Kfold for fintunning model

Can K-fold cross-validation be used for fine-tuning? It seems to be creating different models with each fold when I try to use it