when comparing a RMSE of a train model to the RMSE test, in one model my train is higher than my test and another model my test is higher than my train. the R -square in both case in 0.22 but is expected given the dataset according to my lecturer. I need to know which model to use and why thanks.

Also in both case there is no multi collinearity of independent variables and p-values for all variables are less than 0.05 when using a 5% significance level.