Datasets:

---

yuyijiong

/

difficult_retrieval

like 1

This PR updates the dataset card for "Difficult Long-context Retrieval Tasks".

Key improvements include:

• Adding relevant task_categories (question-answering) and tags (long-context, retrieval, llm-evaluation, benchmark) to the metadata for better discoverability.
• Integrating a summary of the dataset's purpose and findings directly from the associated GitHub repository's README to provide a more comprehensive overview.
• Adding a "Sample Usage" section, detailing how to evaluate LLMs using the provided evaluate.py script and various prompt styles, enhancing the usability of the dataset card.
• Ensuring the paper link is correctly formatted in Markdown.