Jaeger is a small package for performing random search on parameter spaces for functions with a nice interface. I hope to extend it with Bayesian optimization and distributed evaluation techniques.
The approach is that the user defines a function which all the parameters that should vary as inputs. In the next step, he defines a search space consisting of several random variables. She can then sample from that search space in order to pass those parameters into the function of interest.
For an example, see examples/neuralnetwork.py.