Running a trace in eval() mode

jit
1

Is it possible to run a trace of a model in .eval() mode which can latter be used to find the freeze of the model
frozen_module = torch.jit.freeze(traced)