Yandex Laboratory hosted a seminar titled "TabM: Advancing Tabular Deep Learning With Parameter-Efficient Ensembling"

On August 26, 2025, the lab hosted the workshop "TabM: Advancing Tabular Deep Learning With Parameter-Efficient Ensembling"

Яндекс

Akim Kotelnikov,
Научно-учебная лаборатория компании Яндекс: Младший научный сотрудник

The seminar discussed a new work on effective ensembles in tabular data. We talked about the architecture, motivation, results, and analysis of the model. This model is interesting because it competes with gradient boosting (for example, CatBoost) and is simple and fast - this could be a big step forward in the development of neural networks for tabular data.