Module benchmark 

Structs

BenchmarkBaseline

Published baseline accuracy for reference comparison.

BenchmarkResult

Result of running an agent on a benchmark task.

BenchmarkRun

Aggregated benchmark run for an agent against one suite.

BenchmarkTask

A single task loaded from a benchmark dataset.

Enums

BenchmarkSuite

Supported public benchmark suites.

Functions

published_baselines

Hardcoded published baselines for percentile calculation.