Skip to content

Eval Tree API Reference

Hierarchical evaluation orchestration for composing complex evaluation workflows.

from axion.eval_tree import EvalTree
T

EvalTree

Hierarchical scoring model that builds a tree from a configuration and executes metrics using an optimized two-phase batch process.

C

Config-Driven

Define scoring trees via YAML, dict, or Config objects. Supports weighted aggregation, strategy overrides, and nested component hierarchies.

EvalTree

axion.eval_tree.EvalTree 

EvalTree(config: Union[Config, Dict[str, Any], str], max_concurrent: int = 5, summary_generator: Optional[BaseSummary] = None, enable_internal_caching: bool = True, trace_granularity: TraceGranularity = SEPARATE)

Bases: TreeMixin

Hierarchical scoring model that builds a tree from a configuration and executes metrics using an optimized two-phase batch process.

elapsed_time property 

elapsed_time: Union[float, None]

Execution Time

summary property 

summary: Union[Dict[str, Any], None]

Model Summary

execute async 

execute(input_data: Union[DatasetItem, Dict[str, Any]]) -> TestResult

Executes the evaluation for a single data item.

batch_execute async 

batch_execute(evaluation_inputs: Union[List[DatasetItem], Dataset, DataFrame], evaluation_name: str = 'EvalTree Evaluation', show_progress: bool = True) -> EvaluationResult

Executes the evaluation using an optimized two-phase process: 1. Batch Calculation: Runs all leaf metrics across the dataset. 2. In-Memory Aggregation: Computes hierarchical scores from the results.

get_node 

get_node(name: str) -> Optional[Node]

Get node by name

get_metric_summary 

get_metric_summary() -> Dict[str, Any]

display classmethod 

display()

Display Usage Documentation

Hierarchical Scoring Guide Running Evaluations