I'd say turn it on, since otherwise the introduced forces may heat up and finally blow apart the system.

Thank you Norman. I pulled my structures with constant temperature on and Langevin damping coefficient of 5, But today I realize this part in Tutorial:

"The temperature control should be disabled in order to disturb the movement of the atoms as little as possible. Switch off the Constant Temperature Control"

It seems that it is just proper for a tutorial not for a real Simulation though.

thank you again.

One thing that you should usually be concerned about is the fact that using something like Langevin thermostat introduces irreversibility. I'm not sure if this is a factor in steered MD, but if reversibility is assumed in the method (which for example is the case with hybrid MD/MC) you cannot use such methods for temperature control. Instead, there are thermostats based on reversible non-Hamiltonian dynamics such as the MTTK integrator.