Thank you for your great work. Can you provide the training hyperparameter of the 'dino+mixvpr' experiment?I just want to reproduce it