EXAONE Tabular ranked first on TabArena for classification by Elo score, a user reports
A user describes LG AI Research’s classification and regression models as having about 21 million parameters each. The released weights are limited to non-commercial research and education, the user says, while the inference code permits commercial use.
TLDR
A September 11, 2026 post reports that LG AI Research’s EXAONE Tabular ranked first for classification and second for regression on TabArena by Elo score, and that its technical report was available. According to the user, the model handles missing values without requiring them to be filled in first and uses missingness information to represent features. It also uses built-in feature selection for tables that exceed its feature limit. The user notes that commercial use is permitted for the inference code, but the released weights are restricted to non-commercial research and education.
Combined views
—
1 Source, first seen ago
