anonymous-data / README.md
mixcausalbench's picture
Add files using upload-large-folder tool
25b0cde verified
---
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.