Google Research

Overview

Google Research is the research division of Google, spanning fundamental and applied work across machine learning, systems, and AI. In tabular machine learning it released TabFM, a zero-shot foundation model that reframes tabular prediction as an in-context learning problem.

Key Product

TabFM (TabFM-1.0.0) is a zero-shot tabular foundation model for classification and regression. It combines alternating row and column attention with row compression, and is pretrained on hundreds of millions of synthetic datasets generated from structural causal models. The model is released open-weights under Apache-2.0 on GitHub and Hugging Face, with both JAX and PyTorch backends. The repository notes it is “not an officially supported Google product.”

Outlook

Google entering the tabular foundation model space brings substantial compute and research depth to a field so far led by startups and academic labs. As an open-weights release, TabFM lowers the barrier to zero-shot tabular prediction, though its long-term investment and productization within Google’s broader AI stack remain to be seen.