automl/HpBandSter

When I would like to find good configurations for log distributional variables

nabenabe0928 opened this issue · 2 comments

nabenabe0928 commented

Hi, I would like to optimize the configuration which includes some variables which should be sampled after converted into log scale.

I checked your codes and found there I could choose 4 types of variable: "Continuous", "Ordinal", "Integer", "U"(categorical). I would like to make the points clear.

In the former hyperopt or Optuna offered by PFN, I can choose the distributions from "Uniform", "Discrete", "LogUniform", "Categorical". In the HpBandSter, how can I handle with these kind of variables??

I'm looking forward to hearing from you.

shuhei

sfalkner commented

Hello,
The definition of the parameters is done with the ConfigSpace package. There is only limited documentation for that, but your case is rather easy: The Continuous and Integer parameters have an additional argument called log. If you set that to true, the optimizer will assume a uniform prior in log-space, which is what you want, I think.
Hope that helps!

mfeurer commented

Thanks to @PhMueller we now have "real" docs for the ConfigSpace at https://automl.github.io/ConfigSpace/master/