wrom/

HebrewBible_HapaxLegomenon

A dataset created for the NLP Research Course 097920, focusing on Hapax Legomenon — words that appear only once in the entire Hebrew Bible.
This dataset is designed to study LLM understanding of rare words in context, comparing a Hebrew-specific LLM (dicta-il/dictalm2.0-instruct) with a general-purpose LLM (gemini-2.0-flash).

We designed three annotation tasks to evaluate LLM outputs:

1️⃣ Preference… See the full description on the dataset page: https://huggingface.co/datasets/wrom/HebrewBible_HapaxLegomenon.