eval.BenchmarkResult

Result of benchmarking a persona across models.

Usage

Source

eval.BenchmarkResult()

Parameter Attributes

persona: str
scores: dict[str, float]
dimension_scores: dict[str, dict[str, float]]
best_model: str
passed_models: list[str]
eval_results: EvalResults

Attributes

persona: str

The persona name that was benchmarked.

scores: dict[str, float]

Mapping of model string to overall mean score (0.0–1.0).

dimension_scores: dict[str, dict[str, float]]

Mapping of model string to per-dimension scores.

best_model: str

The model with the highest overall score.

passed_models: list[str]

Models that met the threshold.

eval_results: EvalResults
The underlying EvalResults for further analysis.

Methods

Name Description
ranking() Return models ranked by overall score (descending).

ranking()

Return models ranked by overall score (descending).

Usage

Source

ranking()
Returns
list[tuple[str, float]]
List of (model, score) tuples, highest first.