anonymous-data / README.md
mixcausalbench's picture
Add files using upload-large-folder tool
25b0cde verified
metadata
pretty_name: Anonymous Data
configs:
  - config_name: small
    data_files:
      - split: test
        path: graphs/small/*.csv
  - config_name: large
    data_files:
      - split: test
        path: graphs/large/*.csv
  - config_name: synthetic
    data_files:
      - split: test
        path: graphs/synthetic/*.csv

Anonymous Data

This dataset contains benchmark files for evaluating language models on causal-discovery tasks.

The small and synthetic configs are available through the Hugging Face dataset viewer. The large config is provided as direct CSV files under graphs/large/; these files are large enough that automatic viewer/Croissant conversion may omit them, so the NeurIPS Croissant metadata is completed manually from the hosted CSV file layout.