evaluation/__init__.py

"""
DR-Bench Evaluation Pipeline.

Provides LLM-as-judge coverage evaluation for agent outputs against
ground truth pitch points extracted from success stories.

Pipeline:
    1. Extract ground truth pitch points from success stories (extract_pitch_points.py)
    2. Run coverage judge to score agent outputs against GT (coverage_judge.py)
    3. Compute weighted coverage scores (weighted_score.py)
    4. Orchestrate evaluation via CoverageEvaluator (evaluator.py)
"""

from evaluation.evaluator import CoverageEvaluator, EvaluationResult
from evaluation.weighted_score import calculate_score

__all__ = ["CoverageEvaluator", "EvaluationResult", "calculate_score"]