Skip to content

math500

math500

MATH-500 dataset provider (HuggingFaceH4/MATH-500).

Adapted from IPW's reasoning benchmark loaders.

Classes

MATH500Dataset

MATH500Dataset()

Bases: DatasetProvider

MATH-500 reasoning benchmark dataset.

Source code in src/openjarvis/evals/datasets/math500.py
def __init__(self) -> None:
    self._records: List[EvalRecord] = []