TLRD distills tri-level rationales (instance features, dataset distributions, neighbor comparisons) from a teacher into student LLMs to close the accuracy gap with tree ensembles on tabular data while generating grounded explanations.
G raph N arrator: Generating Textual Explanations for Graph Neural Networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
TLRD: Teaching LLMs to Reason over Tabular Data with Tri-Level Rationale Distillation
TLRD distills tri-level rationales (instance features, dataset distributions, neighbor comparisons) from a teacher into student LLMs to close the accuracy gap with tree ensembles on tabular data while generating grounded explanations.