metadata
configs:
- config_name: default
data_files:
- split: all
path: all.jsonl
This dataset consists of LLM-generated items in the style of the Choice of Plausible Alternatives (COPA) benchmark. For more info, see the paper and the Github repo.