To present results in research papers, a random seed is typically used for repeatability. However, it we know that the choice of the random seed can significantly impact the obtained results. This makes me wonder what researches do to report their results metrics.
Do they select the seed that produces the highest or lowest metrics, or report the average.
Aucun commentaire:
Enregistrer un commentaire