What is the collar and overlap setting for computing diarization benchmarks

[ivangtorre]

For reproducibility,
What are the parameters of collar and skip_overlap used to compute the diarization benchmark table?
Are those values obtained by estimating the number of speakers?

pyannote.metrics.diarization.DiarizationErrorRate(collar=¿?, skip_overlap=¿?)

[hbredin]

collar=0.0

skip_overlap=False