"""Auto-generated example: 0 Float, 0 Categorical (with repeats)."""
import random
import bencher as bn
class BaselineCheck(bn.ParametrizedSweep):
"""Measures a fixed baseline metric with no swept parameters."""
baseline = bn.ResultFloat(units="ms", doc="Baseline latency")
def benchmark(self):
self.baseline = 42.0
self.baseline += random.gauss(0, 0.15 * 5)
def example_sweep_0_float_0_cat_with_repeats(run_cfg: bn.BenchRunCfg | None = None) -> bn.Bench:
"""0 Float, 0 Categorical (with repeats)."""
bench = BaselineCheck().to_bench(run_cfg)
bench.plot_sweep(input_vars=[], result_vars=['baseline'], description='A 0 float + 0 categorical parameter sweep with multiple repeats per combination. Repeating measurements reveals the noise structure of your benchmark. If your function is deterministic, all repeats will be identical; if it has stochastic components, repeats let you estimate confidence intervals and distinguish signal from noise. The benchmark function must be pure -- if past calls affect future calls through side effects, the statistics will be invalid. With no input variables, this is a 0D sweep that measures a single baseline metric.', post_description='Swarm/violin plots show the distribution of repeated measurements. If repeat has high variance, it suggests either measurement noise or unintended side effects in the benchmark function.')
return bench
if __name__ == "__main__":
bn.run(example_sweep_0_float_0_cat_with_repeats, subsampling_divisions=4, repeats=10)