Does "Deep Learning on a Data Diet" reproduce? Overall yes, but GraNd at Initialization does not