| 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. | |