Aaron Klein

Hmm for me the example works. I pushed my local changes could you pull it and try again?

It seems a bit odd that BO suggests configurations that are not in the bounds. Which version of RoBO do you use?

The underlying datasets is always the same for all benchmarks, they just differ in the encoding for the adjacency matrix. NASCifar10a encodes the matrix as a bit string, NASCifar10B consists...

good question, I used an older version of [SMAC3](https://github.com/automl/SMAC3) but it should also work with the current master branch. However, for specific problems with SMAC I would recommend to open...

Yes, you just need to run the scripts in the experiment_script folder. In the paper we ran each optimizer multiple times (~100 times IIRC) and the figures show the averaged...

Good catch! Could you send a pull request to change that?

the interface looks the same as for Fabolas just that the additional variable is an integer defining the task rather than a continuous variable: ``` def objective(x, task): return function_value,...

The problem is that you use python2.7. RoBO only supports python3.4 and python3.5. See also here https://github.com/automl/RoBO/issues/62

you need to use our fork of george https://github.com/automl/george However, note that RoBO is not maintained anymore and I moved the most stuff, such as fabolas to [emukit](https://github.com/amzn/emukit)

Hi, unfortunately RoBO doesn't allows for parallel BO runs. However, we are actively working on a new BO method based on Hyperband (see our [workshop paper](http://ml.informatik.uni-freiburg.de/papers/17-BayesOpt-BOHB.pdf)) that can be run...