Improve dataset card: Add task categories, GitHub links, and detailed overview

#1
by nielsr HF Staff - opened

This PR significantly improves the FinAuditing dataset card by:

  • Updating the paper title in the metadata to accurately reflect the official title.
  • Adding question-answering to the task_categories metadata, which better represents the dataset's focus on information retrieval, relation extraction, and mathematical reasoning tasks for LLMs.
  • Including direct links to both the main GitHub repository and the Evaluation Framework repository for easy access to associated code and resources.
  • Adding a detailed overview of the FinAuditing benchmark, extracted from the paper abstract, to provide better context.
  • Incorporating the "Datasets Released" table from the GitHub README, offering clear descriptions of the benchmark's subtasks (FinSM, FinRE, FinMR) and their subsets.
  • Including the BibTeX citation for the paper in the content section.
  • Removing the empty "File information" section.

These updates provide a more comprehensive and accessible resource for researchers and users interested in the FinAuditing benchmark.

Ready to merge
This branch is ready to get merged automatically.

Sign up or log in to comment