OPEN LINK ↗
// 1h agoMODEL RELEASE
Google Research open-sources TabFM model
TabFM is a scikit-learn compatible tabular foundation model developed by Google Research for zero-shot classification and regression. Leveraging in-context learning, it reads training rows as context to predict test targets in a single forward pass using PyTorch or JAX backends.
// ANALYSIS
Hot take: Zero-shot tabular models are the next frontier for AutoML, but their true value depends on whether in-context learning can consistently beat traditional gradient boosted trees on complex, real-world schemas.
- –Dual-backend support allows developers to run inference using either JAX or PyTorch.
- –Integration with scikit-learn API makes it a drop-in replacement in existing pipelines.
- –In-context learning eliminates feature engineering, hyperparameter tuning, and dataset-specific training.
// TAGS
tabfmtabular-datafoundation-modelmachine-learningzero-shotscikit-learnjaxpytorchgoogle-research
DISCOVERED
1h ago
2026-07-03
PUBLISHED
1h ago
2026-07-03
RELEVANCE
8/ 10
AUTHOR
Github Awesome