TabFM: A zero-shot foundation model for tabular data

Google Research has introduced TabFM, a foundation model trained to work directly with tabular data—the spreadsheets and structured databases that power most enterprise systems. Unlike traditional machine learning approaches, TabFM operates in a zero-shot mode, meaning it can perform tasks on new data without requiring specific fine-tuning. According to Google, this advance addresses a gap: while foundation models have transformed how we handle text and images, the vast majority of real-world data remains locked in tables and databases. TabFM aims to bring the same capability and adaptability to structured data that recent models have brought to unstructured domains.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton