Geval summarization base

Base class for GEval Summarization metrics.

This module provides a base class for all GEval summarization metrics to avoid code duplication.

GEvalSummarizationBaseMetric(name=None, evaluation_params=None, model=DefaultValues.MODEL, criteria=None, evaluation_steps=None, rubric=None, model_credentials=None, model_config=None, threshold=0.5, additional_context=None, batch_status_check_interval=DefaultValues.BATCH_STATUS_CHECK_INTERVAL, batch_max_iterations=DefaultValues.BATCH_MAX_ITERATIONS, num_judges=DefaultValues.NUM_JUDGES, aggregation_method=DefaultValues.AGGREGATION_METHOD, max_concurrent_judges=None, strict_mode=False)

Bases: DeepEvalGEvalMetric

Base class for GEval Summarization metrics.

This class provides common functionality for summarization metrics including: 1. Applying the GEval output contract via _apply_contract (score, rubric_score, success, threshold, etc.) 2. Preserving the 1-3 diagnostic rubric_score via _to_rubric_score

Available Fields

• input (str): Source text or transcript.
• actual_output (str): Generated summary.

Scoring

• [0, 1] float: Normalized score after _apply_contract.

Attributes:

Name	Type	Description
higher_is_better	bool	Whether a higher score indicates better quality. Defaults to True.
threshold	float	Pass/fail threshold. Concrete subclasses must override this.