Skip to contents

Compare two benchmark runs (track progress over time)

Usage

fx_benchmark_compare(old, new)

Arguments

old, new

Outputs of fx_benchmark() (e.g. before/after a change).

Value

A list with overall (per-level accuracy and headline deltas, new minus old) and by_data_type (per gold data_type, the type-accuracy delta), with regressions (negative deltas) sorted first.