Right now, it's collective under the hood. Performance under consideration of composition-only and structure-only (e.g. validity metric) should also be tracked and easily accessible. Can be computed from the same match matrix for most of the metrics, and the validity metrics are already fairly delineated.
Right now, it's collective under the hood. Performance under consideration of composition-only and structure-only (e.g. validity metric) should also be tracked and easily accessible. Can be computed from the same match matrix for most of the metrics, and the validity metrics are already fairly delineated.
https://github.com/sparks-baird/matbench-genmetrics/issues/63, https://github.com/sparks-baird/matbench-genmetrics/issues/44