Основні поняття
AutoRE introduces a novel paradigm, RHF, for efficient Document-Level Relation Extraction, achieving state-of-the-art results on the RE-DocRED dataset.
Статистика
大規模言語モデル(LLMs)はテキストの理解と生成に優れている。
Re-DocREDデータセットでAutoREの最高性能を示す。
Цитати
"LLMs still perform poorly in DocRE tasks involving the extraction of multiple relations and triplet facts."
"AutoRE proficiently addresses the intricate task of extracting multiple relations from document-level texts."