Nvidia Kumo Tabular: Pushing the Accuracy-Efficiency Frontier
Comments
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Summary
Nvidia researchers introduced Kumo Tabular, a family of open tabular foundation models for classification and regression in 27M, 61M, and 214M parameter sizes. Pretrained only on artificial tables, they predict new tables without task-specific training.