The non linear model worked fine but in exp() form it says the steady state values are too far away from initial values and I am using the same initial values as I did for the non linear form. I am not sure about what I should do next. Any help is welcome
Did you take the logs of initial values of the non-log version? If yes, instead of using the log of the initial values from the non-log version, use the log of the actual steady state values found by Dynare. The problem is that due to the variable substitution, the numerical optimizers may have trouble finding the same steady state.