Datasets:

Modalities:
Text
Formats:
csv
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
License:
Cohere For AI org
edited Jul 10

Feel free if this is not wanted.

This PR introduces task_id's in the format f"lbpp/{idx}" to each sample, following the column naming from HumanEval/MBPP (and adopting the same naming scheme as HumanEval).
This specifically allows a better "indexing" of sample rows, especially after removing canary samples.

dennlinger changed pull request status to open
mgalle changed pull request status to merged

Sign up or log in to comment