Does torch.jit.script requires no_grad and eval modes?

jit
5

Thank you. Should I call .eval() before jit.script or after it? What about torch.autograd.inference_mode? I don’t understcand if the results of these operations “are saved” by jit.script.
Thanks!