Skip to content

Incorrect ground truth in benchmark at indices 99, 100, 279, 280 #30

@giftedsoda

Description

@giftedsoda

Hi, I identified some labeling errors in the benchmark dataset. These incorrect answers affect the evaluation accuracy. Specifically, the ground truth for indices 99, 100, 279, and 280 appears to be wrong.Since this is a benchmark, these errors might lead to inaccurate model scoring.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions