iCreatAI News
Skip to content
Trending storyWatching

Databricks explains running tabular foundation models on AI Runtime

1 reports1 sources1 days agoUpdated

Understand this story

AI overview

Databricks published an article explaining how to run and fine-tune tabular foundation models (TFMs) on Databricks AI Runtime, and provided a clonable code repository containing zero-shot inference examples for various TFMs, as well as a complete example of classification and regression fine-tuning with KumoTabular. Databricks said its on-demand GPUs (including H100) can support TFMs such as TabPFN-3 scaling to 1 million rows × 200 features. The article also discusses how TFMs work, comparisons of recent models, and trade-offs for enterprise adoption.

AI overview based on source reports · Updated 1 days ago

Coverage timeline

Follow the coverage to explore different perspectives.

Nov 17
  1. Databricks Blog
    Tabular foundation models on Databricks AI Runtime

    This Databricks article explains how tabular foundation models (TFMs) are pretrained on millions of synthetic datasets and make predictions via in-context learning without task-specific training, covering tokenization approaches, row-compression architectures like TabICLv1, and a comparison of recent models (TabICLv2, TabFM, TabPFNv3.5, Mitra v2, KumoTabular) by scale and license. It describes GPU memory challenges for TFM inference and how Databricks AI Runtime provides managed serverless GPU infrastructure, MLflow tracking, Unity Catalog governance, and Model Serving for deploying TFMs.

Activity for this story

Current comparable activity
4
Peak comparable activity
9Nov 17 03:00
Change in the past 24 hours
-49%

Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.