Datasets:

Modalities:
Tabular
Text
Formats:
csv
Size:
< 1K
Libraries:
Datasets
pandas
License:

Incorrect answers undermine the credibility of your leaderboard

#11
by sayhitoday - opened

I would like to point out an issue with one of the labels that appears to be clearly incorrect, which raises concerns about the reliability of the leaderboard.

Please help explain the problem with the following label:

39, Company_Documents, co_dc_nxt_history, 40,
"Based on the provided documents, what is the latest proposed deadline by NEXTDC for certifying SC1, S3, and M3 to ISO 14001? Provide the month and year.",
45444

The question explicitly asks for a specific month and year, yet the labeled answer is β€œ45444”, which is not a valid date and does not match the question format at all.

Why is this kind of answer considered correct in the dataset?
Errors like this directly reduce confidence in the evaluation results and the credibility of the leaderboard.

Sign up or log in to comment