Sefaria/he_berel_gold

该数据集包含近3000个来自Sefaria的标注示例，用于训练希伯来语命名实体识别（NER）模型。目前标注包括以下类别：- מקור（来源）：用于任何引用，包括ibid、להלן和דיבור המתחיל。 - בן-אדם（人物）：用于任何个人、人群（例如בית שמאי）或类似人的实体（如מלאכים和匿名实体如שמעון/לוי）。数据集通过训练BEREL分类器并微调GPT-4o作为分类器来整理，其中包含两个分类器不一致的示例。

This dataset contains nearly 3,000 annotated examples from Sefaria, intended for training Hebrew named entity recognition (NER) models. The current annotation covers the following categories:
- מקור (Source): For any citations, including ibid, להלן, and דיבור המתחיל.
- בן-אדם (Person): For any individuals, groups (e.g., בית שמאי), or human-like entities (such as מלאכים and anonymous entities like שמעון/לוי).
The dataset was curated by training a BEREL classifier and fine-tuning GPT-4o as a classifier, and includes two examples where the two classifiers produced inconsistent results.