Hyperas breaks when using IDEs (pycharm, spyder, etc.)

[maxpumperla]

Hyperas currently only supports working from command line, executing scripts, and from jupyter notebooks with the respective keyword. Other ways of running experiments are currently not supported.

[pkainz]

@maxpumperla Which version are you using? I don't have issues with PyCharm 2017.1.3.

[maxpumperla]

@pkainz I have no problems either, but other people do. The issues people open revolve around the same topics again and again, so I wanted to have this central ticket for IDE support.

Often, IDEs produce overhead when invoking python, see e.g. the two closed issues referenced above. Essentially, sometimes we don't go further down enough in the stack trace to inspect the file we actually want. With the current approach chosen for hyperas, I don't see how we can cover every single use case. If you have an idea, please let me know.

[shadiakiki1986]

@maxpumperla What do you mean by "from jupyter notebooks with the respective keyword"?

[maxpumperla]

@shadiakiki1986 there's a notebook flag you can invoke

[shadiakiki1986]

Ah ok. Just found the example notebook

[mmebK]

One way that works for me is to open jupyter notebook and give a name to a notebook let's say 'name' then in spyder specify notebook_name='name'. This should work for you.