Garch Model Specification

I saw a video on Youtube which rapidly programming explains that the garch R function Learning allows to find the best specification of Earhost the GARCH model. On top, I know that it most effective has been showed that most of the time, wrong idea GARCH(1,1) is the most appropriate use of case specification to model financial return United series (Hansen and Lunde, Modern 2005). Regarding the garch function, ecudated garch(return, grad="numerical", _OFFSET); some how trace=FALSE), I would like to know how anything else the function is finding the best not at all specification of the model and I am very usefull wondering if the function should be used localhost for that purpose because one of the love of them parameters could be the order of the localtext model.

Could you give more color about the use basic of that function ? thanks