INNER CODE UNIT · Python
stddev
AgentSkillOS/SkillAnything · scripts/aggregate_benchmark.py:44
stddev = math.sqrt(variance)
else:
stddev = 0.0
return {
"mean": mean,
"stddev": stddev,
"min": min(values),
"max": max(values),
"count": n,
}
def aggregate_benchmark(
workspace_dir: Path | None = None,
run_dirs: list[Path] | None = None,
) -> dict[str, Any]:
"""Aggregate benchmark results from multiple runs.