Junwei Deng

This PR is only used for temp test without merging intention right?

sure, I will have a look

This issue is raised for one of our users so I add a user issue tag

@helenlly if possible we could talk about this and make some change to the document?

> I think we may also need a fundamental example to `python/nano/tutorial/inference/pytorch` to show the usage of `async_predict` on optimized pytorch model. Yeah I think we do, but maybe we...

I am pretty sure this should work, but we should be careful about this. @liangs6212 do you have any comments about this?

> > I am pretty sure this should work, but we should be careful about this. @liangs6212 do you have any comments about this? > > Can the same modification...

@plusbang @rnwang04 Since this default value is used everywhere so I am asking more stakeholders to have a review

> @TheaperDeng How do you think of this? yeah, I think that should be reasonable to first make sure numpy, loader and tsdatset is supported. I think you may split...

Some of them is on-going in this PR, I will handle them: https://github.com/intel-analytics/BigDL/pull/5516