IRF measure

i am using the model(linear) option .
dynare plots IRFs for the variable X
which in reality is log (X) - log(Xbar).

suppose for the estimation, the data is entered in percentages. for example, the interest rate is given as 3%, 4% etc instead of 0.03, 0.04.

i am not sure if i am clear. example, if dynare shows that output declines to 0.2 after a monetary policy shock, is it a 0.2% decline or is it 20% decline ( from steady state)?
is it necessary to multiply the impulse responses computed by dynare by 100 in this case so that the output appears in % deviations from steady state?
it is only a scaling i am referring to, 100( log (X)- log(Xbar)) is a percentage deviation in my terms.