benchmark2 / benchpress_eval /__init__.py
anon's picture
Add EditJudge-Bench evaluation code
25cbcc2
Raw
History Blame Contribute Delete
470 Bytes
"""Small utilities for evaluating judges on the EditJudge-Bench dataset release."""
from .data import expand_triplets, load_dataset, validate_dataset
from .metrics import (
align_predictions,
compute_overall_metrics,
per_edit_type_auc,
per_negative_type_auc,
)
__all__ = [
"align_predictions",
"compute_overall_metrics",
"expand_triplets",
"load_dataset",
"per_edit_type_auc",
"per_negative_type_auc",
"validate_dataset",
]