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.