I need to optimize my pytorch model. Is there any way to optimize the pytorch model by either updgrading versions (pytorch version or cuda version) or trying out any script ?
Updating to the latest stack is usually recommended (and let us know if you are seeing any regressions).
Besides that the performance guide is a good starter to optimize your use case.