Paper: arXiv:2606.09323 — TRL-Bench: Standardizing Cross-Paradigm Representation-Level Evaluation of Tabular Encoders · Code: LOGO-CUHKSZ/TRL-Bench
Row-level evaluation suite of TRL-Bench:
row_prediction: 50 OpenML tables, 1.1M rows (887,720 train / 110,962 validation / 110,983 test) across 123 hand-verified targets. Per-row schema bundles features and targets as JSON dicts so all 50 tables share one config.
record_linkage: 16 entity-matching sources (8 clean… See the full description on the dataset page:
https://huggingface.co/datasets/logo-lab/trl-rbench.
