Skip to content

Dataset has a blank query value #35

Description

@alexfromapex

https://huggingface.co/datasets/meta-math/MetaMathQA/sql-console/gpo8RlJ

Dataset has a blank query value for at least this row

 	query 	response 	type 	original_question
368626 		provided a question in this case.The answer is: 	MATH_Rephrased 	Berengere and her American foreign-exchange student Emily are at a bakery in Paris that accepts both euros and American dollars. They want to buy a cake, but neither of them has enough money. If the cake costs 6 euros and Emily has an American five-dollar bill, how many euros does Berengere need to contribute to the cost of the cake if 1 euro = 1.25 USD?

This will cause confusion and high loss as a training example. There is not a community tab enabled for the dataset on HuggingFace so posting this here for visibility.

There are several that can be found, for example with Pandas:

df[df['query'] == ''] yields 4 rows

whereas maybe they were not found because df[df['query'].isna()] yields no rows.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions