# XGBoost Fit vs Train

I am trying to do a grid searching using the methodology that mentioned in this post. However, I found that XGBClassifier().fit() is using much more memory than xgboost.train. Does anyone know why? Is this related to sparse matrix?