release.
- what changes
- They can apply a pretrained foundation model to label new rows without task-specific training, shifting from per-task models to in-context learning.
- to do
- Download Kumo Tabular from Hugging Face or GitHub and run it with the open-source library to predict labels on their own tables.
key points
- Open foundation model for tabular classification and regression, no training required.
- Three sizes from 28M to 215M parameters, pretrained on synthetic data.
- Ranks first on TabArena, BeyondArena, TALENT, and ScoringBench.
#nvidia#tabular data#foundation model#open source
score
score 8 out of 10. 0-10: how dense the facts are, multiplied by how much you can do with them after reading. 8+ means the topic's evidence bar is met: benchmarks and availability for a new model, amount and investors for a funding round, revenue figures for a solo-money story. Below 5 an item does not enter the digest. A press release scores 3 or less, a reprint loses 2, anything older than 14 days loses 1, a headline that misleads loses 3.
read the source