← back to paper
arxiv: 2605.00199 · 2 revisions
RSAT: Structured Attribution Makes Small Language Models Faithful Table Reasoners