Metewand

Metewand benchmarks computational methods across languages, evaluates results independently, and records timings, metrics, failures, and provenance.

 

Metewand compares computational methods across languages. You write ordinary Python or R functions and declare the comparison in YAML. The command-line runner executes methods as separate processes, evaluates their answers independently, and retains settings, timings, metrics, failures, and provenance.

The runner supports parameter grids, retries, resuming runs, and exporting results for analysis. Metewand is under active development, with Python and R packages and support for native Rust workers.