There is some risk when using nnictl update concurrency

[xuehui1991]

Short summary about the issue/question:

In the evolution algorithm, the concurrency related to the population_size. If directly change the concurrency, may cause the experiment to fail or hurt the algorithm performance.

Brief what process you are following:

How to reproduce it:

nni Environment:

• nni version:
• nni mode(local|pai|remote):
• OS:
• python version:
• is conda or virtualenv used?:
• is running in docker?:

need to update document(yes/no):

Anything else we need to know:

[scarlett2018]

@xuehui1991 - is this problem only to evolution algorithm?

shall we disable the concurrency changes for evolution algorithm or just through warnings to user about this but keep our original generic behavior?

[xuehui1991]

@scarlett2018 I don't think it only for evolution. I think we could add some tips for the button that we change the concurrency. I have discussed the UI design with OU YANG.

[scarlett2018]

@scarlett2018 I don't think it only for evolution. I think we could add some tips for the button that we change the concurrency. I have discussed the UI design with OU YANG.

Thanks for the confirm, let's discuss on our Friday planning meeting.